GARWOOD, JERRY E. SMITH and WIENER, Circuit Judges, dissenting:
I respectfully dissent from the majority’s decision to reverse the district court’s judgment granting the writ.
It is important to recognize at the outset that whether an actual conflict of interest between an attorney and his client exists is a separate inquiry from whether we apply Cuyler v. Sullivan, 446 U.S. 335, 100 S.Ct. 1708, 64 L.Ed.2d 333 (1980), or Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), when a criminal defendant or, as here, a habeas petitioner challenges his conviction based on the alleged existence of an actual conflict of his trial counsel. Whether an attorney-client conflict exists must be addressed at the commencement of the representation not only by the attorney and his client, but also frequently by the trial court. The same question must be addressed as a threshold issue on appeal or on habeas review. If we allow the context in which the question of the existence of an actual attorney-client conflict arises here — on retrospective review of a conviction — to distort the criteria for determining whether an actual conflict exists, we inevitably skew the same inquiry when it is made at the beginning of the representation. This we cannot do.
The district court’s conclusion that the execution of a media rights contract created an actual conflict of interest between E. Ray Andrews and his client, Betty Lou Beets, is correct, and the majority’s contrary conclusion is legally and factually insupportable. If that conflict of interest was the cause of Andrews’s failure to withdraw and testify on Beets’s behalf — an issue that I would remand to the district court to decide — then Beets will have shown that it had an adverse effect on Andrews’s representation, and applying Cuyler, the writ was properly granted.
Finally, I disagree with the majority’s unprecedented decision to limit the rule of Cuyler to cases involving multiple or serial representation. The court thereby excludes from the ambit of Cuyler an exceptional conflict between an attorney’s self-interest and his client’s interest stemming from a highly particularized and powerfully focused source, *1280a media rights contract. If we reserve Cuyler for extraordinary attorney-client conflicts of that sort, not normally encountered in law practice, and we apply Strickland to alleged deficiencies in an attorney’s performance having their sources in the more common incidents of the attorney-client relationship, we avoid having the Cuyler exception swallow the Strickland rule. At the same time we preserve the benefit of the Cuyler inquiry for those exceptional cases that he at the heart of the principles animating it.
I. BACKGROUND
A. Andrews’s Representation
A full understanding of the issues in this appeal requires a more complete examination of the facts and circumstances surrounding E. Ray Andrews’s representation of Betty Lou Beets than the majority provides. It is clear from the record of Beets’s trial and from the record of the federal habeas proceedings that the testimony of Andrews was critical to Beets’s defense that she did not murder Jimmy Don Beets for remuneration. It is also clear from the record of the federal habeas proceedings that Andrews contemplated obtaining the media rights contract very early in his representation of Beets, long before the trial began.
As the majority notes, in late 1984, more than a year after Jimmy Don’s disappearance, the mobile home in which Beets lived was destroyed by fire.1 The insurance company, apparently suspicious of the claim, resisted paying on the policy. Thus, in his testimony at the federal habeas proceeding, Andrews agreed that Beets “came to [him] looking for insurance benefits, but not with respect to the death of Jimmy Don Beets.” Instead, Andrews testified, Beets approached him for help in collecting the proceeds from the insurance policy covering the mobile home.
At the habeas proceeding, Andrews testified that he believed that he had suggested to Beets, and thought he was obligated to suggest, pursuing any benefits that might be available as a result of Jimmy Don’s disappearance. As Andrews testified, “Ms. Beets never pushed me like some clients would for money, proceeds, and it was ... through independent investigation that I found out that she had money maybe due and payable or owing to her.” Andrews and Beets agreed that Andrews, in a contingent fee arrangement, would help Beets pursue any benefits to which she might be entitled.
After his initial efforts proved unsuccessful, Andrews sought the assistance of brothers Bruce L. and Randell C. Roberts, attorneys who were practicing in Tyler, Texas. According to Randell Roberts’s affidavit that was admitted into the record of the habeas proceeding in lieu of five testimony, Andrews arranged for himself, Beets, and Randell Roberts to meet. Roberts recalled that Andrews did most of the talking at that initial meeting, and that:
With respect to potential life insurance benefits ... Ms. Beets was able to provide ... very little information. It was my impression that she believed at the time that there were probably some life insurance or pension benefits due to her, however, she appeared to know very little about the amount of the benefits in question or the potential insurance companies or other sources which would be responsible for these benefits.
Eventually, Randell Roberts passed the file to his brother Bruce, who began looking for benefits. At Beets’s trial, Bruce Roberts testified that “when [he] first took the case, [Beets’s] primary concern was ... with the fire insurance company.” Bruce Roberts also testified that Beets had what “looked like part of a policy from the credit union in Dallas. She also knew that she had or was asking me to check into pension benefits.” Bruce Roberts further testified, and later reemphasized in his affidavit which was also admitted into the record of the habeas proceeding, that Beets had no idea what benefits she may have been entitled to. Despite Beets’s ignorance about any benefits she may have been due, Bruce Roberts pursued the *1281claims, writing letters and making telephone calls to anyone he thought might have owed Beets money as a result of her husband’s disappearance.
Bruce Roberts’s efforts met with some success, and he had Jimmy Don declared dead and secured a settlement with the City of Dallas for some pension funds. In early June of 1985, before the settlement was finalized, Jimmy Don’s skeletal remains were unearthed from a wishing well in front of the mobile home. Beets was subsequently arrested and charged with murder.2 The case, as the majority notes, generated significant local and national media attention. Andrews agreed to represent Beets in the murder trial, and there is evidence that from very early on in his representation of Beets, Andrews envisioned profiting from the Betty Lou Beets story.
The same month that Beets was arrested and that Andrews began his representation of Beets — June 1985 — Andrews associated Gilbert M. Hargrave to assist in the trial. According to Hargrave’s testimony in the federal habeas proceeding, in June of 1985, long before the trial began and before Har-grave had agreed to work on the case, Andrews stated, “T’m going to get the book rights and I’ll give you twenty percent of the book rights.’ ” Hargrave also testified that “[Andrews] thought the case was a valuable case, that the book rights were valuable, that it was notorious, famous, ... and that it would generate a lot of income-producing type of publicity for himself and myself.”
Additionally, well before Beets’s trial commenced, Andrews undertook efforts to secure the media rights. The record of the federal habeas proceeding contains two draft versions of a contract assigning the media rights of the trial to Andrews’s son. Specifically, there is a typed draft of a media rights contract dated September 23, 1985 and a handwritten draft of the same document, presumably written sometime earlier. Thus, even though the majority notes that “[o]n October 8, just after Beets’s trial commenced, she signed a contract transferring all literary and media rights in her case to Andrews’s son,” Beets v. Collins, 65 F.3d at 1261 (5th Cir.1995) (en banc), it is clear from the record that securing the media rights was on Andrews’s mind virtually from the beginning of his representation of Beets in connection with her indictment for Jimmy Don’s murder.3 As it developed, the assignment of the media rights was the consideration for Andrews’s services in defending Beets.
During the trial, Andrews had two lines of defense. His principal strategy during the guilt phase of the trial was simply to show that Beets did not commit the murder. Andrews, however, left little doubt that his secondary strategy was “to try to attack the State’s proof on their claim that [Beets] did it for remuneration.” As co-counsel Hargrave testified at the habeas proceeding, “[t]he basic theory [of the defense] was that [Beets] was not guilty, that she hadn’t committed the act that she was under indictment for and that if she actually had that she certainly hadn’t done so for remuneration.”
Accordingly, during trial, Andrews repeatedly attempted to make clear to the jury that it was his suggestion that Beets seek out benefits resulting from Jimmy Don’s disap*1282pearance. As noted above, Andrews elicited testimony from Bruce Roberts that pursuing benefits from Jimmy Don’s disappearance was not suggested by Beets. Moreover, during his examination of Beets, Andrews attempted to show that Beets was not interested in any insurance benefits.4 Further, during his closing argument, Andrews again attempted to convey that he had suggested pursuing insurance and pension benefits, stating:
They’re saying that [Beets] killed Jimmy Don Beets for insurance money. Ladies and gentlemen, she didn’t even know anything about insurance, how much insurance he had or anything. Me and other lawyers inquired into this. Never called me in nearly two years.... Does that sound like somebody that’s out after insurance money?
Andrews reemphasized this near the end of his argument, asking the jury:
Did [the prosecutor] ever prove to you, people that she ever collected any of his retirement proceeds? ... [T]he only proof that came in was a lawyer works for money. If a lawyer sees ... there’s a ease there, they’re going to go after it. And I probably should have gone after it faster. I’m certainly glad now I didn’t.
Viewing the record in this case, there is no question that attempting to show that Beets did not act for remunerative purposes was an important aspect of Andrews’s strategy. Andrews’s efforts to accomplish this objective, both in examination and in argument, were neutralized to some extent by the court’s repeated instruction that “what the lawyers say is not evidence.” As Beets’s counsel, Andrews, the only person besides Beets who could testify about exactly how the pursuit of the insurance and retirement benefits began, was precluded from testifying, and the jury was instructed not to consider as evidence any statements which he made about his involvement. Moreover, it is possible that the jury discounted the statements that he did make at trial as impermissible efforts to bolster his client’s case.
Judge Higginbotham’s view about the importance to Beets’s defense of Andrews’s testimony, set out succinctly in his concurrence to the panel opinion in this case, bears repeating:
Andrews’s testimony could have significantly bolstered th[e] defense_ Andrews ... could have told the jury that he mentioned to Beets the possibility of receiving benefits shortly after Jimmy Don’s disappearance. Any later interest or inquiry into benefits could have been attributable to this post-murder information. Moreover, Andrews could have established Beets’s lack of knowledge at a time closer to the murder than Roberts’ evidence. Andrews’s testimony was not merely cumulative. ... It certainly would have been in Beets’s best interest for Andrews to have testified.
Beets v. Collins, 986 F.2d 1478, 1491 (5th Cir.) (Higginbotham, J., specially concurring), reh’g en banc granted, 998 F.2d 253 (5th Cir.1993).
B. The District Court’s Findings
A full grasp of this case also requires a careful consideration of the district court’s findings. The district court began its Order by noting that “it is apparent that the defense counsel, E. Ray Andrews, fought for his chent to the full extent of his ability and energy.... Andrews is well known to the Court as a competent and tenacious criminal lawyer.”
*1283Subsequently, the district court ruled against Beets on most of her habeas claims, and then turned to “the issue which ha[d] proven most troublesome ... [Beets’s] Sixth Amendment claim.” The court started its analysis by stating that “there are actually two conflicts in this case, the conflict created by the media rights contract ... and the conflict arising from the fact that the attorney should have been a witness instead of an advocate....” Although the district court stated that “the two conflicts may be intertwined to a limited extent,” it addressed the conflicts separately.
After examining the framework for analyzing Sixth Amendment challenges based on conflicts of interest, as set forth by the Supreme Court in Cuyler v. Sullivan, 446 U.S. 335, 100 S.Ct. 1708, 64 L.Ed.2d 333 (1980), the district court concluded:
Mere demonstration of an actual conflict is insufficient; the term ‘adverse’ must mean that some negative impact on counsel’s performance is required. After careful consideration, this court is of the opinion that an adverse effect on performance is demonstrated when counsel, laboring under an actual conflict of interest, pursues some course of conduct inconsistent with the best interest of his client.
Applying this standard, the district court first reemphasized that Beets had “demonstrated two actual conflicts of interest in this case, the media rights conflict and the witness/advocate conflict.” The court then turned to the adverse effect prong of its analysis. As to the media rights conflict, the court “simply d[id] not believe that [it] affected Andrews’[s] performance at any conscious level. There is, of course, no adverse effect when there is no effect at all.” The court also noted, however, that “[t]he possibility exists that the media rights contract motivated Andrews at a subconscious level to remain in the case when he should have withdrawn and testified for Petitioner. To that limited extent, the two conflicts are intertwined.” The district court did not explore this relationship, finding instead that “the witness/advocate conflict is a separate conflict which did in fact adversely affect Andrews’s performance. This is sufficient under Cuyler without a detailed analysis of Andrews’[s] possible motivation.”
As to the witness/advocate conflict, the court described Andrews’s knowledge of Beets’s pursuit of benefits resulting from her husband’s death as well as Andrews’s efforts to communicate that knowledge to the jury. The district court found that those efforts were insufficient, stating that “Andrews obviously should have known of his dual status as a witness and advocate prior to trial. Andrews’[s] dual status should have also been apparent to both the judge and district attorney as the trial unfolded.” Although the district court recognized that “the conflict never occurred to any of the participants,” it also noted that “[t]he testimony that Andrews could have provided as an independent witness related to an essential element of the State’s charge of murder for remuneration.” Thus, the district court concluded that “counsel pursued a course of conduct inconsistent with his client’s best interest when he accepted employment or failed to withdraw and testify as a witness on [Beets’s] behalf.” Accordingly, the district court granted Beets’s habeas petition.
With a clear understanding of the factual background and district court findings in the case, I turn to an examination of the substantive issues in this appeal. To establish a Sixth Amendment violation, the Supreme Court has held that a defendant who did not raise the objection at trial “must demonstrate that an actual conflict of interest adversely affected his lawyer’s performance.” Cuyler, 446 U.S. at 348, 100 S.Ct. at 1718. I first present what seem to me to be the threshold inquiries in the context of an asserted Sixth Amendment violation involving a conflict between the interest of a lawyer and the interest of his client. I look then at the question whether an actual conflict of interest existed between Andrews and Beets, and at the question whether any such conflict adversely affected Andrews’s performance. Finally, I address why the Cuyler standard, as opposed to the more stringent Strickland standard, should apply to this case.
II. THE CONFLICT BETWEEN ATTORNEY AND CLIENT
A. The Threshold Inquiries
The Second Circuit’s decision in Winkler v. Keane, 7 F.3d 304 (2d Cir.1993), cert. de*1284nied, — U.S. -, 114 S.Ct. 1407, 128 L.Ed.2d 79 (1994), is particularly instructive in this case because, unlike the many cases addressing the actual conflict issue in the multiple representation context, Winkler addresses a conflict between the interest of the lawyer and the interest of his client. The issue presented by Winkler was whether a contingency fee agreement between a criminal defendant and his attorney created a conflict of interest for the attorney resulting in a violation of the defendant’s Sixth Amendment right to effective assistance of counsel. The court began by noting that an attorney has an actual, as opposed to a potential, conflict of interest “when, during the course of the representation, the attorney’s and defendant’s interests diverge with respect to a material factual or legal issue or to a course of action.” Id. at 307 (emphasis added) (internal quotation omitted). Having defined when an actual conflict of interest exists between an attorney and his client, the court went on to analyze the alleged conflict at issue:
Winkler argues that the contingency fee created an actual conflict of interest for trial counsel because Winkler’s interests in effective representation were pitted against trial counsel’s monetary interest. We agree. The contingency fee agreement in this case provided trial counsel with an extra $25,000 only if Winkler was acquitted or otherwise not found guilty. Thus, trial counsel had a disincentive to seek a plea agreement, or to put forth mitigating defenses that would result in conviction of a lesser included offense. Plainly the contingency fee agreement created an actual conflict of interest.
Id. at 307-08. It is important to note that the Winkler court focused only on the objective divergence of interests between the lawyer and his client to determine whether an actual conflict existed. Having found such a conflict, the court went on to reject Winkler’s argument that proof of adverse effect was not needed to grant relief under the Sixth Amendment. The court held that to prove a Sixth Amendment violation, Winkler must meet the Cuyler standard, and that standard required proof of an adverse effect. See id. at 308.
Winkler argued that he was adversely affected by his counsel’s failure to initiate or to engage in plea bargaining and by his counsel’s failure to develop an intoxication defense to Winkler’s second degree murder charge. According to Winkler, both of these alleged failures were motivated by his counsel’s pecuniary interest in total acquittal, which was the only outcome that would entitle counsel to payment of the $25,000 bonus under the contingency fee agreement. See id. at 309.
To address these adverse effect arguments, the court laid out a test for “prov[ing] adverse effect on the basis of what an attorney failed to do”:
[a defendant first] must demonstrate that some plausible alternative defense strategy or tactic might have been pursued. He need not show that the defense would necessarily have been successful if it had been used, but that it possessed sufficient substance to be a viable alternative. Second, he must establish that the alternative defense was inherently in conflict with or not undertaken due to the attorney’s other loyalties or interests.
Id. (quoting United States v. Gambino, 864 F.2d 1064, 1070 (3d Cir.1988) (quoting United States v. Fahey, 769 F.2d 829, 836 (1st Cir.1985)), cert. denied, 492 U.S. 906, 109 S.Ct. 3215, 106 L.Ed.2d 566 (1989)).
In applying the test, the court looked first at the failure to initiate plea bargaining. The court noted that the state court (which had held a hearing on the defendant’s attorney-conflict claim) had found that in an alleged contract murder case, the prosecution would have been highly unlikely to accept a plea agreement. The court held, however, that:
Winkler need not show that a strategy would have been successful, only that it “possessed sufficient substance to be a viable alternative.” Even if it is likely to be unsuccessful, the negotiation of a plea bargain in a case in which the evidence is strongly against a defendant is a viable alternative.
Id. (citation omitted). The court’s determination that a viable alternative had not been pursued did not end the adverse effect inqui*1285ry. The court noted that the state court had found that plea bargain possibilities were not pursued because Winkler had advised his counsel that he was totally innocent and that he was not interested in pleading to a lesser charge even if the opportunity to do so were offered. See id. Thus, the Winkler court concluded that “trial counsel did not pursue a plea bargain because Winkler rejected this path, not because of trial counsel’s monetary interest in the outcome.” Id. (emphasis added).
The court made the same kind of inquiry into the failure to develop an intoxication defense. Because “Winkler had snorted cocaine and smoked marijuana before the fatal event,” the court found that an intoxication defense also had sufficient substance to be a viable alternative. See id. at 310. Nevertheless, the court found that Winkler’s counsel had discussed the possibility of a conviction of lesser charges on the basis of intoxication, but Winkler had rejected this alternative, again asserting his innocence. See id. The court accepted the state court’s factual conclusion that ‘Winkler failed to establish that the fee arrangement caused trial counsel not to seek a conviction for lesser charges.” Id. (emphasis added). The court ended by concluding that Winkler had “failed to prove that trial counsel’s representation was adversely affected by the conflict of interest. Thus, his Sixth Amendment right to counsel was not violated.” Id.
In summary, the Winkler court made three distinct inquiries in its Sixth Amendment analysis. First, the court determined whether an actual conflict of interest existed between the lawyer and his client by asking whether the attorney’s and defendant’s interests diverged with respect to a material factual or legal issue or to a course of action. Second, in addressing whether there had been an adverse effect, the court inquired as to whether a viable alternative might have been pursued. Third, the court made a proximate cause inquiry, asking whether the viable alternative was not pursued because of the conflict. A Sixth Amendment violation was made out only if all three inquiries were affirmatively answered — i.e., the interests of the lawyer and his client diverged, a viable alternative was not pursued, and the failure to pursue the viable alternative was caused by the divergent interests. Using this threshold framework, I proceed to Beets’s case.
B. Was there a Conflict?
As described above, the district court found that Beets “demonstrated two actual conflicts of interest in this case, the media rights conflict and the witness/advocate conflict.” I will address these two “conflicts” separately, turning first to the media rights conflict.
1. Media Rights
A conflict of interest between Beets and Andrews existed not later than the point at which the formal contract giving Andrews’s son the media rights to the Betty Lou Beets story was executed, and perhaps earlier. The majority “joins other courts, scholars and organizations of the bar who have uniformly denounced the execution of literary and media rights fee arrangements between attorneys and their clients during the pen-dency of a representation.” Beets, 65 F.3d at 1273. What the majority fails to acknowledge is the reason for such uniform condemnation — the extraordinarily high probability that a media rights contract between counsel and client will create a conflict of interest. When Andrews began his representation of Beets on her murder charge, Texas’s rules of ethics provided that “[pjrior to conclusion of all aspects of the matter giving rise to his employment, a lawyer shall not enter into any arrangement or understanding with a client ... by which he acquires an interest in publication rights with respect to the subject matter of his employment.” State Bar Rules, art. X, § 9, DR 5-104(B) (Texas Code of Professional Responsibility) (1984).5
*1286The reason for the rule is clear. Despite the majority’s assertions, media rights contracts are not prohibited primarily because they “encourage counsel to misuse the judicial process for the sake of his [own] enrichment and publicity seeking” or because they “necessarily trade[] on the misery of the victim and his family.” Beets, 65 F.3d at 1273. While ensuring that the judicial process is not misused and discouraging manipulation of the suffering of others for profit are important goals, commentators uniformly agree that the reason media contracts are frowned upon is because “[a]n agreement by which a lawyer acquires literary or media rights concerning the conduct of the representation creates a conflict between the interests of the client and the personal interests of the lawyer.” Laws. Man. on Prof. Conduct (ABA/BNA) 51:702 (1984) (emphasis added); see also John Wesley Hall, Jr., Professional Responsibility of the Criminal Lawyer § 12.13, at 414 (1987) (“A grave conflict of interest can arise from a [media rights contract]....” (internal quotation omitted)); Geoffrey C. Hazard, Jr. & Susan P. Koniak, The Law and Ethics of Lawyering 498 (1990) (“The reason for prohibiting such arrangements is that what makes ‘good copy’ does not necessarily make a good defense.”); Robert P. Schuwerk & John F. Sutton, Jr., A Guide to the Texas Disciplinary Rules of Professional Conduct, 27A Hous.L.Rev. 133 (1990) (“The lawyer’s acquisition from a client of publication rights to portrayals or accounts of the subject of the representation will probably create a conflict of interests.”); Charles W. Wolfram, Modem Legal Ethics § 9.3.3, at 525 (1986) (“The problems [with media rights contracts] are two — conflict of interests and the revelation of client informa-tion_”).
The rules against media rights contracts are designed to prevent the specific conflict resonating in this case; simply put, “a lawyer in a criminal case who obtains from his client television, radio, motion picture, newspaper, magazine, book, or other publication rights with respect to the case may be influenced, consciously or unconsciously, to a course of conduct that will enhance the value of his publication rights to the prejudice of his ehent.” State Bar Rules, art. X, § 9, EC 5-4 (Texas Code of Professional Responsibility) (1984) (emphasis added). Plainly, a media rights contract “ ‘may place the lawyer under temptation to conduct the defense with an eye on the literary aspects and its dramatic potential. If such an arrangement or contract is part of the fee, in lieu of the fee, or a condition of accepting the employment, it is especially reprehensible.’ ” Hall, supra, § 12.12, at 414 n. 19 (quoting ABA Standards, The Defense Function Std 4-3.4, Commentary).6 In fact, a media rights contract is so rife with conflict that under Texas’s rules “client consent will not cure a violation of [the prohibitions of media contracts].” Schuwerk & Sutton, supra, at 134.
In the instant case, the media contract weighed on Andrews’s mind from the beginning of his representation. At the very least, the contract placed him in a situation of divided interests. Before the advent of the media rights contract, Beets’s interest lay in having Andrews withdraw as her counsel and testify at her trial that he had initiated the idea of searching for Jimmy Don’s insurance and pension benefits. As her attorney, this was also Andrews’s interest because he was obligated to see to it that his client’s best defense was put forward. After the media rights contract was confected, the interests of Beets and Andrews sharply diverged. While Beets’s interest remained in having Andrews withdraw and testify, Andrews’s interest now squarely lay in remaining as her counsel because only then would he be entitled to the potentially lucrative media rights.
*1287The record makes clear that the district court did not err in finding that Beets demonstrated that Andrews had an actual conflict of interest in regard to the media rights contract.
2. The Lawyer as Witness
The district court also recognized a second conflict of interest, Andrews’s advocate/witness conflict. The district court and the majority treat this conflict separately from the media rights conflict. In a situation where a lawyer can provide favorable testimony material to his client’s case, his failure to withdraw and testify may or may not stem from an actual conflict — i.e., from a divergence of interests between the lawyer and his client. If the failure to withdraw is caused, for example, by a desire to stay in the ease for the fee involved, a conflict of interest may exist. As one commentator has suggested, when a lawyer has a duty to withdraw and testify in favor of his client, but does not do so, “[s]uch a decision would raise serious questions about either the lawyer’s competence or about the effect of a conflict of interest.” Wolfram, supra, § 7.5.2, at 381 (emphasis added). According to this commentator:
The conflict is between the lawyer’s duty of loyalty to the client, which urges the lawyer to give the needed testimony, and the lawyer’s economic instincts, which may lead the lawyer to remain in the case as advocate in order to continue earning a fee that otherwise would have to be abandoned.
Id. (emphasis added). In my view, while Andrews’s role as both an advocate and a potential witness may well have violated the applicable ethical rules,7 its significance for this case lies not in its possible status as an independent conflict, but rather in its relationship to the media rights conflict. That is, as the district court recognized but did not explicitly resolve, the question remaining in this case is whether the media rights contract is what caused Andrews to remain in the case as counsel.8
C. Was Andrews’s Representation Adversely Affected?
Under Cuyler, a defendant does not have the burden of showing actual prejudice — i.e., the defendant does not have to show that the result of the trial probably would have been different. See Strickland, 466 U.S. at 691-96, 104 S.Ct. at 2066-69; United States v. *1288Greig, 967 F.2d 1018, 1024 (5th Cir.1992). Instead, the defendant needs to demonstrate an adverse effect upon his representation, and “Cuyler*s adverse effect element establishes a relatively low threshold for a petitioner to cross.” Beets, 986 F.2d at 1490 (Higginbotham, J., specially concurring). A limited presumption of prejudice arises from a showing of adverse effect because, as the Supreme Court has noted, “it is difficult to measure the precise effect on the defense of representation corrupted by conflicting interests.” Strickland, 466 U.S. at 692, 104 S.Ct. at 2067. To establish an adverse effect on the basis of what an attorney failed to do, a defendant must demonstrate that some plausible alternative defense strategy or tactic— “a viable alternative” — might have been pursued. See, e.g., Winkler, 7 F.3d at 309.
In Beets’s ease, Andrews’s failure to testify had an adverse effect on her defense, as Andrews’s testimony was clearly a viable alternative. Throughout the trial, Beets attacked the remuneration element of the State’s case on which her capital murder conviction rested. Andrews had significant testimony to offer bearing on the critical issue of whether the killing of Jimmy Don was for a remunerative purpose, specifically to obtain Jimmy Don’s life insurance proceeds and pension benefits. If the jury reasonably doubted that Beets killed her husband for the insurance money, the murder was not a capital offense.
The majority downplays the importance of Andrews’s testimony by referring to it as “cumulative.” Beets, 65 F.3d at 1276, 1277. Of course, as mentioned, Beets does not need to show that Andrews’s testimony would have been successful, but only that it possessed sufficient substance to be a viable alternative. See Winkler, 7 F.3d at 309. Andrews’s testimony clearly meets this standard, and the suggestion that his testimony is cumulative is simply based upon an erroneous reading of the record. As evidence of the cumulative nature of Andrews’s testimony, the majority points to Beets’s own testimony that she was unaware of Jimmy Don’s death benefits before she visited Andrews. But the jury may well have discounted Beets’s testimony because of its self-serving nature.
The majority also points to the testimony of Bruce Roberts, who stated that more than a year after Jimmy Don’s murder, Beets seemed ignorant of his insurance and benefits. Roberts’s testimony, however, was damaging in certain respects to Beets’s defense. Although Roberts did testify that Beets’s “primary concern was ... with the fire insurance company” and that Beets never pressured him to collect money from the City of Dallas, Roberts was also asked what Beets knew about benefits when she first came to him. In response, Roberts stated:
At the time I talked to her, she had one-well, it looked like part of a policy, as I recall, from the credit union in Dallas. She also knew that she had or was asking me to cheek into pension benefits. Basically, that’s all the information she could give me.
Thus, Roberts did not definitively testify that Beets was ignorant about the possibility of collecting benefits. Further, Roberts did not, and could not have, testified that Andrews suggested to Beets that they pursue Jimmy Don’s insurance and pension benefits because Roberts was not present at the meeting between Beets and Andrews at which that suggestion was made.
Other than Beets, only Andrews could have told the jury that he initiated the discussion regarding Jimmy Don’s death benefits with Beets, and only Andrews could have testified that she appeared to lack knowledge of any such benefits. Further, only Andrews could have testified that it was at his suggestion that those death benefits were pursued. Any later interest or inquiry into benefits by or on behalf of Beets could have been attributable to this meeting between Beets and Andrews that took place long after Jimmy Don’s murder.
Similarly, the majority’s invocation of Denny Burris’s testimony to downplay any adverse effect from the absence of Andrews’s testimony is erroneous. Burris, a disinterested witness, stated that Beets inquired about Jimmy Don’s insurance soon after the murder. The suggestion is that because Beets discussed insurance with Burris, a *1289chaplain, the biased testimony of Andrews was unlikely to have swayed the jury and its absence was unimportant. First of all, it is again worth mentioning that any suggestion that Andrews’s testimony would not have been successful to Beets’s defense is irrelevant. Second, although Burris was a chaplain, his assignment was to discuss with Beets the benefits due to the widow of a fireman. “[I]t was not a mission to console a widow with prayer where the widow’s interest was insurance not intercession.” Beets, 986 F.2d at 1491 (Higginbotham, J., specially concurring). Thus, the evidence of Beets’s early focus on insurance, as the State would have it, is not so compelling. Finally, Burris’s specific testimony did not show that Beets already knew about Jimmy Don’s insurance and pension benefits. Rather, Burris testified that, several days after Jimmy Don’s disappearance, Beets “asked about insurance, if she would be covered and things like that.”
In short, it is all too clear that Andrews’s testimony was a viable alternative, and his failure to testify had an adverse effect on Beets’s defense. I agree with the assessment of Judge Higginbotham in his special concurrence to the panel opinion:
Andrews’s testimony was not merely cumulative. I cannot conclude that it would not have been helpful to Beets at trial. It certainly would have been in Beets’s best interest for Andrews to have testified. Given the low threshold established by Cuyler, I would not reject [the district court’s] conclusion that Andrews’s failure to give this evidence at trial adversely affected the conduct of her defense.
Id. at 1491-92 (Higginbotham, J., specially concurring).
Our task is not yet complete, however, as the third Winkler inquiry still remains on the table: whether Andrews’s withdrawal and testimony — the “viable alternative” — was not pursued because o/the media rights conflict. This is a fact-bound question that the district court did not directly answer.9 After hearing evidence, the district court found that “Andrews obviously should have known of his dual status as witness and advocate prior to trial.” The district court also concluded that Andrews “pursued a course of conduct inconsistent with his client’s best interest when he accepted employment or failed to withdraw and testify as a witness on Petitioner’s behalf.”
On the other hand, the court concluded that the media rights contract did not affect Andrews’s performance “at any conscious level.” The court noted, however, that “[t]he possibility exists that the media rights contract motivated Andrews at a subconscious level to remain in the case when he should have withdrawn and testified for Petitioner. To that limited extent, the two conflicts are intertwined.” These statements suggest that the district court did not definitively resolve whether Andrews’s failure to withdraw and testify was attributable in some fashion to his actual conflict of interest arising from the media rights contract.
The majority holds that there is no need to remand this case to the district court for an explicit finding on whether the media rights contract caused Andrews to fail to withdraw for two reasons. First, the majority invokes, for the first time in the five years that this ease has been in federal court, the presumption of correctness afforded by 28 U.S.C. § 2254(d) to the findings of fact made by the state trial court judge on Beets’s state habe-as petition. The state trial court found as a fact that “[p]etitioner’s grant of ‘book rights’ to the son of her counsel had no effect on the strategy of defense counsel.” This finding was based on the trial court’s own personal recollection of the trial and on Andrews’s affidavit which stated that:
defense attorney had no conflict of interest throughout the proceedings by agreeing that book rights would be his entire fee, said defense attorney has no book rights, these rights were given to said attorney’s son and were only given after the trial was *1290into its third or fourth day, the State Bar of Texas has found said defense attorney committed no improprieties with regard to this matter.
At no point during the course of Beets’s federal habeas proceedings has the State sought to invoke the presumption of correctness afforded by § 2254(d). Beets filed a motion for an evidentiary hearing along with her federal habeas petition. When the State filed its response (and an amended response) to Beets’s petition, it did not object to the hearing and, in responding to Beets’s claim of an actual conflict of interest, the State asserted:
Respondent denies that there was any conflict of interest in this case. However, because the Court has scheduled an evi-dentiary hearing on the issue, rather than argue the claim at this time, Respondent mil rely on the facts developed at that hearing.
(emphasis added). Finally, the State (appellant in this court) did not raise the preclusive effect of the state court’s habeas findings in any of the many briefs it has filed with this court. Neither the panel opinion nor Judge Higginbotham’s special concurrence mentions § 2254(d) or Sumner v. Mata, 449 U.S. 539, 101 S.Ct. 764, 66 L.Ed.2d 722 (1981), or any of its progeny. In a last ditch effort to avoid dealing with the ambiguities in the federal district court’s fact findings, the presumption of correctness has been resurrected despite the State’s unwillingness to invoke it. Because no party has addressed the presumption of correctness, we cannot determine whether any of its exceptions applies. For example, we do not know the position of the State or Beets on whether the factfinding procedure employed by the state court was adequate to afford a full and fair hearing or whether the material facts were adequately developed at the state court hearing. See 28 U.S.C. § 2254(d)(3) and (4). This case should not be decided at this late date on a basis not raised by the parties.
The majority asserts as its second reason why, applying Cuyler, there is no need to remand for an explicit finding on causation the district court’s statement that “the court simply does not believe that the media rights contract affected Andrews’[s] performance at any conscious level.” The majority ignores the district court’s recognition of a possible connection between the media rights contract and what it termed the “witness/advocate conflict” and its failure to resolve the ultimate question whether they were related. The majority also ignores the contradiction inherent in the district court’s opinion in finding the witness/advocate ethical problem to be an actual conflict (thereby implying that it stemmed from a divergence of interests between Andrews and Beets) while finding at the same time that the media rights contract (the likely source of the divergence) had no conscious effect on Andrews’s performance.
In my view, because the district court did not explicitly decide whether Andrews’s failure to withdraw and testify was caused by the actual conflict engendered by the media rights contract, the wiser course is to vacate the district court’s judgment granting the writ and to remand the case so that the district court may consider the question in the first instance. If the district court determines on remand that Andrews’s failure to withdraw and testify was caused by the actual conflict of interest arising from the execution of the media rights contract, with its powerful incentive to remain in the case, then Beets will have successfully demonstrated “that an actual conflict of interest adversely affected [her] lawyer’s performance,” Cuyler, 446 U.S. at 350, 100 S.Ct. at 1719, and she will be entitled to habeas relief. The judgment granting the writ should then be reinstated. If, on the other hand, the district court concludes that Andrews’s failure to withdraw and testify was not caused by the actual conflict stemming from the media rights contract, then that failure should be analyzed under Strickland. On this record, there is no reasonable probability that the outcome of Beets’s trial would have been different, and Beets has failed to satisfy the prejudice prong of Strickland. The writ should then be denied.
D. The Majority Opinion
Before explaining why Cuyler, as distinguished from Strickland, applies to this case, *1291I turn to an examination of the problems in the majority’s approach to the existence of a conflict in this case.
The majority is squarely faced with the district court’s fact-bound conclusion that Beets “demonstrated two actual conflicts of interest in this case, the media rights conflict and the witness/advocate conflict.” The majority holds, however, that “only a potential and not an actual conflict arose between Beets and her lawyer.” Beets, 65 F.3d at 1260.
The majority discounts the district court’s conclusion that an actual conflict existed by reason of the media rights contract, noting that:
[T]he media rights contract posed a serious potential conflict of interest, [but] Beets failed to show how it hindered Andrews’s presentation of her defense or prejudiced her by rendering the result of her criminal prosecution fundamentally unreliable. Beets has not asserted that Andrews manipulated the case to enhance publicity or that the contract generally clouded his good judgment. Beets has shown no actual influence of the media rights contract on the conduct of her defense.
Beets, 65 F.3d at 1274. Moreover, the majority later seems to resurrect the vacated panel opinion, noting that two judges held in that opinion “[a]s to the media rights contract, there was no ‘actual conflict’ ... because ... [t]he record does not demonstrate that the contract induced Andrews to compromise his zealous representation of Beets in favor of his own pecuniary interest.” Id. at 1277.
Similarly, in discussing whether there was a conflict in Andrews’s failure to withdraw and testify, the majority states that “[b]e-eause Andrews’s potential testimony for Beets was cumulative, he was not a necessary witness for her defense and did not face a substantial advocate/witness conflict.” Beets, 65 F.3d at 1276-77. Again, the majority seems to resurrect the conclusion of the panel opinion that “Beets alleged, at most, a merely hypothetical or speculative witness/advocate conflict, which did not materialize into an actual conflict that forced Andrews to choose between his self-interest and his duty to Beets.” Beets, 65 F.3d at 1277.
The majority’s conclusions fail on two levels. First, the majority suggests that an actual conflict does not exist until an attorney makes a “choice” between his interest and the interest of his client. Building on the requirement of a “choice,” the majority seems to add (as did the panel) a scienter element to conflict analysis, inserting a requirement, heretofore alien to the law, that an attorney must consciously recognize that he is operating under a conflict before that conflict can be said to exist in fact. Second and most noticeably, the majority conflates the existence and effect elements of the analysis by concluding that a conflict did not exist because Beets failed to demonstrate an effect.
1. The Function of “Choice” in Conflict Analysis
The majority looks to statements by the Seventh, Tenth, and Eleventh Circuits to support the panel’s contention that an actual conflict does not exist until an attorney makes a “choice” between his interest and the interest of his client. See Beets, 65 F.3d at 1277 (citing Stevenson v. Newsome, 774 F.2d 1558, 1561-62 (11th Cir.1985), cert. denied, 475 U.S. 1089, 106 S.Ct. 1476, 89 L.Ed.2d 731 (1986); United States v. Litchfield, 959 F.2d 1514, 1518 (10th Cir.1992); United States v. Acevedo, 891 F.2d 607, 610 (7th Cir.1989); United States v. Horton, 845 F.2d 1414, 1419 (7th Cir.1988)). The concept of “making a choice” is typically used in cases where an attorney has a potential conflict (frequently between two clients) — i.e., a situation where interests have not yet diverged but could do so in the future. The courts look to whether a “choice” has been made only to signal that a divergence of interests has occurred — i.e., only to signal that a potential conflict has become an actual conflict. The central question is whether the interests have diverged, and the concept of “making a choice” is an analytical tool used to answer this question.
The cases cited by the majority support this proposition. See Stevenson, 774 F.2d at *12921562 (noting that “[t]here is no evidence in this case that [the attorney] was subject to divided loyalties sufficient to establish an actual conflict of interest,” thus, rejecting Stevenson’s allegations upon a determination that divergent interests were absent from the case) (emphasis added); Horton, 845 F.2d at 1420 (focusing on the absence of divergent interests in finding no conflict was created by counsel’s application for a position as a United States Attorney by stating that “[i]n any event, a candidate for a high federal position in his professional field would not advance his own interest by demonstrating that he is a weak or unskilled attorney on behalf of his client’s interests.”); Acevedo, 891 F.2d at 610 (failing to find conflict, but noting that if Acevedo had alleged in her affidavit that her attorney was involved with her in the criminal activity, he “would have an obvious interest in preventing Acevedo from testifying and thus implicating him in the illegal scheme”); Litchfield, 959 F.2d at 1518 (rejecting defendant’s claim that a conflict of interest arose because trial counsel, concerned that defendant was going to commit perjury, held an ex parte conference with judge, noting that “[t]he situation presented counsel with a difficult dilemma, and we cannot say that his ex parte discussion with the district court was a violation of his ethical duty or evidence of a conflict of interest.”)
This focus on divergent interests to determine whether an actual conflict of interest exists is also the focus in our circuit: “‘[a] conflict exists when defense counsel places himself in a position conducive to divided loyalties.’” United States v. Vaquero, 997 F.2d 78, 89 (5th Cir.) (quoting United States v. Carpenter, 769 F.2d 258, 263 (5th Cir.1985)), cert. denied, — U.S. -, 114 S.Ct. 614, 126 L.Ed.2d 578 (1993); accord Mitchell v. Maggio, 679 F.2d 77, 79 (5th Cir.), cert. denied, 459 U.S. 912, 103 S.Ct. 222, 74 L.Ed.2d 176 (1982).
In Beets’s case, divergent interests existed, at the latest, when Andrews executed the media rights contract. At that point, it was in Beets’s interest for Andrews to withdraw and testify, while it was in Andrews’s interest to remain as counsel so that he would receive the value of the media rights.
The majority recognizes that disagreements between the majority and this dissent exist not only on whether there was an actual (as opposed to a potential) conflict but also on whether the conflict should be judged from an objective standpoint. Beets, 65 F.3d at 1278. In my view, it is important to be clear that determining whether divergent interests are present such that an actual conflict exists contemplates an objective evaluation of the situation in which counsel is placed.10 As the Ninth Circuit recently noted, “[t]he existence of an actual conflict cannot be governed solely by the perceptions of the attorney; rather, the court itself must examine the record to discern whether the attorney’s behavior seems to have been influenced by the suggested conflict.” Sanders v. Ratelle, 21 F.3d 1446, 1452 (9th Cir.1994).
This objective evaluation makes perfect sense, for if the rule were as the majority suggests, counsel’s actions benefiting himself and harming his client would not be actual conflicts, irrespective of their effect on the proceedings, as long as counsel was too obtuse, insensitive, or selfish to recognize that the pursuit of his own goals was coming at the expense of his client’s defense. As the Supreme Court commented in an analogous context, “[i]t is unlikely that [an attorney] would concede that he continued improperly to act as counsel.” Wood v. Georgia, 450 U.S. 261, 265 n. 5, 101 S.Ct. 1097, 1100 n. 5, 67 L.Ed.2d 220 (1981) (describing how the conflict of interest was properly presented when the lawyer who allegedly had the conflict of interest had prepared the brief and the petition for certiorari). The fact that Andrews arguably chose to continue his representation thoughtlessly as opposed to deliberately does not obviate the fact that given the possible decisions he could have made as an attorney, he undertook a course of action *1293that benefited himself while hindering Beets’s defense. Simply put, an actual conflict is demonstrated when a defendant objectively shows that his interest and his attorney’s interest diverged with respect to a material factual or legal issue or to a course of action, and such a divergence occurred in this case.11
2. Separating the Existence of a Conflict from the Effect of a Conflict
The majority also contends that there was no actual conflict in the context of the media rights contract “because ... [t]he record does not demonstrate that the contract induced Andrews to compromise his zealous representation of Beets in favor of his own pecuniary interest.” Beets, 65 F.3d at 1277. Similarly, in the lawyer as witness context, the majority contends that there was no actual conflict “[bjecause Andrews’s potential testimony for Beets was cumulative, [and] he was not a necessary witness for her defense.” Id. at 1277. As I have pointed out above, this conclusion has no factual support in the record. As a legal conclusion, it has no support in the case law surrounding Cuyler; determining whether there was an actual conflict (as distinguished from determining whether the Sixth Amendment has been violated) does not require a showing of an adverse effect.
As the majority concedes, Cuyler incorporates a standard less rigorous than Strickland. See Strickland, 466 U.S. at 692, 104 S.Ct. at 2067 (noting that unlike general ineffective assistance of counsel claims, the criminal justice system “maintain[s] a fairly rigid rule of presumed prejudice” in the conflict of interest context). The reason for this lighter burden in conflict cases is clear. As the Supreme Court has noted, it “is difficult to measure the precise effect on the defense of representation corrupted by conflicting interests,” Strickland, 466 U.S. at 692, 104 S.Ct. at 2067, and accordingly, the Court has “refused to indulge in nice calculations as to the amount of prejudice attributable to the conflict.” Cuyler, 446 U.S. at 349, 100 S.Ct. at 1719 (internal quotation omitted). The majority, however, engages in such a calculation to determine whether a conflict even existed.
The question of whether there actually was a conflict plays an important role in separating cases where interests diverge — i.e., where the attorney places his own or another’s interest above the client’s interest — from those situations where the conflict remains potential. Whether the conflict actually affected the representation is a separate inquiry from the question of whether there was an actual conflict. Even when addressing whether a conflict had an adverse effect, the degree of prejudice caused by the conflict is not material once any real effect is shown. See Cuyler, 446 U.S. at 349,100 S.Ct. at 1719 (“[A] defendant who shows that a conflict of interest actually affected the adequacy of his representation need not demonstrate prejudice.”).12
*1294The rule that the majority espouses goes even further than that prohibited in evaluating actual effect. The majority collapses the question of effect into the question of actual conflict. Just as it is improper to evaluate a foregone strategy’s potential for success when determining whether there is an adverse effect, it is also improper to evaluate that strategy’s potential for success when determining whether there is an actual conflict of interest.13
III. WHICH STANDARD — CUYLER OR STRICKLAND ?
The majority holds that Strickland (rather than Cuyler) governs the analysis of Beets’s claim. The majority contends that Cuyler’s analysis is applicable only to conflicts stemming from multiple representation, and it opines that “Strickland offers a superior framework for addressing attorney conflicts outside the multiple or serial client context.” Beets, 65 F.3d at 1265. First, I disagree with the majority’s conclusion that Cuyler and the other Supreme Court cases addressing attorney conflicts support its decision to limit Cuyler to the multiple representation context.
Second, drawing on those cases and on some of the cases at the circuit level that apply Cuyler to attorney-client conflicts, I would apply Cuyler to a conflict between the attorney and his client which has a highly particularized and powerfully focused source, of a kind not frequently or normally encountered in the practice of law. It is these exceptional situations, where the divergence between the lawyer’s self-interest and his client’s interest poses an extraordinary threat to the lawyer’s duty of loyalty, that warrant the protection of Cuyler. As this court and other courts have recognized, the conflict stemming from a media rights contract is such a conflict, as are the conflict arising from the kind of contingent fee arrangement at issue in Winkler and the conflict arising from an attorney’s involvement in the allegedly criminal conduct of his client. If we reserve Cuyler for extraordinary attorney-client conflicts of that sort, not normally encountered in law practice, and we apply Strickland to alleged deficiencies in an attorney’s performance having their sources in the more common incidents of the attorney-client relationship, we avoid having the Cuyler exception swallow the Strickland rule. At the same time we preserve the benefit of the Cuyler inquiry for those exceptional cases that lie at the heart of the principles animating it.
A Conflict of Interest Jurisprudence
To test the majority’s hypothesis that Cuyler applies only to multiple representation cases, I look first at what the Supreme Court and other courts have said about attorney conflicts. It is well-settled that “[wjhere a constitutional right to counsel exists ... there is a correlative right to representation that is free from conflict of interest.” Wood, 450 U.S. at 271, 101 S.Ct. at 1103; see also Cuyler, 446 U.S. at 335, 100 S.Ct. at 1711; Holloway v. Arkansas, 435 U.S. 475, 98 S.Ct. 1173, 55 L.Ed.2d 426 (1978). The Supreme Court revisited Cuyler in Wood v. Georgia, 450 U.S. 261, 101 S.Ct. 1097, 67 L.Ed.2d 220 (1981), and it applied its framework to a conflict created by a third-party’s payment of *1295counsel. After examining the record, the Court noted that the defendants’ employer had paid for the defendants’ legal assistance, for the defendants’ bond fees, and for some of the other fines that the defendants incurred, but it had failed to pay the fines which resulted in the defendants’ incarceration. Wood, 450 U.S. at 267, 101 S.Ct. at 1101. The Court further observed that:
The fact that the employer chose to refuse payment of these fines, even as it paid other fines and paid the sums necessary to keep petitioners free on bond in this case, suggests the possibility that it was seeking — in its own interest — a resolution of the equal protection claim raised [in the case].
Id. (footnote omitted). The Court recognized that because the attorney was being paid by the employer, and was therefore the employer’s agent, there was a “clear possibility of conflict of interest.” Id. In light of this possibility, the Court remanded the case to the state court, instructing the lower court to apply the Cuyler framework and to determine “whether the conflict of interest that th[e] record strongly suggests actually existed at the time of the probation revocation or earlier.” Id. at 273, 101 S.Ct. at 1104.
The Supreme Court next discussed conflicts of interest in Strickland. In that case, the Court was called upon to determine the “proper standards for judging a criminal defendant’s contention that the Constitution requires a conviction ... to be set aside because counsel’s assistance at the trial ... was ineffective.” Strickland, 466 U.S. at 671, 104 S.Ct. at 2056.
Notably, when describing the standard for evaluating the prejudicial effect of a counsel’s failings, the Court distinguished ineffectiveness claims predicated on conflicts of interest. Specifically, the court noted that these claims warranted a limited presumption of prejudice, stating that “prejudice is presumed when counsel is burdened by an actual conflict of interest.” Strickland, 466 U.S. at 692, 104 S.Ct. at 2067 (citing Cuyler, 446 U.S. at 345-50, 100 S.Ct. at 1716-19). When there is an actual conflict, the Court emphasized that “counsel breaches the duty of loyalty, perhaps the most basic of counsel’s duties.” Id. Additionally, the Court found that a limited presumption of prejudice was warranted because “it is difficult to measure the precise effect on the defense of representation corrupted by conflicting interests.” Id. Further, the Court reasoned that “[g]iv-en the obligation of counsel to avoid conflicts of interest and the ability of trial courts to make early inquiry in certain situations likely to give rise to conflicts, it is reasonable for the criminal justice system to maintain a fairly rigid rule of presumed prejudice for conflicts of interest.” Id. (citation omitted).
The Supreme Court has not specifically addressed whether Cuyler applies to cases involving conflicts stemming from sources other than multiple representation. See Illinois v. Washington, 469 U.S. 1022, 1023, 105 S.Ct. 442, 443, 83 L.Ed.2d 367 (1984) (White, J., dissenting from denial of certiorari). Nevertheless, as the majority concedes, this court, as well as every circuit court facing the issue, has applied the rule of Cuyler to many types of conflicts of interest.14 In fact, the *1296Seventh, Ninth, and Eleventh Circuits have applied the Cuyler framework to conflicts stemming from media rights contracts. See United States v. Marrera, 768 F.2d 201, 205-09 (7th Cir.1985) (employing Cuyler framework to claim predicated on “conflict of interest between [the] lawyer’s financial interest in proceeds from the movie rights and [defendant’s] interest in acquittal”), cert. denied, 475 U.S. 1020, 106 S.Ct. 1209, 89 L.Ed.2d 321 (1986); Zamora v. Dugger, 834 F.2d 956, 960 (11th Cir.1987) (noting that “[t]he standard developed in Cuyler has been applied to cases in which defendants argue that their lawyers were more interested in publicity than in obtaining an acquittal,” and employing the Cuyler analysis); United States v. Hearst, 638 F.2d 1190, 1193 (9th Cir.1980) (recognizing that the conflict in Cuyler was based on multiple representation, and observing that the case before it was “based on private financial interests” of the lawyer, but applying Cuyler because “[t]hese differences are immaterial.”), cert. denied, 451 U.S. 938, 101 S.Ct. 2018, 68 L.Ed.2d 325 (1981).
Nevertheless, the majority boldly asserts that all of these other courts have misread Cuyler and the Supreme Court’s subsequent cases, stating that “[o]ne cannot read Cuyler [as] analyzing] conflicts of interest in a context broader than that of multiple client representation.” Beets, 65 F.3d at 1266. As noted above, however, the Supreme Court did just that in Wood, applying Cuyler to a conflict of interest stemming from the fact that defendants’ counsel was being paid by a third party. The majority attempts to distinguish this case by stating that the “lawyer was at least in the functional equivalent of a joint representation.... Both the theater and the employees expected him to advance their interests, yet to serve one might require him to fail the others, while doing nothing could harm both.” Beets, 65 F.3d at 1267. The majority forces Wood into the multiple representation category by focusing on the common denominator of all conflicts: divided loyalties or divergent interests between two or more entities.
B. Divided Loyalties: The Ethical Principles
The majority accurately notes that representation of two or more clients whose interests are best served by divergent litigation tactics presents a situation in which an attorney’s loyalties may be pulled in different directions by his various clients. When such a situation arises, an attorney may be forced to choose the interest of one client at the expense of the interest of the other client, or the attorney may choose to do nothing and neglect the interests of both clients. See Geoffrey C. Hazard & W. William Hodes, 1 The Law of Lawyering § 1.7:101 (2d ed. Supp.1992); Model Rules of Professional Conduct Rule 1.7, 1.9 (specifically addressing conflicts of interest arising from concurrent representation and serial representation).
Multiple representation situations, however, are not the only circumstances in which a conflict of interest may test an attorney’s duty of loyalty. A lawyer’s duty of loyalty may also be compromised when his own interests diverge from his client’s interests. See Wolfram, supra, § 7.1.3, at 317 (“The principle of loyalty runs throughout conflicts thinking but is most prominent in the areas of simultaneous conflicts and conflicts involving the lawyer’s personal interests.”). In fact, the general rule against conflicts of interest provides that “[a] lawyer shall not represent a client if the representation of that client may be materially limited by the lawyer’s responsibilities to another client or to a third person, or by the lawyer’s own interests.” Model Rules of Professional Conduct Rule 1.7(b); see also Wolfram, supra, § 7.1.2, at 315 (describing how an older version of the rules governing conflicts “deal[t] with two central situations — when a lawyer’s *1297personal interests clash with those of a client and when a lawyer represents at the same time clients with differing interests”). This potential for a conflict rooted in the attorney’s self-interest is so severe that the Model Rule of Professional Conduct 1.8 is devoted almost entirely to prohibitions and restrictions aimed at preventing such conflicts.15 The reason for these rules is clear. Just as an attorney’s loyalty may be pulled in different directions by clients’ divergent interests, an attorney’s loyalty can be sorely tested when his own self-interest runs counter to the interests of his client.
Thus, the majority’s attempt to draw the Cuyler line at multiple representation is ill-considered, for there is no logical reason why the distinction could not be used to classify all conflicts (including those involving the attorney’s self-interest) as “multiple representations.” Simply put, there is no intuitive reason why the Cuyler line should be drawn at conflicts where the interests of only third parties cause the divergence facing the attorney, as distinguished from conflicts where the interest of the attorney himself causes the divergence that he confronts. Indeed, there is a powerful intuitive reason why, in some situations, that line should not be (and has not been) drawn there. There are exceptional conflicts involving the attorney’s self-interest that, human nature being what it is, are far more likely to impair the lawyer’s ability to satisfy his duty of loyalty to his client than are the more ordinary conflicts between clients.
C. Where Should the Cuyler Line Be Drawn?
I recognize that not every conflict of interest pitting a lawyer’s self-interest against his client’s interests should trigger the analysis outlined in Cuyler. As one commentator notes, “[i]n a sense, every representation begins with a lawyer-client conflict. If the representation is for a fee, the lawyer’s economic interest will be to maximize the amount of the fee and the client’s will be to minimize it.” Wolfram, supra, § 7.1.1, at 313. Conversely, if the representation is for a flat fee, the attorney’s interest will be to minimize the amount of time spent on the case, and the client’s interest will be to maximize it. Similar conflicts inure in any contract for the sale of goods or services; the seller’s interest is to maximize the amount the buyer spends and minimize his own costs, and the buyer’s interest is to minimize the amount that he spends and maximize the quality of the goods or services.
Thus, the Cuyler exception would swallow the Strickland rule if it were applied to every case in which a criminal defendant complains that his lawyer failed to investigate a witness or a defense, neglected to perform an experiment, did not hire a witness, or otherwise failed to take action because the attorney decided that it was not worth the time or the expense. We have recognized that Cuyler is not meant to cover these types of cases. Strickland appropriately governs claims for failure to investigate16 and the like, and courts have had little difficulty in treating such claims under Strickland’s ineffectiveness rubric. See, e.g., Williams v. Calderon, 52 F.3d 1465, 1473 (9th Cir.1995) (refusing to apply Cuyler when a defendant alleged that “the fact that payment for any investigation or psychiatric services could have come from counsel’s pocket forced counsel to choose be*1298tween [the defendant’s] interests and his own”); United States v. Zackson, 6 F.3d 911, 921 (2d Cir.1993) (finding that Strickland, not Cuyler, was applicable to a claim that defense counsel was “plagued by a conflict of interest, namely that he was under enormous time constraints in regard to prior trial commitments” (internal quotations omitted)); Yohey v. Collins, 985 F.2d 222, 227 (5th Cir.1993) (finding that a failure to hire an expert was not a conflict in the Cuyler sense and applying Strickland to the alleged conflict).17
In addition to conflicts that are more properly treated under Strickland as claims about competence and diligence, there are other attorney-client conflicts frequently or normally encountered in the practice of law that will be better handled under Strickland. For example, the conflict claimed to exist in United States v. Sayan, 968 F.2d 55, 64-65 (D.C.Cir.1992), involving a lawyer who allegedly faded to request a continuance because he was afraid the judge would take action against him and his law firm if he made such a request, would arise with some frequency, as would the conflict claimed to exist in Zamora v. Dugger, 834 F.2d 956, 960 (11th Cir.1987), that the lawyer was more concerned with publicity than with his client’s fate. Both these charges can be made, with some credibility, in a good number of cases, and where they form the basis for a claim for post-conviction relief, they should be evaluated under Strickland.
While the great majority of alleged attorney-client conflicts arising in post-conviction proceedings — those frequently or normally encountered in the practice — will be better handled under Strickland, there are exceptional conflicts between an attorney’s self-interest and his client’s interest, stemming from highly particularized and powerfully focused sources, of the sort not normally encountered in law practice, that demand the application of Cuyler. A media rights contract is such a source,18 as are the kind of contingent fee arrangement at issue in Wink-ler and an attorney’s involvement in the allegedly criminal conduct of his client. These circumstances present situations so fraught with the temptation for the lawyer to sacrifice his chent’s best interest for his own benefit that they constitute particularly serious threats to the duty of loyalty. Not coincidentally, the Supreme Court and lower courts have applied the Cuyler presumption to these very types of cases.19
The majority posits that in these cases, unlike in the multiple representation context, the risk of prejudice is not plain, and that “[w]hen the duty of loyalty is challenged by an attorney’s self-interest, the range of possible breaches ... is virtually limitless.” Beets, 65 F.3d at 1271. I disagree. The risk is all too plain. Further, Cuyler has been *1299the law for fifteen years, and it cites precedents at the circuit level (including this circuit’s decision in Foxworth v. Wainwright, 516 F.2d 1072 (5th Cir.1975)), that are even older. The inescapable fact is that the courts have not had difficulty with the boundary problems described by the majority, as courts have been able to separate ordinary ineffective assistance claims (even those dressed in conflict language) from the exceptional cases that warrant the Cuyler standard. But even if we do encounter problems with cases at the boundaries, that is no reason to change the rule in a case that lies at the heart of the principles animating Cuyler.
In short, there is no authority whatsoever for limiting Cuyler to the multiple representation situation, and, as many courts have recognized, it makes no sense to do so in those exceptional cases where an attorney’s self-interest poses a serious threat to the duty of loyalty.
IV. SUMMARY
Under Cuyler, relief is proper on a Sixth Amendment claim of ineffective assistance of counsel when a defendant “demonstrate^] that an actual conflict of interest adversely affected his lawyer’s performance.” Cuyler, 446 U.S. at 348, 100 S.Ct. at 1718. In the instant case, Andrews was faced with an actual conflict because, while Beets’s interest lay in having Andrews withdraw and testify, Andrews’s interest lay in remaining as her counsel, because only then would he be entitled to the potentially lucrative media rights. Additionally, because Andrews did not withdraw and testify, Beets’s representation was adversely affected. A Sixth Amendment violation will be shown if the district court concludes that the conflict was the cause of Andrews’s failure to withdraw and testify. I would vacate the district court’s judgment and remand with instructions to resolve that issue. If the district court concludes that the conflict was the cause of Andrews’s failure to withdraw, then a Cuyler claim has been successfully established and the judgment granting the writ would be reinstated. If the district court concludes that the conflict was not the cause of Andrews’s failure to withdraw, then that failure should be evaluated under Strickland. Under that test, Beets has failed to show prejudice, i.e., that the result of her trial would have been different had Andrews withdrawn and testified. In that ease, the writ should be denied.
. The mobile home was Jimmy Don’s separate property, but until Jimmy Don's disappearance, Betty Lou and Jimmy Don resided in the mobile home together. After Jimmy Don disappeared, Betty Lou continued to reside in the mobile home.
. According to Randell Roberts's affidavit, after Beets was arrested, he and his brother “consulted with Mr. Andrews with respect to [their] further involvement in either of [the fire or the life] insurance matters. It was agreed that we would withdraw from further involvement in either matter...." Additionally, Randell Roberts noted that "[i]n deciding to withdraw from these matters my brother and I knew that we might be called to testify on behalf of Ms. Beets during her trial. We did not think that it would be appropriate for us to continue to represent her in the other matters if we were needed to testify on her behalf.” As noted above, Bruce Roberts did testify.
. The majority comments that "Andrews testified at the federal habeas hearing that this contract was signed after negotiations fell through to obtain his fee from Beets's children." Beets, 65 F.3d at 1261-62. This is technically true; the contract was not signed until after the trial began, and Andrews did note that he and one of Beets’s daughters "discussed finances prior to trial. It didn’t come through.” This, however, does not support the notion that the media rights contract was a last-second alternative to a fee. As noted above, the record clearly indicates that Andrews contemplated obtaining the media rights long before the trial started.
. Specifically, during his examination of Beets, the following colloquy took place:
Andrews: Whose suggestion was it that we try to collect retirement and insurance money?
Beets: I don't know that anybody suggested it.
Andrews: Was it some two years later?
Beets: Yes, it was about two years later.
Andrews: Was it a lawyer [who] suggested it?
Beets: I came to you.
Andrews: Did I send you ... to some other lawyer?
Beets: I talked to Randy Roberts in your office.
Andrews: Okay. Did you ever push me to just, "Let’s get that money. Let’s get that money and the whole bit.” Did you ever do that?
Beets: No, I didn't expect to get any of it. Additionally, near the conclusion of his examination of Beets, Andrews inquired whether Beets sought the settlement from the city “on my [Andrews’s] recommendation?” Beets replied yes.
. Similarly, the American Bar Association's Model Code of Professional Responsibility and Model Rules of Professional Conduct prohibit (and continue to forbid) a lawyer from obtaining media rights to his client’s case. See Model Rules of Professional Conduct Rule 1.8(d) (mandating that "[p]rior to the conclusion of representation of a client, a lawyer shall not make or negotiate an agreement giving the lawyer literary or media rights to a portrayal or account based in substan*1286tial part on information relating to the representation”); Model Code of Professional Responsibility EC 5-4 (mirroring the language of the Texas rule).
. See also Laws. Man. on Prof. Conduct, supra, at 51:702 (stating that the purpose of the prohibition is to "avoid the conflict of interest that would arise if the course of action that would further the client's cause would at the same time diminish the value of the lawyer’s publication rights"); Schuwerk and Sutton, supra, at 134 (noting that a lawyer who is the beneficiary of a media rights contract “may be tempted to take various actions in the representation of the client based on their effect upon the value of the publication rights.”).
. The Texas standards of professional conduct in effect at the time of Beets's trial frowned upon an individual's acting as counsel in a case when he may also be called upon to serve as a witness. Specifically, the rule provided that:
If, after undertaking employment in contemplated or pending litigation, a lawyer learns or it is obvious that he ... ought to be called as a witness on behalf of his client, he shah withdraw from the conduct of the trial and ... shall not continue representation in the trial.
State Bar Rules, art. X, § 9, DR 5-102 (1984). The Rule provided for certain exceptions as well. See Texas State Bar Rules, art. X, § 9, DR 5-101(B) (1984).
The rationale behind this rule is somewhat different from the reasons animating the prohibition against media rights contracts. Thus, while "the dual role [of advocate and witness] may be detrimental to the client’s interests because the lawyer may be more impeachable on grounds of bias,” it is also recognized that:
The most cogent rationale for the advocate-witness rule rests on protection of the fact-finding process. [The] adversary system works best when the roles of the judge, of the attorneys, and of the witnesses are clearly defined. Any mixing of those roles inevitably diminishes the effectiveness of the entire system.
Schuwerk & Sutton, supra, at 317-18 (internal quotations omitted). Other commentators agree, noting that:
The rationales [for rules limiting a lawyer by prohibiting media contracts and limiting a lawyer from acting as a witness] are simple. First, the attorney has built-in bias which must be argued to the finder of fact. Second, it is assumed that lawyers as advocates would bend the facts for the client or that the jury would give the lawyer’s testimony too much credence. Third, the lawyer-witness role may inhibit effective cross-examination.
Hall, supra, § 12.10, at 412 (footnotes omitted).
. Specifically the district court stated:
The possibility exists that the media rights contract motivated Andrews at a subconscious level to remain in the case when he should have withdrawn and testified for [Beets], To that limited extent, the two conflicts are intertwined. However, this Court has determined that the witness/advocate conflict is a separate conflict which did in fact adversely affect Andrews’[s] performance. This is sufficient under Cuyler without a detailed analysis of Andrews'[s] possible motivation.
. I recognize that the ultimate question whether a conflict of interest existed here is a mixed question of law and fact, see Strickland, 466 U.S. at 698, 104 S.Ct. at 2070 (citing Cuyler, 446 U.S. at 342, 100 S.Ct. at 1714-15), but its fact-bound components, such as whether Andrews’s failure to withdraw and testify was caused by the media rights contract, should be addressed in the first instance by the district court.
. To say that the inquiry whether an actual conflict exists contemplates an objective evaluation of the situation in which counsel is placed, i.e., that counsel's subjective perceptions cannot control the outcome of that inquiry, is supported by the Court’s conclusion that the question whether a conflict exists is a mixed question of law and fact. See Cuyler, 446 U.S. at 342, 100 S.Ct. at 1714-15.
. Until the panel's decision (which echoes in the majority opinion), there had been no question that the conflict inquiry is objective in the ordinary civil context. For example, in situations where a law firm is alleged to have a conflict of interest stemming from serial representation, whether the attorneys in that firm recognized or even knew about the conflict is not controlling. Instead, if an objective standard is met, i.e., if the “prior representations are substantially related to the present case,” then "the court will irrebuttably presume that relevant confidential information was disclosed during the former period of representation.” In re American Airlines, 972 F.2d 605, 614 (5th Cir.1992), cert. denied, - U.S. -, 113 S.Ct. 1262, 122 L.Ed.2d 659 (1993). In such cases, there is little doubt that a law firm could not avoid disqualification by protesting that it neither believed nor realized that it had a conflict of interest. See In re Martin, 817 F.2d 175, 182 (1st Cir.1987) (noting that in examining whether an actual conflict of interest exists, "[s]incerity or protestations of good faith, no matter how genuine, will not be enough. The test must be more [of] an objective one.").
. Circuit courts have frequently applied this principle. Thus, as the First, Second, and Third Circuits have noted:
[a defendant first] must demonstrate that some plausible alternative defense strategy or tactic might have been pursued. He need not show that the defense would necessarily have been successful if it had been used, but that it possessed sufficient substance to be a viable alternative. Second, he must establish that the alternative defense was inherently in conflict with or not undertaken due to the attorney’s other loyalties or interests.
Winkler, 7 F.3d at 309 (quoting Gambino, 864 F.2d at 1070 (alteration in original) (emphasis *1294added)); accord Fahey, 769 F.2d at 836; see also Foxworth v. Wainwright, 516 F.2d 1072, 1077 n. 7 (5th Cir.1975) (noting, before Cuyler or Strickland, that “[i]f an actual, significant, conflict is found ... the degree of prejudice is not to be considered.”).
. There had been some question in this court about the other aspect of the test articulated in Cuyler; that is, whether a petitioner was required to show that an actual conflict of interest actually affected his representation. In Baty v. Balkcom, 661 F.2d 391, 395 (5th Cir. Unit B Nov. 1981), cert. denied, 456 U.S. 1011, 102 S.Ct. 2307, 73 L.Ed.2d 1308 (1982), we stated our belief that "a requirement of proof of adverse effect of a conflict of interest on counsel, in addition to proof of an actual conflict, was not the intent of the [Supreme] Court in Cuyler." Baty, 661 F.2d at 397 n. 13. After the Supreme Court’s decision in Strickland, we revisited that determination, holding that "proof of some adverse effect is required before prejudice will be presumed from a showing that the attorney had an actual conflict of interest.” Nealy v. Cabana, 782 F.2d 1362, 1365 (5th Cir.), cert. denied, 479 U.S. 819, 107 S.Ct. 83, 93 L.Ed.2d 37 (1986). In none of those cases, however, did we suggest a different standard for determining the first prong of the Cuyler inquiry — whether there was an actual conflict.
. For cases applying Cuyler, see Garcia v. Bunnell, 33 F.3d 1193, 1198 n. 4 (9th Cir.1994) (applying the Cuyler standard to conflict created by attorney accepting job with prosecution office prior to trial, but noting that "[i]t is not logically necessary that the approach of [Cuyler ] also apply to conflicts between a defendant’s and the attorney's own personal interests; however, we conclude that precedent so requires”), cert. denied, — U.S. --, 115 S.Ct. 1374, 131 L.Ed.2d 229 (1995); Winkler v. Keane, 7 F.3d 304, 307 (2d Cir.1993) (applying Cuyler to conflict created by attorney working on contingency fee in criminal case), cert. denied, - U.S. -, 114 S.Ct. 1407, 128 L.Ed.2d 79 (1994); United States v. Sayan, 968 F.2d 55, 64-65 (D.C.Cir.1992) (upholding application of Cuyler's adverse effect test to alleged conflict created by lawyer’s fear of antagonizing judge); United States v. Michaud, 925 F.2d 37, 40 (1st Cir.1991) (analyzing conflict of interest stemming from attorney’s association with prosecuting IRS under Cuyler framework); United States v. Horton, 845 F.2d 1414, 1418-21 (7th Cir.1988) (applying Cuyler to conflict generated by defense attorney’s candidacy for U.S. Attorney); United States v. Andrews, 790 F.2d 803, 811 (10th Cir.1986) (finding that Cuyler applies in situations involving "counsel's ability to represent his client fairly, loyally or impartially”), cert. denied, 481 U.S. 1018, 107 S.Ct. 1898, 95 L.Ed.2d 505 (1987); Roach v. Martin, 757 F.2d 1463, 1479 (4th Cir.) (applying Cuyler when alleged conflict of interest was rooted in fact that defense attorney was under investigation by state *1296bar grievance committee), cert. denied, 474 U.S. 865, 106 S.Ct. 185, 88 L.Ed.2d 154 (1985); Ware v. King, 694 F.2d 89, 92 (5th Cir.1982) (per curiam) (using Cuyler framework to analyze claim of conflict of interest stemming from separate civil and criminal lawsuits pending between defense counsel and prosecutor), cert. denied, 461 U.S. 930, 103 S.Ct. 2092, 77 L.Ed.2d 302 (1983); United States v. Knight, 680 F.2d 470, 471 (6th Cir.1982) (per curiam) (undertaking Cuyler analysis in evaluating claim of conflict of interest stemming from attorneys’ knowledge that they were under investigation for stealing documents during trial), cert. denied, 459 U.S. 1102, 103 S.Ct. 723, 74 L.Ed.2d 950 (1983).
. For example, Model Rule 1.8(a) restricts an attorney from entering into business transactions with a client. Similarly Rule 1.8(d) prohibits an attorney from acquiring media rights from a client prior to the conclusion of the representation of that client. Rule 1.8(f) severely restricts the ability of an attorney to receive compensation from someone other than his client, and Rule 1.5(d)(2) prohibits attorneys from entering into contingent fee arrangements in criminal cases. See also State Bar Rules, art. X, § 9, DR 5-101(A) (Texas Code of Professional Responsibility) (1984) ("[A] lawyer shall not accept employment if the exercise of his professional judgment on behalf of his client will be or reasonably may be affected by his own financial, business, property, or personal interests."); Hazard & Hodes, supra, § 1.8:101 (noting that many of the transactions prohibited in Model Rule 1.8 “involve transactions in which the lawyer's own self-interest threaten to adversely affect the quality of the representation to be provided”).
. Strickland itself was a claim for failure to investigate, and in adopting its test, the Court noted that Strickland’s standard ”require[d] no special amplification in order to define counsel's duty to investigate.” Strickland, 466 U.S. at 690, 104 S.Ct. at 2066.
. In differentiating between conflicts that merit the more stringent test of Strickland and conflicts that warrant the Cuyler methodology, courts may, as instructed to do in Strickland when determining the reasonableness of attorney conduct, receive some guidance from prevailing norms of professional responsibility. Ineffective assistance claims rooted in the failure to investigate or to devote proper attention to a case have rightfully been treated under the uncontroversial standards of competence and diligence. See Strickland, 466 U.S. at 690, 104 S.Ct. at 2065-66; Zackson, 6 F.3d at 921; Model Rules of Professional Conduct Rule 1.1. (“A lawyer shall provide competent representation to a client. Competent representation requires the legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation.”); Model Rules of Professional Conduct Rule 1.3 ("A lawyer shall act with reasonable diligence and promptness in representing a client.").
. The majority states that "[t]he dissent has agreed that a witness/advocate conflict alone is not the sort that even under their approach should be governed by a Cuyler inquiry.” Beets, 65 F.3d at 1277. To the contrary, a witness/advocate “conflict” having its source in a media rights contract, as may be the case here, is exactly the kind of conflict that should be governed by Cuyler.
.For examples of these cases see Wood, 450 U.S. at 271-72, 101 S.Ct. at 1103-04 (applying Cuyler to a case in which a defense attorney was paid by a third party with a possibly conflicting interest); Winkler, 7 F.3d at 308 (using Cuyler in a case where a criminal defense attorney was paid on a contingency fee basis); Marrera, 768 F.2d at 207 & n. 6 (employing Cuyler's framework to a conflict based on a lawyer’s financial interest in media rights); Hearst, 638 F.2d at 1193 (same); Acevedo, 891 F.2d at 610-11 (employing the Cuyler test to a situation in which an attorney may be involved in the criminal conduct that his client is alleged to have committed).