dissenting:
A procedure designed to avoid a simple bookkeeping error, that avoids the error all but once in twenty years, is at least “reasonably adapted” to avoid that error. The majority disagrees. I dissent.
Here, Arthur Engelen sued Rosen & Loeb because Rosen garnished Engelen’s wages to collect a debt that Engelen had already paid. The garnishment violated Engelen’s rights under the federal Fair Debt Collection Practices Act and an analogous state law. Rosen, otherwise conceding liability, moved for summary judgment, invoking the statute’s bona fide error defense.
That defense allows debt collectors like Rosen to avoid judgment if they can show that their violations were “not intentional and resulted from a bona fide error notwithstanding the maintenance of procedures reasonably adapted to avoid any such error.” 15 U.S.C. § 1692k(c). Our case law divides that showing into three parts. The debt collector must show that the error (1) was unintentional, (2) bona fide, meaning made in good-faith, and (3) resulted notwithstanding “reasonably adapted” procedures to avoid it. McCollough v. Johnson, Rodenburg & Lauinger, LLC, 637 F.3d 939, 948 (9th Cir.2011). The district court faithfully applied this case law, granted Rosen’s motion, and should be affirmed.
Rosen’s evidence was simple but convincing. Alan Rosen is an experienced San Diego lawyer whose law firm, Rosen & Loeb, collects commercial debts for creditors. It was hired to collect Enge-len’s debt by Erin Capital Management, an enterprise that buys uncollected debts in bulk. Rosen adopted a brief, written procedure to collect the debts now owed to Erin, including Engelen’s. Rosen employed that procedure for recovering Erin’s debts, but did not create it especially for Erin. Rosen had used the same procedure with other similar clients for years and repeatedly trained its employees on it. Rosen’s employees were also regularly trained, tested, and then retrained based on the test results on how to comply with debt collection laws generally.
Rosen’s procedure worked like this: first, Erin would receive a payment. Then, Erin would notify Rosen that the payment information was available for download on Erin’s website. Rosen’s office manager would download the payment information, save it to Rosen’s computers, and print-off a copy. The office manager would then hand deliver the printed copy to Rosen’s bookkeeper, who would then enter the payments into Rosen’s own software. If a debtor’s balance zeroed after the payment information was entered, then Rosen automatically stopped collection efforts. Periodically, the office manager would spot check the bookkeeper’s work, making sure that payments were accurately credited.
This procedure should have prevented Rosen’s unlawful garnishment of Engelen’s wages. Erin did, in fact, place notice on its website of Engelen’s full payment of his debt; Rosen’s office manager did, in fact, download that payment information, print it, and deliver it to the bookkeeper. But, for some unknown reason, the bookkeeper simply failed to record the payment. If she had, Engelen’s account would have zeroed, and Rosen would not have initiated the garnishment. But for a single, lost print-out, we wouldn’t be here.
And losing that print-out was an anomaly. Rosen himself declared and testified that the procedure had worked without fail until the firm received the Engelen debt and has worked without fail since-a period of more than twenty years with only a single mistake. The officer manager and *711bookkeeper both backed Rosen’s claim with their own sworn declarations. And Engelen produced no contrary evidence.
This anomaly was a bona fide error. It was unintentional, it was in good faith, and it occurred notwithstanding procedures “reasonably adapted” to prevent it. My colleagues appear to agree that Rosen’s error was unintentional and bona fide. We part company over whether Rosen’s procedures were “reasonably adapted.” I explain why I believe they were.
First, Rosen’s procedures were reasonably adapted to avoid the error because, had they been used correctly, they would have avoided the error. Had the bookkeeper entered Engelen’s payment, his account would have zeroed and Rosen would have stopped all collection activities. The unlawful garnishment would thus never have happened.
Second, a procedure that only fails once in twenty years-i.e., once in likely thousands or tens of thousands of applieations-is reasonably adapted to avoid the errors that would result from its failure. The majority tries to turn this single mistake into several by noting that “ten other recording errors occurred on the same day as the error that affected Engelen.” The majority doesn’t acknowledge that its eleven errors all have the same genesis. The print-out with Engelen’s payment information also contained information for 10 other accounts.
In my book, a single, lost print-out is one error. But my analysis would be the same even under the majority’s semantics — eleven errors in twenty years is still overpowering evidence of reasonably adapted procedures. By indicating that even this record isn’t good enough, the majority sends a chilling message to debt collectors in our jurisdiction: reasonableness requires less than one error every twenty years; reasonable means perfect. This ignores the statutory defense Congress provided to this otherwise strict liability statute.
Third, our prior cases are all either off point or, to the extent relevant, support Rosen, In Evon v. Law Offices of Sidney Mickell, 688 F.3d 1015, 1030 (9th Cir.2012), the error was intentional and there were no avoidance procedures in place. In McCollough v. Johnson, Rodenburg & Lauinger, LLC, 637 F.3d 939, 948 (9th Cir.2011), the debt collector’s procedures were inadequate because, unlike Rosen’s, they would not have prevented the error at issue. And in both Clark v. Capital Credit & Collection Servs., Inc., 460 F.3d 1162, 1177 (9th Cir.2006), and Fox v. Citicorp Credit Servs., Inc., 15 F.3d 1507, 1514 (9th Cir.1994), the debt collectors offered no evidence of the existence of any avoidance procedures at all.
Reichert v. Nat’l Credit Sys., Inc., 531 F.3d 1002, 1006 (9th Cir.2008), faults the debt collector for relying on a single, “con-clusory” declaration to prove-up its procedures. Rosen’s evidence was simple, but far from eonclusory. Reichert also approvingly cites two out-of-circuit cases that favor Rosen. Id. One approves of regular training like Rosen’s. Jenkins v. Heintz, 124 F.3d 824, 834-35 (7th Cir.1997). And the other makes the common-sense observation that a debt collector need not employ complicated procedures to address simple problems. Wilhelm v. Credico, Inc., 519 F.3d 416, 421 (8th Cir.2008).
The most factually similar, circuit-level authority is Smith v. Transworld Systems, Inc., 953 F.2d 1025, 1031 (6th Cir.1992). There, as here, the debt collector violated the law by taking a collection action after it received information that should have stayed its hand. Id. And there, as here, the debt collector made a simple showing — two employee declarations and a five-*712page procedures manual. Id. (Rosen’s manual was shorter but he has an extra declaration.) That was enough for the court to conclude that the problem was “at most a clerical error” and to affirm summary judgment. We should do the same.
In reversing, the majority relies on three arguments. First, the disposition concludes that, other than the office manager’s “periodic spot-checking, none of Ro-sen & Loeb’s procedures was aimed at preventing wrongful wage garnishments caused by the bookkeeper’s failure to record payment information.” This is wrong. It’s wrong because Rosen’s other procedures — the procedure for downloading, printing, and inputting payment notifications and the monthly trainings on those procedures-are all meant to avoid unlawful wage garnishments by preventing the bookkeeper from failing to input payment information in the first place. The majority’s failure to credit all of Rosen’s procedures is unexplained and unfair.
Also unfair is the majority’s assertion that the “the periodic spot-checking did not consistently reduce the likelihood of recording errors.” How they reached that conclusion on a record of one error in twenty years is a mystery. Their cite to Jerman v. Carlisle, 559 U.S. 573, 587, 130 S.Ct. 1605, 176 L.Ed.2d 519 (2010), is meaningless. Of course the spot-checking, considered in isolation, is not itself a “mechanical” process with “orderly” steps. No one ever said it was. Rather, it is but one orderly step in the mechanical process Rosen follows: download, print-out, hand-off, input, spot-check.
The majority’s next complaint is that Rosen “presented no evidence of any regular redundancies designed to catch human recording errors.” But what is the office manager’s spot-checking if not a redundancy? And why, without argument or authority, does the majority seem to insist that Rosen — and therefore all debt collectors — spot-check its spot-checking? The majority does not explain why the absence of a second redundancy, or perhaps a more regular one, makes Rosen’s procedure unreasonable.
Finally, the majority finds that Rosen’s procedures were unreasonable because they did not require, prior to seeking the writ of garnishment, contacting Erin or contacting the state court that issued the writ. This is just more spot-checking of spot-checking. Rosen did have a procedure for communicating with Erin, it just happened to break down this one time; but not because Erin forgot to tell Rosen that Engelen had paid the debt. As to the court, the record is silent, but at oral argument it was disputed whether checking with the court would have helped. That the court issued the writ on a satisfied debt like Engelen’s suggests that consulting it may not have proved useful. Moreover, the majority offers nothing — no authority or reason — to explain why Ro-sen’s procedures are unreasonable because they did not require these steps.
Of course, on de novo review of summary judgment we draw all inferences from the evidence in the non-movant’s favor, here Engelen’s. See Anderson v. Liberty Lobby, 477 U.S. 242, 248, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986). But drawing those inferences doesn’t permit us to ignore evidence or make it up. We cannot ignore Rosen’s uncontradicted evidence that its procedures never failed before En-gelen. We can infer, from that evidence, that Rosen is simply unaware of any possible prior failures because no one ever complained. But we cannot invent prior failures to justify deeming Rosen’s procedures unreasonable.
Thus, even when properly discounted, Rosen’s evidence establishes that its procedures are reasonably adapted. The law *713contains an exception to liability for bona fide mistakes. I would affirm because no reasonable juror could find against Rosen on its bona fide error defense, even after drawing all reasonable inferences from the evidence in Engelen’s favor. Reasonable just means reasonable, not perfect. If this isn’t a bona fide error, I don’t know what would be.
I respectfully dissent.