delivered the opinion of the Court,
in which Justice Johnson, Justice Willett, Justice Guzman, Justice Lehrmann, and Justice Devine joined.This appeal involves an excéss insurance policy that covered multiple commercial properties and limited the coverage to “the least” of three alternative amounts. When fifteen of the covered properties were damaged in one occurrence, the insurer calculated “the least” of the three alternative limits'separately for each covered item at each damaged property, on an item-by-item basis. The insured contends that “the least” of the three limits applies just once in any one occurrence to the total of all losses from all covered items at all of the damaged properties. The trial court agreed with the insurer, and a split court of appeals agreed with insured. We conclude that both constructions are reasonable and the policy is therefore ambiguous. Because we construe an insurance policy’s ambiguous language in favor of coverage for the insured, we affirm.
I.
Background
The Lynd Company manages the insurance needs of more than 100 commercial properties located in eleven different states. In 2005, Lynd purchased two layers of insurance to cover the properties, which Lynd ultimately valued at a total of about $1 billion. The first layer, a primary policy from Westchester Fire Insurance Company, provided coverage up to $20 million per occurrence. The second layer, an excess policy from RSUI Indemnity Company, covered losses that exceeded $20 million, up to $480 million per occurrence. Instead of identifying the covered properties, the RSUI policy required Lynd to provide a list of the properties to RSUI before the policy became effective. The policy’s “Scheduled Limit of Liability” endorsement referred to this list as the “Statement of Values.” In addition to the properties’ names, addresses, and other details, the Statement of Values included the values that Lynd or the properties’ owners assigned to each of three types or “items” of coverage the policy provided at each location: (1) the building replacement value, (2) the value of the building’s contents, and (3) the value of one year’s rental income from the property.
The Scheduled Limit of Liability endorsement provided that the policy’s premium would be based on the values that *117Lynd reported in the Statement of Values. The policy also included a Reporting Endorsement, which identified the premium rate as $0,025 per $100 of reported value. The Reporting Endorsement required Lynd to update the Statement of Values quarterly, and sooner for any newly acquired properties that Lynd valued over $1 million. It also provided that RSUI would recalculate the premium every quarter by applying the same $0.025/$100 rate to all values reported for that quarter. Throughout the policy’s term, as Lynd added additional properties or increased the values of properties it had previously reported, the parties added four separate endorsements to confirm quarterly increases in the reported values and premium.
The Scheduled Limit of Liability endorsement also provided that the policy would only cover losses “at the locations listed on the latest Statement of Values” and only if “a value is shown for [the] scheduled item.” But the policy did not require Lynd to list any particular properties or report the values with any particular accuracy. As a result, Lynd effectively controlled the amount of the premium it paid through the values it reported in the Statement of Values. For its highest-valued property, Lynd listed about $18.2 million for building replacement, $10,000 for contents, just over $4-million for rental income, and a total insured value of about $22.3 million. It was thus at least possible that a single occurrence could cause losses at that property in excess of Westchester’s primary policy’s $20 million limit. None of the other listed properties had a total insured value exceeding $20 million. But ■with the exception of the one most valuable property, Westchester’s $20 million primary policy was sufficient to cover all losses that a single occurrence could cause to any one property. In other words, only an occurrence that was catastrophic enough to damage more than one of the listed properties was likely to trigger coverage under RSUI’s excess policy.
In September 2005, the Gulf Coast experienced such an occurrence: Hurricane Rita. The storm damaged fifteen of the properties that Lynd had listed in the Statement of Values. The parties agree that a single “occurrence” caused all of the damage and that the combined total of Lynd’s resulting losses was just over $24.5 millibn. After Westchester paid its $20 million limit, RSUI refused to pay the remaining $4.5 million and instead paid Lynd about $750,000; In calculating this amount, RSUI included the actual adjusted amounts of the losses incurred at thirteen of the properties, because those amounts were less than 115% of the values Lynd had listed for those items in the Statement of Values. But one property, the Willow Bend apartment complex, incurred about $5 million in losses, which was about $2 million more than the total insured value that Lynd listed for that property, and another, the Le Chateau apartment complex, sustained about $11 million in losses, nearly $3.5 million more than Lynd’s reported total value. For these two propertiés, RSUI paid 115% of the reported values, rather than the actual adjusted amount of the loss. The parties dispute whether the policy limited RSUI’s liability in that manner.
Lynd filed this suit against RSUI to recover the difference between its $24.5 million in losses and the $20,750,000 that Westchester and RSUI had paid, plus statutory .prompt-payment penalties, prejudgment interest, and attorney’s feés. The parties stipulated to the amounts at issue: about $4.2 million in disputed coverage, $2.5 million in penalties, and $866,000 in interest and attorney’s fees, totaling just over $7.5 million. On cross-motions for summary judgment, the trial court agreed *118with RSUI’s construction of the policy and ordered that Lynd recover nothing. The court of appeals agreed with Lynd’s construction, reversed, and rendered judgment awarding Lynd the full $7.5 million. That court denied RSUI’s motion for rehearing en banc, but three justices dissented, and one of the remaining four wrote separately to urge this Court to grant review and resolve the parties’ dispute.
II.
Governing Legal Principles
An insurance policy is a contract, generally governed by the same rules of construction as all other contracts. Gilbert Tex. Constr., L.P. v. Underwriters at Lloyd’s London, 327 S.W.3d 118, 126 (Tex.2010). When construing a contract,' our primary concern is to ascertain the intentions of the parties as expressed in the document. Amedisys, Inc. v. Kingwood Home Health Care, LLC, 437 S.W.3d 507, 514 (Tex.2014). We begin our analysis with the language of the contract because it is the best representation of what the parties mutually intended. Gilbert Tex. Constr., 327 S.W.3d at 126; see also Anglo-Dutch Petroleum Int'l, Inc. v. Greenberg Peden, P.C., 352 S.W.3d 445, 451 (Tex.2011). Unless the policy dictates otherwise, we give words and phrases their ordinary and generally accepted meaning, reading them in context and in light of the rules of grammar and common usage. See Gilbert Tex. Constr., 327 S.W.3d at 126; Forbau v. Aetna Life Ins. Co., 876 S.W.2d 132, 133 (Tex.1994). We strive to give effect to all of the words and provisions so that none is rendered meaningless. See Gilbert Tex. Constr., 327 S.W.3d at 126; Forbau, 876 S.W.2d at 133. “No one phrase, sentence, or section [of a contract] should be isolated from its setting and considered apart from the other provisions.” Forbau, 876 S.W.2d at 134 (quoting Guardian Trust Co. v. Bauereisen, 132 Tex. 396, 121 S.W.2d 579, 583 (1938)).
When construing an insurance policy, we are mindful of other courts’ interpretations of policy language that is identical or very similar to the policy language at issue. Trinity Universal Ins. Co. v. Cowan, 945 S.W.2d 819, 824 (Tex.1997). “Courts usually strive for uniformity in construing insurance provisions, especially where ... the contract provisions at issue are identical across the jurisdictions.” Nat’l Union Fire Ins. Co. of Pittsburgh, Pa. v. CBI Indus., Inc., 907 S.W.2d 517, 522 (Tex.1995); see also Zurich Am. Ins. Co. v. Nokia, Inc., 268 S.W.3d 487, 496-97 (Tex.2008) (“We have repeatedly stressed the importance of uniformity ‘when identical insurance provisions-will necessarily be interpreted in various jurisdictions.’ ”) (quoting Cowan, 945 S.W.2d at 824).
RSUI and Lynd offer conflicting constructions of the Scheduled Limit of Liability endorsement. If only one party’s construction is reasonable, the policy is unambiguous and we will adopt that party’s construction. See Grain Dealers Mut. Ins. Co. v. McKee, 943 S.W.2d 455, 459 (Tex.1997). But if both constructions present reasonable interpretations of the policy’s language, we must conclude that the policy is ambiguous. See id. at 458; Balandran v. Safeco Ins. Co. of Am., 972 S.W.2d 738, 741 (Tex.1998). In that event, “we must resolve the uncertainty by adopt ing the construction that most favors the insured,” and because we are construing a limitation on coverage, we must do so “even if the construction urged by the insurer appears to be more reasonable or a more accurate reflection of the parties’ intent.” Nat’l Union Fire Ins. Co. of Pittsburgh, Pa. v. Hudson Energy Co., 811 S.W.2d 552, 555 (Tex.1991). “This widely followed rule is an outgrowth of the gener*119al principle that uncertain contractual language is construed against the party selecting that language,” and is “justified by the special relationship between insurers and insureds arising from the parties’ unequal bargaining power.” Balandran, 972 S.W.2d at 741 n.1 (citing Steven Putt, et al., 2 Couch on Insurance § 22.14 (3d ed. 1997); Arnold v. Nat’l Cnty. Mut. Fire Ins. Co., 725 S.W.2d 165, 167 (Tex.1987)).
In contract law, the terms “ambiguous” and “ambiguity” have a more specific meaning than merely denoting a lack of clarity in language. Universal C.I.T. Credit Corp. v. Daniel, 150 Tex. 513, 243 S.W.2d 154, 157 (1951). “An ambiguity does not arise simply because the parties offer conflicting interpretations.” Am. Mfrs. Mut. Ins. Co. v. Schaefer, 124 S.W.3d 154, 157 (Tex.2003). Instead, “a contract is ambiguous only when the application of pertinent rules of interpretation to the face of the instrument leaves it genuinely uncertain which one of two or more meanings is the proper meaning.” Daniel, 243 S.W.2d at 157; see Balandran, 972 S.W.2d at 741. Thus, a contract is ambiguous only if, after applying the rules of construction, it remains “subject to two or more reasonable interpretations.” Balandran, 972 S.W.2d at 741. Our task in this case is to determine whether Lynd’s construction of the RSUI policy is reasonable. If it is, we must enforce that construction, even if RSUI’s construction is also reasonable.
III.
Construing the Scheduled Limit of Liability Endorsement
The Scheduled Limit of Liability endorsement provides in full:
This Endorsement Changes The Policy. Please Read It Carefully.
SCHEDULED LIMIT OF LIABILITY
This endorsement modifies insurance provided under the following:
ALL COVERAGE PARTS
It is understood and agreed that the following special terms and conditions apply to this policy:
1. In the event of loss hereunder, liability of the Company shall be limited to the least of the following in any one “occurrence”:
a. The actual adjusted amount of the loss, less applicable deductibles and primary and underlying excess limits;
b. 115% of the individually stated value for each scheduled item of property insured at the location which had the loss as shown on the latest Statement of Values on file with this Company, less applicable deductibles and primary and underlying excess limits. If no value is shown for a scheduled item then there is no coverage for that item; or
c. The Limit of Liability as shown on the Declarations page of this policy or as endorsed to this policy.
2. Coverage under this policy is provided only at the locations listed on the latest Statement of Values on file with this Company or as endorsed on to this policy.
3. The premium for this policy is based upon the Statement of Values on file with this Company or attached to this policy.
The term “occurrence”, where used in this policy, shall mean any one loss, disaster, casualty or series of losses, disasters, or casualties arising from one event.
When the term “occurrence” applies to a loss or series of losses from the perils of tornado, cyclone, hurricane, windstorm, *120hail, flood, earthquake, volcanic eruption, riot, riot attending a strike, civil commotion and vandalism and malicious • mischief, one event shall be construed to be all losses arising during a continuous period of 72 hours. When filing a proof of loss the insured may elect the moment at which the 72 hour period shall be deemed to have commenced, which shall not be earlier than the first loss to occur at any covered location.
Paragraph 1 of this endorsement describes three alternative liability limits and provides that RSUI’s liability “shall be limited to the least of the [three] in any one ‘occurrence.’ ” The parties agree that all of the losses at issue resulted from one occurrence. The parties disagree, however, on how to compare and apply the three alternative limits when one occurrence causes losses at multiple locations. RSUI compared the three alternatives and applied “the least” of them, but it did so multiple times, once for each of the coverage items at each of the damaged locations. At thirteen properties, it applied the limit under Part la (the “actual adjusted amount of the loss”), because that amount was less than the amounts under Part lb (“115% of the individually stated value for each scheduled item”) and Part lc ($480 million). But for Willow Bend and Le Chateau, it applied Part lb’s limit (115%' of the stated values) because that amount was less than the limits under Part la and Part lc. Lynd contends that RSUI must compare the three alternatives and apply “the least” of them only once, using the total amounts of the losses and stated values related to “any one ‘occurrence,’ ” and should have applied the Part la limit here because that amount was less than the amounts under Parts lb and lc. So Lynd asserts that RSUI should have paid the $24.5 million — the “actual adjusted amount” of all losses at all of the properties, including Willow Bend and Le Chateau — less the policy’s $25,000 deductible and the Westchester policy’s $20 million limit.
The court of appeals agreed with Lynd, and RSUI petitioned for our review. RSUI contends that the court of appeals erred because (A) consistent with insurance industry practices, the “structure” of the policy confirms that it is a “scheduled” policy rather than a “blanket” policy or some “unknown hybrid” of the two; (B) the only reasonable construction of the plain language of the policy is that the alternative limits apply on an item-by-item basis; and (C) Lynd’s construction conflicts with the vast majority of other courts’ constructions of the same or similar language and makes Texas an outlier on this issue. We begin our analysis with the policy’s language, and then address the remaining arguments in turn.
A. The Language of the Endorsement
RSUI and Lynd both contend that the plain language of the Scheduled Limit bf Liability endorsement, considered within the context of the policy as a whole, supports only their proposed construction. We have carefully considered the endorsement, including its title, its cautionary statements, Paragraph l’s introductory statement, the three alternative limits, and all of the language as a whole. Having done so, we conclude that the endorsement reasonably can be read to support either party’s proposed construction and is therefore ambiguous.
1. The Title
The endorsement is titled “Scheduled Limit of Liability.” (Emphasis added.) The policy does not define the term “scheduled,” but RSUI contends that it is widely accepted within the insurance industry that a “scheduled” policy provides “scheduled coverage,” meaning coverage *121that is limited on an item-by-item basis. This, RSUI contends, “is the ‘hallmark’ of a scheduled policy.” Because this endorsement is titled “Scheduled Limit of Liability,” RSUI asserts, it necessarily limits the coverage on an item-by-item basis. (Emphasis added.) Lynd agrees that “scheduled” is a term of art within the insurance industry, but disputes that “scheduled” policies necessarily limit coverage on an item-by-item basis when one occurrence damages multiple items. We will address this dispute about industry practices further below.1 (Emphasis added.) For now, we focus on the policy’s language.
We agree that the “title, like every other portion of a contract, may be looked to in determining its meaning,” E.H. Perry & Co. v. Langbehn, 113 Tex. 72, 252 S.W. 472, 474 (1923), because headings and titles provide context and can inform the meaning of the sections they label. See, e.g., Michiana Easy Livin’ Country, Inc. v. Holten, 168 S.W.3d 777, 793 (Tex.2005) (declining to construe a provision as limited to one topic when “the title of the paragraph show[ed] that two different topics were addressed in it”); see also Reliance Ins. Co. v. Orleans Parish Sch. Bd., 322 F.2d 803, 807 (5th Cir.1963) (noting “considerations which persuade us that the parties intended what the special form plainly stated — ‘School Form-Blanket’ policy”). Generally, courts should construe contractual provisions in a man-' ner that is consistent with the labels the parties have given them.
Assuming that the word “scheduled” in the endorsement’s title means “limited on an item-by-item basis,” as RSUI contends, we agree that the title of the Scheduled Limit of Liability endorsement provides some support for RSUI’s construction. Nevertheless, although “courts may consider the title of a contract provision or section to interpret a contract, ‘the greater weight must be given to the operative contractual clauses of the agreement.’ ” Enter. Leasing Co. of Hous. v. Barrios, 156 S.W.3d 547, 549 (Tex.2004) (quoting Necee v. A.A.A. Realty Co., 159 Tex. 403, 322 S.W.2d 597, 600 (1959)). Thus, titles and headings are not determi native, and when they are inconsistent with the plain meaning of the provision’s operative language, we afford greater weight to the operative language. So we must consider the endorsement’s operative language, to which we afford the greater import, to determine whether it is consistent only with the meaning that RSUI attributes to the title. See Barrios, 156 S.W.3d at 549.
2. The Cautionary Language
The Scheduled Limit of Liability endorsement contains cautionary language immediately . above and below the title: “This Endorsement Changes The Policy. Please Read It Carefully,” and “This endorsement modifies insurance provided under .... ALL COVERAGE PARTS.” The *122parties agree that RSUI’s excess policy is a “follow form” policy, meaning that it is generally subject to the terms and conditions of Westchester’s primary policy except where RSUI’s policy expressly modifies those terms.2 See, e.g., GenCorp, Inc. v. Am. Int’l Underwriters, 178 F.3d 804, 809 n.3 (6th Cir.1999) (describing “follow form” excess policies); KLN Steel Prods. Co. v. CNA Ins. Cos., 278 S.W.3d 429, 443 (Tex.App.-San Antonio 2008, pet. denied) (same). The parties also agree that West-chester’s primary policy was not a “scheduled” policy. So the cautionary language could be read to mean that, even though Westchester’s policy was not a scheduled policy, and even though RSUI’s policy was a follow form policy, the endorsement confirms that the RSUI policy differs from the Westchester policy in the limits it imposes. We agree with this reading of the cautionary language, but it only begs the question of how the RSUI’s policies’ limits differ. Both parties agree that the Scheduled Limit of Liability endorsement imposes coverage limits that the Westchester policy does not impose, but they disagree on when and how those limits apply. The cautionary language does not address those questions, and therefore does not indicate which party’s construction of the endorsement is correct.
3. Paragraph l’s Introductory Statement
The first sentence of Paragraph 1 of the Scheduled Limit of Liability endorsement imposes the three alternative coverage limits “in the event of loss” under the policy. The policy’s Excess Physical Damage Coverage Form defines the term “loss” to mean “a loss or series of losses arising out of one event or occurrence.” (Emphasis added.) The term loss in Paragraph 1 may thus refer to either “a loss” or to the entire “series of losses” that resulted from Hurricane Rita, which the parties agree was “one event or occurrence.” In other words, we could read the phrase to mean that the alternative limits apply “in the event of [each loss],” as RSUI proposes, but we could also read it to mean that they apply “in the event of [the entire series of losses],” as Lynd proposes. Because the policy defines the term “loss” to mean “a loss” or a “series of losses,” we conclude that the phrase “in the event of loss,” standing alone, is equally consistent with both parties’ constructions of the endorsement.
The remainder of Paragraph l’s introductory statement limits RSUI’s liability to the least of the three alternative limits “in any one ‘occurrence.’ ” The endorsement defines the term “occurrence” to “mean any one loss, disaster, casualty or series of losses, disasters, or casualties arising from one event.” It then provides that when the occurrence is a hurricane or other significant peril, “one event” includes “all losses” arising during a 72-hour period. (Emphasis added.) Thus, at least for purposes of a significant peril, the policy defines the word “occurrence” essentially the same way it defines the word “loss,” uses the word “occurrence” to define the word “loss,” uses the word “loss” or “losses” to define both “occurrence” and *123“event,” and uses the word “event” to define both “loss” and “occurrence”:
LOSS: “a loss or series of losses arising out of one event or occurrence.”
OCCURRENCE: “any one loss ... or series of losses ... arising from one event.”
EVENT: “all losses” in a 72-hour period.
The circularity of these definitions complicates our attempt to construe the endorsement. Here, at least, the parties agree that all of the losses at issue arose from just one “event.” But because the policy defines the term “occurrence” to include “any one loss” from that event, we could read the phrase “in any one ‘occurrence’ ” to limit RSUI’s liability to the least of the three alternatives based on the amount of each individual loss, consistent with RSUI’s item-by-item approach. (Emphasis added.) But the endorsement defines “occurrence” to mean “any one loss ... or senes of losses,” which would include all of the losses arising from the one event (which is defined for these purposes as “all losses” in a 72-hour period). (Emphasis added.) Under this definition, we could read the phrase “in any one ‘occurrence’ ” to mean that RSUI’s liability is limited to “the least” of the three alternatives “in” all losses — that is, the entire “series of losses” — consistent with Lynd’s aggregate approach. (Emphasis added.) We thus conclude that the phrase “in any one ‘occurrence,’ ” standing alone, is consistent with both parties’ proposed constructions.
Reading the two key phrases of Paragraph l’s introductory sentence together does not change that result. Because the policy defines both “loss” and “occurrence” to mean either “a loss” or “a series of losses,” we could construe the sentence to provide for either an item-by-item or an aggregate limit. The introductory sentence could mean that, “in the event of [a loss],” RSUI’s liability is limited to one of the three alternatives “in any one [single] loss,” as RSUI proposes. But it could also mean that, “in the event of [a series of losses ],” RSUI’s liability is limited to one of the three alternatives “in any one [series of losses ],” as Lynd proposes. Here, the one event in fact caused not just “a loss” but “a series of losses,” and we agree with Lynd that, “in the event of [a series of losses],” the introductory sentence can be construed to apply the alternative limits on an aggregate basis.
4. Part la’s Limit
The first alternative limit, Part la of the endorsement, limits RSUI’s liability to “[t]he actual adjusted amount of the loss, less applicable deductibles and primary and underlying excess limits.” Once again, this language refers to “the loss” in the singular, but because the policy defines that term to mean “a loss or series of losses,” it could be read to refer either to each loss or to the entire series of losses at issue. (Emphasis added.) In this case, for example, it could refer to the amount of $11 million, which was the actual adjusted amount to repair or replace the buildings at Le Chateau, or it could refer to the amount of $24.5 million, which is the actual adjusted amount of the entire “series of losses” at issue. Lynd challenges the first construction by pointing out that the clause does not expressly refer to “each loss” or to “individual losses,” and RSUI challenges the second construction by pointing out that it does not expressly refer to “the total ” or “aggregate ” of the series of losses. (Emphasis added.) But the definition of “loss” essentially provides those terms as two different alternatives. We agree with both RSUI and Lynd that the use of these or similar phrases could have provided some clarity as to the endorsement’s meaning. But as written, *124Part la of the endorsement refers only to “the loss,” and the definition of that term is consistent with both parties’ constructions.
5. Part lb’s Limit
The second alternative limit, Part lb of the endorsement, limits RSUI’s liability to “115% of the individually stated value for each scheduled item of property insured at the location which had the loss as shown on the latest Statement of Values ..., less applicable deductibles and primary and underlying excess limits.” It is important to note that this alternative limit is based not on the amount of the “loss,” but on the amount of the “stated value” at the location that “had the loss.” RSUI contends that the reference to the “individually stated value for each scheduled item ” insured “at the location” that had “the loss” refers to the specific amount that Lynd reported for each of the three coverage items (building replacement, contents, and rental income) at each of the property locations. (Emphasis added.) For the Le Chateau property, for example, Lynd listed “individually stated values” of about $7.6 million for building replacement coverage, $10,000 for contents coverage, and $1.18 million for rental income coverage. According to RSUI, each of these amounts is an “individually stated value” for a unique “scheduled item” where loss occurred, and is thus a separate limit under Part lb. And, according to RSUI, because the endorsement requires a determination of whether that amount or the amount under Part la is “the least,” Part la must also require that the actual adjusted amount of the “loss” or “series of losses” for each item be determined on an item-by-item basis. We agree that this is a reasonable construction of this language, and that it is consistent with RSUI’s construction of the endorsement.
By contrast, Lynd contends that the phrase “individually stated value for each scheduled item ... at the location which had the loss” refers to the total of all the stated values for all the items for which Lynd sustained a loss. We find several problems with this construction. First, it ignores the word “individually,” which means “of, relating to, or associated with an individual” or “a single member of a category.” Merriam-Webster Dictionary and Thesaurus 431 (2014) (defining “individual”). Lynd suggests that we should read “individually” to mean “collectively,” but those words simply are not synonymous. Id. at 51 (defining “collective” to mean “of, relating to, or denoting a group of individuals considered as a whole”). Second, Lynd’s construction requires us to read the phrase “each scheduled item” to mean “all scheduled items,” yet “each” and “all” have essentially opposite common meanings. Id. at 259 (defining “each” to mean “every individual one” or “being one of the class named”); id. at 23 (defining “all” to mean “the whole number, quantity, or amount” or “the whole of’). We cannot construe these undefined words to mean things they do not ordinarily mean. See, e.g., Anglo-Dutch Petroleum, 352 S.W.3d at 451; Forbau 876 S.W.2d at 133.3 And finally, Lynd’s approach requires us to add the words “the total of every” before the phrase “individually stated value,” and we cannot add words to the policy’s language. Am. Mfrs. Mut. Ins., 124 S.W.3d at 162.
*125Third, while the phrase “at the location which had the loss” could support an aggregate approach to the endorsement’s limits, it cannot support the aggregate approach that Lynd proposes. (Emphasis added.) Specifically, because the phrase “at the location which had the loss” modifies the phrase “individually stated value for each scheduled item,” we could construe the sentence to require aggregating the individually stated values of each scheduled item at each of the fifteen damaged locations and then applying the limit on a location-by-location basis. But this would not support the occurrence-by-occurrence aggregation that Lynd proposes.
We thus conclude that, standing alone, the phrase “individually stated value for each scheduled item of property insured at the location which had the loss” supports RSUI’s item-by-item construction of the endorsement, and does not support Lynd’s aggregate approach.
But our analysis of Part lb cannot end here, because the language of Part lb does not end here. The dissenting opinion cleverly misstates our analysis when it asserts that we find that Part lb, “the key passage, ‘does not,’ in the Court’s own words, ‘support Lynd’s [construction]’.” Post at 141. Simply put, we are not finding or “[conceding that, standing alone, element (b) does not support Lynd’s position ...,” post at 142; rather, we find that, standing alone, one phrase contained within part lb does not support Lynd’s position. All artifice aside, our conclusion is clear: “standing alone, the phrase ‘individually stated value for each scheduled item of property insured at the location which had the loss’ ... does not support Lynd’s aggregate approach.” But as we explain in the following section, that phrase does not stand alone in Part lb, and we must construe it within its context, “giving effect to every word, clause, and sentence.” In re Office of Atty. Gen., 422 S.W.3d 623, 629 (Tex.2013). We reach our conclusion only after considering all of Part lb, in context with all of the policy’s provisions, and the conclusion that the dissent purports to find “baffling” is simply not our conclusion at all.
6. Subtracting the Deductible and Primary Policy Limit
There'is more to both Part la and Part ,1b than the phrases we have discussed thus far. In particular, to determine the amounts of the alternative limits that each part describes, each part requires subtraction of the amount of “applicable deductibles and primary and underlying excess limits.” Both parties agree that Lynd had a deductible of $25,000 per occurrence and an underlying primary limit of $20 million per occurrence.4 To calculate the amount of the limit described in Part la, we must determine the “actual adjusted amount of the loss,” and then subtract the deductible and primary limit from that amount. And to calculate the amount of the limit described in Part lb, we must subtract the deductible and primary limit from the “individually stated value for each scheduled item of property insured at the location which had the loss.”
The requirement of subtracting the deductible and primary limit from each of the amounts described in Parts la and lb makes RSUI’s item-by-item construction problematic, at best. As noted, the deductible and primary limit apply “per occurrence,” and the policy defines “occurrence” to mean a “loss ... or series of losses.” (Emphasis added.) RSUI construes the *126word “occurrence,” as it is used in Paragraph l’s introductory sentence, to mean “each loss,” rather than the entire “series of losses.” Yet even RSUI does not contend that the “per occurrence” deductible and “per occurrence” primary limit apply to “each loss.” Indeed, if RSUI must subtract $20,025,000 from the actual adjusted amount of “each loss” for each covered item (under Part la) or from the “individually stated value of each item” (under Part lb), there would be no coverage at all.
One way to avoid this outcome while still accepting RSUI’s item-by-item construction of the endorsement would be to interpret the word “occurrence” to mean “each loss” when it appears in Paragraph l’s introductory statement, but to mean “the series of losses” when it refers to the deductible and primary limit. Generally, the law recognizes “a natural presumption that identical words used in different parts of the same act are intended to have the same meaning.” Atl. Cleaners & Dyers v. United States, 286 U.S. 427, 433, 52 S.Ct. 607, 76 L.Ed. 1204 (1932). But this presumption “is not rigid,” and it “readily yields” when the text uses the words in such various connections that it “warrant[s] the conclusion that they were employed in different parts of the act with different intent.” Id. So although we find RSUI’s attempt to attribute two different definitions to the word “occurrence” to be at best a stretch, we do not completely dismiss the possibility that we could construe the language in that manner.
A second way to avoid this outcome while still accepting RSUI’s item-by-item construction of the endorsement is to construe Parts la and lb to require subtraction of a pro rata portion of the deductible and primary limit from the amounts of “each loss” and the “individually stated value of each scheduled item.” But this approach would require us to read words into the endorsement’s language, to require subtraction of “[a pro rata portion of] applicable deductibles and primary ... limits.” As we have explained, we cannot add words to the policy’s language unless the context requires us to do so. Am. Mfrs. Mut. Ins., 124 S.W.3d at 162.
A third way to avoid this outcome while still accepting RSUI’s item-by-item construction of the endorsement is to construe Parts la and lb together to require subtraction of the deductible and primary limit only once, from the aggregate amount of “the least” of each alternative as to each individually covered item. An obvious problem with this approach is that it requires some aggregation — specifically, aggregation of all of the amounts resulting from the application of the alternative limits as to each separate covered item— before subtracting the amounts of the deductible and primary limits. To do this, we would have to construe the same words to permit an item-by-item approach for part of the calculation and require aggregation for the remainder. So again, we would have to ignore the “natural presumption” that the words were intended to have the same meaning. More importantly, this approach ignores the language and structure of the parts themselves. Part la plainly requires that we subtract the deductible and primary limit from the “actual adjusted amount of the loss.” Separately, Part lb requires that we subtract them from the “individually stated value of each scheduled item.” And the introductory language requires that we apply “the least” of the two differences, after subtracting the deductible and primary coverage from each. This approach, in short, would require us to ignore the plain language and the structure of the two parts.
The dissent nevertheless embraces this approach, finding our conclusion to be “twisted,” “strange,” “unrealis*127tic,” “nonsensical,” “untroubled by realities and consequences,” “glaring[ly] peculiar[ ],” and “simply wrong.” Beneath all this bluster, the dissenting opinion is wholly lacking in brawn. Avoiding any threat of ever being accused of “textual fiyspeck-ing” or of “parsing] every jot and tittle of policy text,” the dissent makes no effort to analyze the policy’s words or structure at all. “Our primary objective ⅛ [construing an insurance policy] is to ascertain and give effect to the parties’ intent as expressed by the words they chose to effectuate their agreement.” In re Deepwater Horizon, — S.W.3d -, -, 2015 WL 674744 (Tex. 2015). “[W]e look at the language of the policy because we presume parties intend what the words of their contract say.” Gilbert Tex. Constr., 327 S.W.3d at 126. Eluding these fundamental principles, the dissent accepts RSUI’s argument, which “is simply that deductibles and other coverage come off the bottom line, as one would expect.” Post at 141. The dissent “take[s] that as a given,” post at 140 n.1, concluding:
The liability limitation quite naturally reads:
liability ... shall be limited to the least of ...
[t]he actual adjusted amount of the loss ... [or]
115% of the individually stated value for each scheduled item of property
[[Image here]]
less applicable deductibles and primary and underlying excess limits.
Post at 141. While this construction may seem “quite natural[ ],” and even a “given” in light of the dissent’s view of what the policy ought to mean, it is simply not what the policy says.
Nevertheless, other language in the policy may provide some support for the third way to avoid this outcome while still accepting RSUI’s item-by-item construction of the endorsement. Specifically, the policy’s Excess Physical Damage Coverage Form independently provides that RSUI’s liability “attaches ... only after the primary and underlying excess insurer(s) have paid or have admitted liability for the full amount of their respective ... liability.” Thus, the policy requires subtraction of the primary limit even without the subtraction language in Parts la and lb of the Scheduled Limit of Liability endorsement. RSUI may have added that language in Parts la and lb just to confirm that the policy only covers amounts that exceed those that the primary insurance has paid, and that nothing in the Scheduled Limit of Liability endorsement changes that result.5 But even if that’s the case, it added the language separately, into each of the two parts, rather than provide for a single *128deduction from the aggregate of all amounts derived from the least of the two parts. By separately requiring the subtraction of the deductibles and primary limit in both parts, the language of the endorsement makes RSUI’s proposed item-by-item construction difficult at best.
By contrast, Lynd’s approach more closely follows both the language and the structure of the parts and honors the “natural presumption” that the word “occurrence” has the same meaning throughout the endorsement. Under Lynd’s approach, the endorsement requires RSUI to (1) determine the actual amount of the “series of losses” and subtract the deductible and primary limit from that amount, (2) determine “[the total of every] individually stated value of each scheduled item” and subtract the deductible and primary limit from that amount, and then (3) compare the two differences to determine which is “the least.” Since, as we have found, the word “loss” in Part la can be construed to refer to the “series of losses,” the first step of this process is not problematic. The second step is problematic, however, because, as explained above, the language of Part lb, standing alone, cannot be construed to refer to “[the total of every ] individually stated value” without writing language into the policy.
But that language in Part lb does not stand alone. We must construe it within its context, which includes the requirement that RSUI subtract the deductible and primary limit from “the individually stated value of each scheduled item.” We agree with Lynd’s contention that this is one appropriate way to construe Part lb, and that doing so supports Lynd’s aggregate approach for applying the alternative limits.
7. The Part lc Limit
Finally, Part lc of the endorsement limits RSUI’s liability to “[t]he Limit of Liability as shown on the Declarations page of this policy or as endorsed to this policy.” The parties agree that the “Limit of Liability” refers to the policy’s overall or upper limit of $480 million.6 Because the policy separately imposes this $480 million per occurrence limit, outside of the Scheduled Limit of Liability endorsement, we ultimately conclude that Part lc is superfluous and neither supportive of nor inconsistent with each party’s proposed construction.
Lynd contends that, like the primary policy’s limit of $20 million per occurrence, Part lc’s $480 million overall limit necessarily applies to the aggregate of all losses arising out of any one occurrence, and not to each individual loss. Therefore, Lynd contends, we must also construe Parts la and lb to produce aggregate limits because, otherwise, we could not compare the amounts that each part produces to determine which is “the least.” We do not agree, however, that Part lc’s $480 million limit necessarily applies to the aggregate of all losses arising from an occurrence. It is at least possible to construe Part lc to impose a $480 million limit on the.coverage provided for “each loss” to “each scheduled item” resulting from an occurrence. Under the facts of this case, of course, it is highly unlikely that a $480 million limit on “each loss” would ever be necessary, because the highest value Lynd stated for *129any one item was $18.2 million. Even if Lynd understated the true value of that item, it is highly improbable that damages to that item (or to any of the other, lesser-valued items) could ever be greater than $480 million.
In theory, however, it was at least possible that Lynd could have acquired a new property that was so valuable that it was capable of sustaining damages in excess of $480 million. In light of that possibility, Part lc could be construed to limit the coverage to $480 million for that one item, regardless of its stated value or. the amount of damages it sustained. Construed in this manner, the Part lc limit of $480 million would not be an “overall” limit on the amount of coverage per occurrence, but would instead be a maximum limit on the amount of coverage for “each loss” incurred for “the individually stated value for each scheduled item.” And it would not increase the overall limit to allow recovery of up to $480 million for every loss sustained in one occurrence because, as we’ve already noted, the policy separately imposes a $480 million “per occurrence” limit, separate and apart from the Scheduled Limit of Liability endorsement.
But even if the Part lc limit does not necessarily apply to the aggregate of all losses from one concurrence, we do agree with Lynd that it can reasonably be construed in that manner. We find support for this construction in the fact that Part lc does not say that its limit is “the amount, of $480 million,” or even “the amount of the Limit of Liability” under the policy. Instead, it expresses the Part lc limit as “[t]he Limit of Liability as shown on the Declarations page of this policy or as endorsed to this policy.” By incorporating “the Limit of Liability,” rather than the amount of that Limit, Part lc refers not just to an amount, but to the specific “per occurrence” amount that limits RSUI’s overall liability under the policy. Thus, even RSUI agrees that, “if a single occurrence such as a hurricane damages more than one portfolio property, RSUI agreed [in Part lc] to pay the sum of the individually scheduled and sub-limited losses up to the per-occurrence cap.” (Emphasis added.) RSUI agrees, in other words, that Part lc’s limit applies once to the' aggregate total of all losses resulting from a single occurrence, and not multiple times to each individual loss.
Ultimately, however, Part lc is superfluous under either party’s construction. Whether construed as a limit on the losses to each scheduled item from one occurrence or a limit on the total losses to all items collectively from one occurrence, Part lc is superfluous because the overall Limit of Liability already limits the policy’s coverage to $480 million per occurrence, regardless of whether the occurrence damages only one item or more than one. But the important thing about Part lc is not its content but its contribution to the structure of the endorsement. Like the subtraction requirements in Parts la, and lb, RSUI may have added Part lc to clarify that the endorsement’s limits do not replace or negate the policy’s overall per occurrence limit.7 But whatever the rea*130son, RSUI added it as one of three alternatives that must be compared to each other to determine which is “the least.” As a result, we must construe Part lc in a manner that is consistent with, and comparable to, Parts la and lb. Based on the actual language and structure of the endorsement, if we construe those two parts to provide for an item-by-item comparison, as RSUI urges, we must construe Part lc to impose an item-by-item limit, even though the Limit of Liability would have the same effect. If we construe Parts la and lb to provide for a comparison of the aggregates of all losses against all stated values, we must construe Part lc to impose an aggregate cap. This is because each part must result in an amount that can be compared to the amounts of the other parts, so that “the least” can be determined. As we have said, we could construe Part lc to impose either an item-by-item or an aggregate limit. Ultimately, because of its reference to the “Limit of Liability,” which both parties agree applies in the aggregate, we think it more readily supports Lynd’s aggregate approach. Even if that were not the most likely construction, it is at least an appropriate construction, and thus Part lc can be construed in a manner that supports Lynd’s constructions of Parts la and lb.
8. The Policy as a Whole
Although we have discussed various provisions that inform the meaning of the Scheduled Limit of Liability endorsement’s language, we are not conducting a balancing test. The law does not ask which party can identify the greatest number of ways the contract’s structure or language supports its construction. Nor does the law deem a contractual provision ambiguous merely because both parties can point to words or phrases that, read in isolation, favor different constructions of the contract, or because both parties can identify language that, through the lens of hindsight, could have been more clearly stated. See, e.g., Balandran, 972 S.W.2d at 741; Daniel, 243 S.W.2d at 157. Few contracts could withstand that scrutiny. “An ambiguity does not arise simply because the parties offer conflicting interpretations.” Schaefer, 124 S.W.3d at 157. To be ambiguous, both interpretations must be a reasonable interpretation of the words chosen by the parties when read in the context of the policy as a whole. See Gilbert Tex. Constr., 327 S.W.3d at 126; Forbau, 876 S.W.2d at 133.
Thus, we must consider all of the endorsement’s relevant terms, and construe them together. Doing so here, we conclude that the language supports both parties’ constructions. Although some aspects of the endorsement are more consistent with Lynd’s interpretation than RSUI’s, nothing in the endorsement, when read as a whole within the context of all the policy’s language, is irreconcilable with RSUI’s interpretation. And although some aspects of the endorsement are more consistent with RSUI’s interpretation than Lynd’s, nothing in the endorsement, when read as a whole within the context of all the policy’s -language, is irreconcilable with Lynd’s interpretation. We thus conclude that the policy’s lan*131guage could reasonably be construed to support them both.
B. Is Lynd’s Proposed Construction Reasonable?
RSUI contends that Lynd’s proposed construction of the Scheduled Limit of Liability endorsement is unreasonable, and therefore does not establish ambiguity, because it ignores the common knowledge and practices of the insurance industry. Specifically, RSUI argues that the insurance industry recognizes two different types of commercial property policies: those that provide “scheduled” (or “specific”) coverage, and those that provide “blanket” coverage. See Br. for Pet’r at 13, No. 13-0080, — S.W.3d -, 2013 WL 5785791 at *13 (citing David Russell, Insuring the Bottom Line: How to Protect Your Company from Liabilities, Catastrophes and Other Business Risks 50 (Silver-lake 1st ed. 2004) for the proposition that commercial property policies insure property on either a specific, schedule, or blanket basis). According to RSUI, the “distinction between the two types of coverage is simple: blanket coverage aggregates all covered properties to provide one coverage limit applied collectively, while scheduled coverage provides scheduled limits for each property independently.” See id. (citing Steven Plitt, et al., 12 Couch on Insurance §§ 175:90, 177:72 (3d ed.1997)).
RSUI explains that a scheduled policy “is a popular tool for insuring multiple properties, often located in multiple states, because it is cost efficient.” This is because “the premium is based on the value of each item of property as declared (or ‘scheduled’) by the property owner,” and the policy covers the actual loss for each item “only up to the scheduled amount.” When a single occurrence damages multiple properties covered by a scheduled policy, “the insurer’s total liability for these individually determined per-property losses is capped at an agreed-upon level.” This item-by-item limitation of the insurer’s liability, RSUI explains, “is the ‘hallmark’ of a scheduled policy.” Using RSUI’s vernacular, a “scheduled policy” is one that provides “scheduled coverage,” meaning coverage that is limited on an item-by-item basis. And because the policy at issue here is a scheduled policy, RSUI contends, its liability limits apply on an item-by-item basis.
The court of appeals concluded that it “need not decide whether the RSUI policy provides for scheduled coverage or blanket coverage,” but must instead decide what the policy’s language means “without regard to how the policy or the coverage it provides is labeled.” RSUI contends that, by “avoiding the question” of whether this is a scheduled or blanket policy, the court of appeals “judicially create[d] an entirely new type of policy that is unknown in the insurance industry, producing absurd results that put Texas at odds with other jurisdictions that have interpreted comparable policies.” We agree with RSUI that we cannot “avoid the question” of whether this is a scheduled or blanket policy, if by those labels RSUI means we must decide whether the policy limits RSUI’s liability on an item-by-item or an aggregate basis. But we agree with the court of appeals that we cannot simply label the policy as “scheduled” or “blanket” and then construe its language to reach that result.
In construing the policy’s language, RSUI contends that we cannot simply look at the words themselves, but must also consider the policy’s overall “structure,” which confirms that it is a scheduled policy. Specifically, RSUI points out that the policy (1) covered multiple properties at multiple locations, (2) required Lynd to submit a schedule of the covered properties and their values, and (3) provided that *132RSUI would rely on the reported values to assess its risks and determine the amount of the premium.8 In addition, RSUI asserts that we must characterize the policy as a scheduled policy because, otherwise, the policy would have the absurd effect of encouraging Lynd to understate the properties’ values. We conclude that none of these reasons individually, nor all of them together, is sufficient to establish that the policy must be a scheduled policy with item-by-item limits, or to render Lynd’s proposed construction of the policy unreasonable.
1. Multiple Properties at Multiple Locations
RSUI explains that a scheduled policy “is a popular tool for insuring multiple properties, often located in multiple states, because it is cost efficient.” But it' also acknowledges that “[e]ither scheduled or blanket policies can be used to cover multiple properties at multiple locations — the kind of coverage that Lynd needs for its multistate portfolio.” In fact, the parties agree that Westchester’s primary policy, which covered the same properties, provided blanket coverage with limits applying to all aggregated losses arising from any one occurrence. The fact that the policy covered multiple properties at multiple locations is thus consistent with either a scheduled or a blanket policy, and does not indicate that it is one or the other.
2. Required List of Properties and Values
According to RSUI, “[ujnder a scheduled policy, before coverage is placed, the property owner submits to the insurer a ‘Statement of Values”listing the value of each item of property to be insured.” Because this policy required Lynd to provide a Statement of Values before coverage was placed, RSUI asserts, the policy provides scheduled coverage. RSUI cites to language in several cases to support this proposition. See, e.g., Reliance Nat’l Indem. Co. v. Lexington Ins. Co., No. 01 C 3369, 2002 WL 31409576, at *8 (N.D.Ill. Oct. 23, 2002) (“Because the Lexington Policy separately scheduled different items of property, it is a scheduled policy with specific limits for particular items and not a blanket coverage policy.”); Anderson Mattress Co. v. First State Ins. Co., 617 N.E.2d 932, 935 (Ind.Ct.App.1993) (finding policies were scheduled policies “because they separately scheduled values for individual pieces of property”); Vernon Fire & Cas. Ins. Co. v. Sharp, 264 Ind. 599,349 N.E.2d 173, 177 (1976) (holding policies were scheduled policies because “the property insured was separately scheduled and valued in the contracts”).9
Several' other courts, however, have recognized that insurance policies often “schedule” the covered properties and their values for reasons other than to impose individual limits. For example, “it is not uncommon to find a blanket policy with a separate, specific listing of individual pieces of property with values, the values of which are averaged to calculate the insurance premium.” Aerosonic Corp. v. *133Liberty Mut. Fire Ins. Co., No. 8:09-CV-1355-T-26EAJ, 2009 WL 5068616, at *8 (M.D.Fla. Dec. 17, 2009). When schedules are required and used to calculate an average or overall premium, or to ensure compliance with a 'co-insurance requirement, or for some purpose other than to impose separate limits on each scheduled item, the use of a schedule “was not at all inconsistent” with blanket coverage. Reliance Ins., 322 F.2d at 807. A blanket policy may use a schedule for one of these purposes, and “not to establish a scheduled policy.” Core-Mark Int’l Corp. v. Commonwealth Ins. Co., No. 05CIV0183WHP, 2006 WL 2501884, at *6 (S.D.N.Y. Aug. 30, 2006). So “the bare fact that a schedule of values is used ... is not enough to establish that a policy is scheduled and not blanket as a matter of law.” ARM Props. Mgmt. Grp. v. RSUI Indem. Co., No. A-07-CA-718-SS, 2008 WL 5973220, at *9 (W.D.Tex. Aug. 25, 2008) (citing Reliance Ins. Co., 322 F.2d at 807).
We agree with these courts. It is one thing to say that every scheduled policy (that is, every policy that limits liability on an item-by-item basis) includes or incorporates a schedule of covered items and their values, but it is quite another to say that every policy that includes a schedule of covered items and their values limits liability on an item-by-item basis. The Scheduled Limit of Liability endorsement at issue here expressly provided that “[t]he premium for this policy is based upon the Statement of Values.” We thus conclude that the “bare fact” that this policy required Lynd to file a Statement-of Values does not indicate that it limited liability on an item-by-item basis.
3. Reliance on Values to Determine Risk and Premium
RSUI acknowledges that an insurer may use a schedule for other purposes, and it admits that it used the Statement of Values in this case to evaluate its risks and determine the amount of the premium. But RSUI contends that even those purposes demonstrate that this was a scheduled policy that limited liability on an item-by-item basis. As we understand this argument, RSUI contends it evaluated its risks and determined the premium based on the stated values for each covered item, and this demonstrates that the policy was “structured” as a scheduled policy with item-by-item limits.
There is nothing in this policy, however, or even in this record, that explains how RSUI evaluated its risks and determined the premium, much less evidence that it did so on an “item-by-item” basis. To the contrary, the policy reflects that RSUI provided the coverage at an average or flat rate of $0,025 per $100 of reported value, and that this rate applied to all items at all properties at all locations in the Statement of Values. For purposes of setting the premium, in other words, RSUI did not distinguish between properties and assign different premiums to different properties based on their risk of sustaining a loss. Instead, RSUI essentially averaged all of the covered risks and based the premium on a flat rate that applied equally to the value of every scheduled item at every location. For a scheduled policy, the premium is usually determined based on the individual risks assigned to each individual covered property, while “[a]n average rate usually indicates a blanket policy.” Reliance Ins. Co., 322 F.2d at 806. The fact that RSUI based this policy’s premium on an average rate for all scheduled items is thus more consistent with a blanket policy than a scheduled policy.
RSUI asserts that it would have charged a much higher rate — much more than $0,025 per $100 of value — if the policy’s limits applied on an aggregate rather than item-by-item basis. In fact, RSUI and the *134amici that support it10 warn that, if they are unable to limit their liabilities on an item-by-item basis, many if not most insurers will simply stop writing policies to cover multiple properties at multiple locations. As a result, they contend, companies like Lynd, which need insurance for multiple properties, would lose the ability to obtain the insurance, at least at the significant discount that a scheduled policy offers.
Since, as this case demonstrates, applying the alternative limits on an aggregate basis may result in the insurer being liable for more in any one occurrence than if the limits applied on an item-by-item basis, it certainly makes sense that RSUI would charge a higher average rate for a policy that applies the limits on an aggregate basis. But there is nothing in this policy or this record to establish what the “appropriate” rate would be, much less that the $0.025/$100 rate is consistent only with item-by-item limits. And even if there were, we could not conclude that RSUI’s decision to set the rate based on item-by-item limits necessarily establishes that the language of the policy actually imposes the limits on that basis.
Nor do we find RSUI’s (and its amici’s) warnings compelling. By finding the policy at issue in this case to be ambiguous, and thus construing it in favor of the insured, we are not holding that insurers cannot offer scheduled policies that limit their liabilities on an item-by-item basis. We are simply holding that when they offer such policies they must use language that unambiguously confirms that they are doing so. They can avoid all of the calamities that RSUI and its amici contend today’s decision will cause to the commercial property insurance market by writing their policies more clearly. We have no fear for the insurance market, because we have no doubt that RSUI and its competitors can do so.
In any event, because RSUI based the premium for Lynd’s policy on an average rate applied to all scheduled items regardless of their relative risks, and in the absence of any evidence establishing that the rate it charged was consistent only with limits that apply on an item-by-item basis, we conclude that the fact that RSUI based the premium on Lynd’s reported values does not establish that the policy was a scheduled policy with item-by-item limits.
4. Encouraging Understatement of Values
Finally, in discussing the policy’s “structure,” RSUI contends that unless we construe this policy to be a scheduled policy with item-by-item limits, we will cause the absurd result of encouraging insureds to understate the values of their covered items. As RSUI explains it, “[a] scheduled policy protects the insurer by limiting its liability to the amount of the insured’s loss, but only up to the scheduled amount, upon which the premium was based.” At the same time, RSUI contends:
The structure of a scheduled policy protects the insured by providing full recovery whether the loss is partial or total, but only if the insured gave an accurate value when scheduling the property (because the scheduled value is intended to be the cost to replace the property in *135the event of a total loss). However, if the insured undervalued the property either by mistake or to gain the benefit of a correspondingly lower premium, then — by its own doing — it will not be fully compensated in the event of a total loss.
RSUI’s construction of its policy thus affords RSUI complete protection against Lynd underinsuring its properties by more than 15%. Put another way, under RSUI’s construction, if Lynd understates the value of the insured properties and pays premiums based on the lower values, Lynd bears the risk (above the 15% cushion) associated with its decision to underin-sure.11 RSUI contends that Lynd’s construction, by comparison, affords RSUI protection against undervaluation by Lynd only in limited circumstances that are not present here. RSUI and its amici argue that Lynd’s construction encourages property owners to underinsure because they will frequently be able to recover more insurance than they purchased. Here, for example, Lynd seeks to recover about $11 million to replace the buildings at the Le Chateau, which is nearly $3.5 million more than the building replacement value of $7.6 million that Lynd reported on the Statement of Values, and on which the premiums were based.
Lynd responds that RSUI’s concern is “unfounded” because under Lynd’s construction, “[a]n insured would be harmed by undervaluing its properties in the vast majority of claims, which involve only isolated events at a single location.” Thus, Lynd asserts, “[n]o reasonable insured would render its own excess policy largely illusory in order to save a little money in the unlikely event of a major catastrophe.” 12 In addition, Lynd argues that “the only way an insured could come out ahead” by underreporting values “is if it knew ahead of time which properties would be completely destroyed and which would receive only mild damage, and then report the values accordingly — all to save a few thousand dollars in premiums.”
Regardless of whether we agree with Lynd’s response, the answer to RSUI’s concerns is that the benefits or detriments of either approach (scheduled, blanket, or an “unknown hybrid”) ultimately depend on how RSUI underwrites the risk and calculates the premiums. Certainly, if RSUI set the premium for Le Chateau based on the scheduled value of $7.6 million and on the assumption ■ that RSUI would never have to pay more than 115% of that amount, then the amount of the premium would be insufficient if RSUI were required to pay more than the 115%. But this analysis overlooks the reality that the policy covered, and that Hurricane Rita damaged, more than just Le Chateau. Lynd does not contend that RSUI would have to pay more than 115% of the stated *136value of Le Chateau if Le Chateau were the only property damaged in any one occurrence. If an occurrence only damaged Le Chateau, and the amount of the loss exceeded 115% of the stated value of the,scheduled items at that location, Part lb would limit RSUI’s liability to 115% of the stated value.
Similarly, under Lynd’s proposed construction, if an occurrence damages property at more than one location, and the amount of the loss exceeded 115% of the total of the stated values of the scheduled items at those locations, Part lb would still limit RSUI’s liability to 115% of the total of the stated values, even if one or more of the properties incurred losses less than the total of the stated values at those locations. The only difference is that RSUI could be liable for more than 115% of the stated values at one or more locations, but only if the losses at that location exceeded 115% of those stated values and the total of all losses did not exceed 115% of the total of the stated values at all damaged locations. Here, for example, the total combined value that Lynd reported for all of the covered properties was just over $793 million, and RSUI charged a flat or average rate premium of $0.025/$100 for all of that amount. As it turned out, Hurricane Rita damaged fifteen of the listed properties, for which Lynd had reported a total combined value of just over $154 million. The total amount of the adjusted loss for those fifteen properties was just under $25 million, (including the amounts that exceeded the reported values for the Le Chateau and Willow Bend complexes). Though it was merely fortuitous that two of the properties sustained losses beyond their stated values, it is just as fortuitous that the thirteen other properties did not sustain more losses. Ultimately, though, the extent of RSUI’s risk was the same: 115% of the total of the stated values for the scheduled items at the locations that incurred a loss.
If Lynd had significantly understated the values of all of the scheduled items at all fifteen properties, it would have paid lower premiums but would have faced the risk that the coverage would be insufficient, particularly if all of the properties were seriously damaged. But neither Lynd nor RSUI could know in advance which of the properties would be seriously damaged, or even damaged at all. Ultimately, then, under Lynd’s proposed construction, the insured is encouraged to accurately state the values because it cannot understate all of the values and it cannot know which properties to strategically undervalue in advance of an occurrence.
The dissent concludes that the “most serious” flaw in our analysis is the “glaring peculiarity of [our] interpretation — that the policy pays more of the losses for one property if others are damaged at the same time.” Post at 142. But the dissent makes no effort to explain why this result is so “peculiar” when RSUI receives the exact same premium rate for all of the items at all of the properties, whether those properties are ever damaged or not, and its liability is always capped at 115% of the values stated for all of those identically treated properties, regardless of whether one, some, or all of them are damaged in a single occurrence. Ultimately, the dissent concludes only that, for the dissent, it is “difficult to imagine” that the parties could have intended such an arrangement. Post at 142. Fortunately, we need not attempt to imagine what the parties intended, and instead need only construe the “parties’ intent as expressed by the words they chose to effectuate their agreement.” In re Deepwater Horizon, — S.W.3d at -.
In the same vein, we cannot agree with RSUI’s argument that Lynd’s construction *137of the endorsement permits Lynd to recover benefits for which it paid no consideration. Specifically, RSUI contends that, for Le Chateau and Willow Bend, Lynd seeks to recover amounts in excess of 115% of the stated values of the scheduled items at those locations, but it did not pay premiums for those excess amounts. But this argument ignores the fact that RSUI charged the same rate for Le Chateau and Willow Bend that it charged for the coverages at all of the other properties. And it ignores the fact that Lynd did pay the premium that RSUI set to cover up to 115% of the stated value of the scheduled items at all of the damaged locations, and all it is seeking is to recover up to that amount. Because not all of the properties sustained losses in excess of 115% of the stated values, Lynd is actually seeking far less than that amount. And in any event, if the $0,025/100 premium rate was insufficient to constitute adequate consideration for the coverage Lynd seeks, RSUI could have charged a higher rate. Or, as we have explained, RSUI could have drafted the policy to unambiguously apply the item-by-item limits it seeks to impose.
Ultimately, we agree with RSUI that the distinction between a scheduled policy and a blanket policy is critical to an insurer’s determination of the risk that the policy shifts from the insured to the insurer, and thus the premiums it charges for undertaking that risk. But we agree with Lynd that the language of this policy ultimately dictates how the policy operates, not the label the parties affix to it, and that parties can choose contract terms that' are unusual or even ill advised. We are also mindful that, while we may- decide “the meaning to which a certain term or phrase is most reasonably susceptible” in light of the meaning “regularly observed in the place, vocation, trade, dr industry so ‘as to justify an expectation that it will be observed with respect to a particular agreement,’ ” we may not consider extrinsic evidence to “contradict or vary the meaning of the explicit language of the parties’ written agreement.” Nat’l Union Fire Ins., 907 S.W.2d at 521 & n.6 (quoting Restatement (Seoond) of Contracts § 222(1)). For . these reasons, we conclude that RSUI’s arguments regarding the insurance industry and the “structure” of the Scheduled Limit of Liability endorsement are insufficient to render Lynd’s proposed construction of the endorsement’s language unreásonable.
C. Other Courts’ Decisions
The parties cite to several cases that demonstrate that RSUI’s Scheduled Limit of Liability endorsement, like many other endorsements and insurance policy parts, is similar to forms that RSUI and other insurers have used in commercial property policies across the nation for decades. See, e.g., LES Realty Trust “A”, 977 N.E.2d at 570; Orleans Parish Sch. Bd. v. Lexington Ins. Co., 99 So.3d 723 (La.Ct.App.2012); RSUI Indem. Co. v. Benderson Dev. Co., No. 2:09-cv-88-FtM-29DNF, 2011 WL 32318 (M.D.Fla. Jan. 5, 2011); Axis Specialty Ins. Corp. v. Simborg Dev., Inc., No. 07-C-5906, 2009 WL 765298 (N.D.Ill. Mar. 20, 2009); Gulfport-Brittany, LLC v. RSUI Indem. Co., No. 1:07cv1036HSO-JMR, 2008 WL 4951468 (S.D.Miss. Nov. 7, 2008), aff'd, 339 Fed. Appx. 413 (5th Cir.2009) (per curiam). When we construe insurance policies, we “usually strive for uniformity in construing insurance provisions, especially where ... the contract provisions at issue are identical across the jurisdictions.” Nat’l Union Fire Ins. Co. of Pittsburgh, 907 S.W.2d at 521-22; see also Zurich Ant., 268 S.W.3d at 496-97 (“We have repeatedly stressed the importance of uniformity “when identical insurance provisions will necessarily be interpreted in various jurisdictions.’ ”) *138(quoting Cowan, 945 S.W.2d at 824). We do this to afford both insureds and insurers the heightened predictability of a body of interpretive case law.
RSUI and its amici argue that adopting Lynd’s construction of this endorsement will place Texas out-of-line with all but one of the other jurisdictions that have addressed this issue and will preclude insurers from offering this sort of insurance for reasonable rates in Texas.13 In response, Lynd asserts that its construction does not conflict with other jurisdictions’ interpretations of the same or similar endorsements because the cases RSUI cites involve different language or damage at a single location, not at multiple locations like occurred here. Lynd contends that only one court has construed this same endorsement in a situation involving damage to multiple locations caused by a single occurrence, and that court rejected RSUI’s interpretation. See ARM Props. Mgmt., 2008 WL 5973220, at *7-9. Lynd also cites to the decisions of other jurisdictions that have found that similar policy language created an aggregate limit, or was at least ambiguous as to the limits it imposed. See, e.g., Berkshire Refrigerated Warehousing LLC v. Commercial Underwriters Ins. Co., No. 05 C 1953, 2006 WL 862877, at *2 (N.D.Ill. Mar. 27, 2006); Core-Mark Int’l, 2006 WL 2501884, at *6; Reliance Ins., 322 F.2d at 807.
Of course, Lynd attempts to distinguish RSUI’s cases in a variety of ways, just as RSUI attempts to distinguish Lynd’s. We agree that many of them, on both sides, are distinguishable for a number of reasons. In the end, however, we conclude that the national jurisprudence on this issue is insufficient to require us to ignore the ambiguity of this policy, for at least three reasons.
First, courts follow other jurisdictions’ constructions of identical policy language to create uniformity among the jurisdictions, so that the parties can predict and rely on the language even if they are in different jurisdictions. This is particularly important when, as here, the exact same policy covers one insured’s multiple properties located in numerous jurisdictions, and is similarly important when the same policy form, using identical language, covers different insureds in different locations. But as the cases the parties cite demonstrate, the various iterations of the allegedly “scheduled policies” that RSUI and other insurers have used throughout the years have not always used identical language, and there is no judicial uniformity even among those that do. In Core-Mark International, for example, the court addressed a policy that contained language similar to the language at issue here, but with a key difference: Part lb referenced the “total stated value for the property involved, as shown on the latest statement of values” rather than the “individually stated value of each scheduled item of property insured as shown on ... the Statement of Values.” 2006 WL 2501884, at *2, *7 (emphasis added). The court held that the language of that policy was ambiguous as to whether it provided scheduled or blanket coverage. Id. at *6. In doing so, it expressly distinguished other cases construing the “individually stated value of each scheduled item” language present in this case, noting that those poli*139cies “unambiguously established scheduled coverage.” Id. at *7. Yet, prior to that court’s decision, another court had construed a nearly identical “total stated value” provision as unambiguously creating item-by-item limits. Anderson Mattress, 617 N.E.2d at 935-37 (construing similar “total stated value” provision as creating item-specific rather than blanket coverage).
Second, as Lynd points out, the commercial property policy at issue in this case is not a standard form policy, and the Scheduled Limit of Liability endorsement in particular is subject to revisions that RSUI may choose to make or accept in negotiations with each individual insured.14 In cases “involving a standard form policy mandated by a state regulatory agency, ... the actual intent of the parties is not what counts (as they did not write it), but the ordinary, everyday meaning of the words to the general public.” Fiess v. State Farm Lloyds, 202 S.W.3d 744, 746 (Tex.2006). Conversely, in cases involving a non-standard form, like the one at issue here, our task is to determine and enforce the mutual intent of the parties to the contract. In doing so, the decisions of other courts in cases involving other parties and other circumstances are not as compelling, particularly when those decisions are themselves less than uniform.
Finally, the numerous and varied decisions that the parties cite, which date back to as early as 1963 and continue until as recently as 2012, offer us no easy opportunity to establish uniformity and yet demonstrate an urgent need for clarity in this arena. The endorsement at issue here is not ambiguous simply because it includes the phrase “in any one ‘occurrence’ ” in Paragraph l’s introductory statement or the disjunctive “or” between Parts lb and lc. Rather, as our effort to understand and explain the endorsement illustrates, the language presents many obstacles to easy interpretation, including the circuitous definitions of the words “loss,” “occurrence,” and “event,” the varying usages of and confusing relationships between those words, and the complicating incorpo-rations of the deductible, the primary policy limits, and the excess policy’s overall limit into the calculations of the endorsement’s three alternative limits. Because of these obstacles, we are convinced that the language is ambiguous, even if other jurisdictions are not.
We are also convinced that one more court decision will do little to meet the need for clarification. Instead, only RSUI, and other insurers who use the same or similar endorsements, can provide the clarity needed to resolve these issues by significantly revising the endorsement to say what the insurers really want it to mean. We do not mean to suggest that doing so will be easy, but it will be easier than trying to construe the current language or create uniformity among the current jurisprudence. For these reasons, we conclude that the numerous and divergent decisions *140of other jurisdictions attempting to construe this and similar endorsements are insufficient to ignore the ambiguities we have identified.
IV.
Conclusion
We hold that the Scheduled Limit of Liability endorsement at issue in this case is reasonably subject to both parties’ proposed constructions and that the endorsement is therefore ambiguous. Because our rules require us to construe an insurance policy’s ambiguous coverage limitation in favor of coverage for the insured, we affirm the court of appeals’ judgment adopting Lynd’s proposed construction.
Chief Justice Hecht filed a dissenting opinion, in which Justice Green and Justice Brown joined.. When construing undefined contractual terms, courts may consider the terms’ commonly accepted meanings within the relevant industry. Cf. Heritage Res., Inc. v. Nations-Bank, 939 S.W.2d 118, 122 (Tex.1996) (applying “commonly accepted meaning in the oil and gas industry” of phrase "market value at the well”). Those meanings are not necessarily determinative, however, and may not apply when the language and its context demonstrate that the parties intended a different meaning. Lynd agrees that Part lb’s limit imposed "scheduled limits” as one of the three alternatives, in the sense that the limit under that alternative was based on "the value for each scheduled item of property” listed on the Statement of Values, but Lynd contends that the language of the endorsement did not permit RSUI to impose that limit on an item-by-item basis. We discuss these arguments in more detail below.
. The Excess Physical Damage Coverage Form states that the policy is "subject to the same warranties, terms and conditions (except as regards the premium, the amount and limits of liability other than the deductible or self-insurance provision where applicable, and the renewal agreement, if any; and EXCEPT AS OTHERWISE PROVIDED HEREIN) as are contained in or as may be added to the policy(ies) of the primary Insurer(s) prior to the happening of a loss for which claim is made hereunder, and should any alteration be made in the premium for the policy(ies) of the primary Insurer(s), then the premium hereon shall be adjusted accordingly.”
. See also Axis Specialty Ins. Corp. v. Simborg Dev., Inc., No. 07-C-5906, 2009 WL 765298, at *4 (N.D.Ill. Mar. 20, 2009) (rejecting insured's argument that term "each" in similar endorsement was analogous to "total” because "each” means "every one of the two or more individuals composing the whole, considered separately from the rest”) (citing Orr v. Edgar, 283. Ill.App.3d 1088, 219 Ill.Dec. 355, 670 N.E.2d 1243, 1249 (1996)).
. Because Lynd purchased only two layers of insurance, there was no "underlying excess" policy limit.
. The policy's Declarations page did not “show” a "Limit of Liability,”, but it did incorporate an "Excess Physical Damage Schedule,” which listed the "Limit Insured” as $480 million per occurrence. The Excess Physical Damage Coverage Form and the Reporting Endorsement (both of which the Declarations page also incorporated) stated that "the limits of [RSUI’s] liability shall be” $480 million per occurrence, and RSUI "will not pay more than” that amount.
. Gulfport-Brittany, LLC v. RSUI Indemnity Co. involved an excess policy with a Scheduled Limit of Liability endorsement very similar to the one at issue here. 339 Fed.Appx. 413, 415 (5th Cir.2009) (per curiam). RSUI had issued the excess policy in that case, and the insured was an owner of two apartment complexes that were damaged by Hurricane Katrina. Id. at 414. The property owner argued that there was a conflict between the policy’s Excess Physical Damage Schedule . and Excess Physical Damage Coverage Form, which only referenced the overall limit on liability (the "Limit Insured”), and the limitations in the Scheduled Limit of Liability endorsement. Id. at 415. The property owner *130argued that the policy language was ambiguous as to whether both limits applied or whether only one or the other applied. Id. Including Part lc in the Scheduled Limit of Liability endorsement helps to convey that the limits in Parts la, lb, and lc all apply to produce the "least” amount and that the Scheduled Limit of Liability does not displace the policy’s upper limit stated elsewhere in the policy forms. See id. (holding that the policy unambiguously created "an overall $140 million per occurrence limit with scheduled sub-limits for individual properties”).
. RSUI also relies on Part lb of the Scheduled Limit of Liability in this context. We have addressed Part lb above.
. See also Gulfport-Brittany, LLC v. RSUI Indem. Co., No. 107CV1036HSO-JMR, 2008 WL 4951468, at *3 (S.D.Miss. Nov. 7, 2008) (“courts in other jurisdictions have determined that where a policy contains a scheduled limit of liability endorsement, and the description of the premises is listed as per schedule on file with the company, a scheduled-rather than blanket-policy is created, and the insurer's liability as to that particular property is limited to the value shown on the statement of values on file with the company”), aff'd, 339 Fed.Appx. 413 (5th Cir.2009) (per curiam).
. We have received amicus filings in support of RSUI from the Property and Casualty Insurers Association of America; Axis Surplus Insurance Company and Axis Insurance Company; the National Association of Professional Surplus Lines Offices, and the National Association of Mutual Insurance Companies. We have also received amicus filings in support of Lynd from San Antonio Apartment Association, Inc.; Texas Apartment Association, Inc. and Texas Hotel & Lodging Association; and a group of Texas policyholder coverage lawyers.
. We do not read RSUI’s arguments as implying that the decision to underinsure is necessarily a nefarious one, only that the decision creates exposure, which is generally born by the decision maker. For example, a person may choose to purchase only $15,000 in coverage for damage to her car even though the car is worth more than that amount. In the event that the car is totaled, she may recover $15,000 from her insurer but will have to incur any additional costs for replacing or repairing the car.
. We note that the decision to insure a piece of property for less than its full value does not render the insurance “largely illusory” — it simply allows the insured to pay lower premiums based on the lower insured value by retaining the risk of damage in excess of the. insured value. The complaint made by RSUI and its amici is that Lynd's construction of the policy allows Lynd to pay lower premiums based on the lower insured value while still shifting most of the risk of damage in excess of the insured value to the insurance company.
. RSUI cites to a number of cases, including LES Realty Trust, 977 N.E.2d at 570; Orleans Parish School Board, 99 So.3d at 726-27, 729; Benderson Development, 2011 WL 32318, at *5; Axis Specialty Insurance, 2009 WL 765298 at *3-4; Gulfport-Brittany, 339 Fed. Appx. at 415-16; Reliance National Indemnity, 2002 WL 31409576, at *8; Fair Grounds Corp. v. Travelers Idem. Co., 742 So.2d 1069, 1071-72 (La.App.1999); Anderson Mattress, 617 N.E.2d at 935-37; and Sharp, 349 N.E.2d at 176-77.
. RSUI acknowledges, in fact, that it accepted an altered version of the endorsement in a policy it issued for a different insured after it issued the Lynd policy, which omitted the phrase “in any one 'occurrence' " in the introductory statement in Paragraph 1 and omitted the disjunctive "or” between Parts lb and lc. Lynd argues in its brief that RSUI made these revisions to the endorsement to make it consistent "with the policy interpretation it argues for in this case.” RSUI denies this accusation in its reply brief, and explains that the revised endorsement was actually "inserted by the broker (not by RSUI) for that particular insured, in the negotiation of that particular manuscript policy.” Regardless of whether or why RSUI revised the endorsement for that particular insured, RSUI acknowledges that this is not a standard form policy, and its explanation confirms that it can make or accept revisions based its negotiations with each insured.