Stephens, J.
¶1 This case involves the proper interpretation of a “resulting loss” clause in an all-risk insurance policy. It also provides an opportunity to clarify application of the efficient proximate cause rule. The Court of Appeals overturned a jury verdict in favor of the insured, reasoning that the resulting loss clause did not apply in the absence of a secondary covered peril that proximately caused the loss. The court remanded for a jury determination as to the efficient proximate cause of the insured’s loss, holding that if the efficient proximate cause was not itself a covered peril, then the policy did not provide coverage. We reverse the Court of Appeals. Because the loss at issue was not excluded under the policy, coverage exists under the ensuing loss provision. And, because there is no rule of law excluding coverage under an efficient proximate cause analysis, and the insurer is precluded from changing the ground for its denial of coverage, there is no basis for a jury to determine the efficient proximate cause of the loss. Accordingly, we reinstate the judgment of the trial court.
FACTS AND PROCEDURAL HISTORY
¶2 Vision One and Vision Tacoma (collectively Vision) joined forces to develop a condominium project in downtown [506]*506Tacoma. Vision contracted with D&D Construction to pour concrete for the building. D&D in turn subcontracted with Berg Equipment & Scaffolding Company to supply the shoring that would temporarily support the poured concrete slabs.
¶3 Several weeks after the shoring was installed, D&D began pouring concrete for the first floor. Shortly after finishing the first section of the floor, the shoring underneath the concrete gave way. The framing, rebar, and newly poured concrete came crashing down onto the lower level parking area, where the wet concrete eventually hardened. It took several weeks to clean up the debris, repair the damage, and reconstruct the collapsed floor.
¶4 Vision timely submitted a claim to Philadelphia Indemnity Insurance Company. Philadelphia hired BT & Associates, an independent engineering firm, to investigate the cause of the collapse. Following its investigation, BT & Associates concluded it was “more likely than not” that the collapse resulted from “[a] marginal shoring design” and “[s]everal problems with the shoring installation.” Clerk’s Papers (CP) at 6118. Its report noted that while “[t]he marginal shoring design alone may not have caused the shoring to collapse,” the “[marginal shoring design] in combination with various shoring installation problems ... on a more likely than not basis, caused the shoring to collapse.” Id.
¶5 At the time of the collapse, Vision had a builders’ risk policy with Philadelphia.1 Under the policy, Philadelphia was required to “pay for direct physical loss’ to Covered [507]*507Property caused by or resulting from any of the Covered Causes of Loss.” CP at 5973. “Covered Property” was defined as “[Vision’s] property or the property of others for which [Vision is] liable, consisting of: . . . [m] ateríais, supplies, machinery, equipment, or fixtures which will become a permanent part of the building, structure, or project.” Id. The policy defined “Covered Causes of Loss” as “Risks of Direct Physical ‘Loss’ to Covered Property unless the ‘loss’ is excluded.” CP at 5974.
¶6 The policy specifically excluded certain losses, including those “caused by or resulting” from deficient design or faulty workmanship. CP at 5977-78. Under the policy, Philadelphia would not pay “for loss or damage caused by” any excluded event. CP at 5971, 5976. The policy explained when a loss was “caused by” an excluded event:
Loss or damage will be considered to have been caused by an excluded event if the occurrence of that event:
(1) Directly and solely results in loss or damage; or
(2) Initiates a sequence of events that results in loss or damage, regardless of the nature of any intermediate or final event in that sequence.[2]
CP at 5971. “Collapse” was not listed as an excluded event.
¶7 The exclusion for faulty workmanship contained a resulting loss clause providing that “if loss or damage by a Covered Cause of Loss results, [Philadelphia] will pay for the loss or damage caused by that Covered Cause of Loss.” CP at 5972, 5978. This provision was likewise included in Vision’s policy as an endorsement. CP at 5972.
¶8 Following BT & Associates’ investigation into the cause of the collapse, Philadelphia denied Vision’s claim. The denial letter explained:
[508]*508The damage to the construction project was a sole and direct result of the marginal shoring design and faulty installation of the shoring. The policy excludes loss caused by deficiency in design and loss caused by faulty workmanship. Coverage will exist for any resulting loss caused by another insured event or peril. In this instance, the only peril, which caused the loss, was defective design and faulty workmanship, therefore there is no coverage for Vision One’s claims. To the extent any portion of the claim can be considered a resulting loss, other policy exclusions and limitations apply.
CP at 13136.
¶9 Vision asked Philadelphia to reconsider. Philadelphia responded by letter, clarifying its reason for denying the claim:
While the faulty workmanship exclusion contains an exception for resulting loss from a Covered Cause of Loss, in [this] case, the only cause of the loss was defective design and faulty workmanship. There is no separate and independent loss that resulted in the claimed damage. Therefore, the faulty workmanship exclusion bars coverage for this loss, and the ‘resulting loss’ provision contained therein does not apply.
CP at 13142 (emphasis omitted).
¶10 Vision sued Philadelphia for breach of contract; bad faith; and violations of the Consumer Protection Act, chapter 19.86 RCW.3 As trial approached, the parties thoroughly briefed how they believed various policy provisions applied for purposes of instructing the jury.
fll One of the parties’ points of contention involved causation. Vision claimed the collapse was caused by faulty equipment — a covered peril under the policy — while Philadelphia argued it was caused by faulty workmanship and [509]*509defective design, which were excluded perils.4 Vision further claimed that if an excluded peril and a covered peril both contributed to the property damage, then the policy would cover the loss. Philadelphia disagreed, arguing that in such a case, coverage follows only if the fact finder first determines that the covered peril was the efficient proximate cause of the loss.
¶12 The trial court agreed with Vision and entered the following order: “If it is found that the loss was caused by one or more non-excluded event(s) in combination with one or more excluded event(s); the loss is covered.” CP at 6588; see also Verbatim Report of Proceedings (VRP) (July 18, 2008) at 18. The trial court denied Philadelphia’s motion for reconsideration.
¶13 Vision and Philadelphia also disputed application of the resulting loss clause under the faulty workmanship exclusion. Because the wet concrete, rebar, and framing were “separate” from the faulty shoring installation, the trial court held that damage to the concrete, rebar, and framing constituted resulting losses. The court entered the following order on the resulting loss clause:
As a matter of law, for purposes of the faulty workmanship resulting loss clause in the contract between Vision One and Philadelphia, the shoring equipment is separate and distinct from the concrete, rebar and wood forms. Thus, any resulting loss or damage caused by the concrete collapse is covered by the policy language.
CP at 7099-7100.
¶14 Philadelphia moved for reconsideration, reiterating its argument that coverage under the resulting loss clause could apply only if the jury decided that faulty workmanship was the efficient proximate cause of the loss. VRP (Sept. 16, 2008) at 10. The trial court and Philadelphia’s counsel engaged in the following colloquy:
[510]*510[COUNSEL]: Well, you still have to have the jury determine what caused the loss, because -
THE COURT: I think they’re going to determine the damages. I think that’s where we’re at. That’s what I’m having a hard time understanding. I understand the wordsmithing that’s going on, but you, in your denial, say it’s bad installation; therefore, bad workmanship. And I’ve said the concrete is separate than the shoring.
[COUNSEL]: We’ve also said it’s bad design, Your Honor.
THE COURT: That’s okay.
[COUNSEL]: Well, it would only be okay if you decided that the only thing that caused this loss was workmanship.
THE COURT: We’ve been over whether there needs to be one cause or two, or multiple causes. If there is a cause that should be covered, then it’s all going to be covered.
[COUNSEL]: But it’s up to the jury to decide what the cause is. What if the jury says it’s all design? We don’t have to prove that, but what if the jury says that? Then you don’t have any resulting loss.
THE COURT: How can they say that when you say it’s installation, and your expert says it’s installation and it’s faulty workmanship? How can I let you present that to the jury and say, well, it’s not poor workmanship.
[COUNSEL]: We’re not going to say it’s not po[or] workmanship, but we’re also going to say it’s partly design. Then it’s up to the jury to decide.
Id. at 18-19.
f 15 The trial court disagreed with Philadelphia’s position, noting that Philadelphia’s letter denying coverage claimed the loss was caused by faulty workmanship as well as faulty design. The court ruled that Philadelphia’s denial letter precluded it from arguing that faulty workmanship was not a cause of the collapse. As a result, the court held that coverage existed as a matter of law and that the only issues remaining for trial were causation, bad faith, and damages.
f 16 The case proceeded to trial, and the jury returned a verdict in Vision’s favor. The jury awarded $251,023 in [511]*511damages for losses associated with repairing and reconstructing the collapsed portion of the floor. Vision did not receive damages for the cost of repairing the shoring, as the court had determined that the shoring loss was not covered under the policy. The jury also awarded Vision $724,605 for delay expenses resulting from the concrete collapse. Additional damages were awarded based on Philadelphia’s bad faith and violations of the Consumer Protection Act. The trial court entered judgment for Vision in the principal amount of $1,148,428, in addition to attorney fees.
¶17 Philadelphia appealed, arguing that the trial court committed legal error in its rulings on causation, resulting loss, and coverage. Philadelphia also assigned error to the damages and attorney fees award. Vision cross appealed, arguing that the trial court erred in its interpretation of the policy’s extra-expense endorsement. Vision claimed that the court’s error prevented Vision from presenting evidence of millions of dollars in lost profits resulting from the collapse.
¶18 The Court of Appeals agreed with Philadelphia that the trial court erred by failing to analyze coverage through the lens of the efficient proximate cause rule. Vision One, LLC v. Phila. Indem. Ins. Co., 158 Wn. App. 91, 105, 241 P.3d 429 (2010). Because the parties disagreed on which of three possible perils was the efficient proximate cause of the loss — faulty equipment (covered peril), defective design (excluded peril), or faulty workmanship (excluded peril subject to resulting loss clause) — the court remanded for a jury determination of causation. Id. at 106. The court also reversed the trial court’s ruling on the resulting loss clause, interpreting this clause to require a secondary covered peril to proximately cause an ensuing loss. Id. at 107. Finding no secondary covered peril that triggered Vision’s losses, the court reversed the trial court’s coverage determination and remanded for a new trial.5 Id. at 111. Vision petitioned this [512]*512court for review, which we granted. Vision One, LLC v. Phila. Indem. Ins. Co., 171 Wn.2d 1001, 249 P.3d 182 (2011).
ANALYSIS
¶19 Interpretation of language in an insurance policy is a question of law we review de novo. Allstate Ins. Co. v. Peasley, 131 Wn.2d 420, 423-24, 932 P.2d 1244 (1997) (citing Rones v. Safeco Ins. Co. of Am., 119 Wn.2d 650, 654, 835 P.2d 1036 (1992)).
¶20 Courts in Washington construe insurance policies as the average person purchasing insurance would, giving the language “ ‘ “a fair, reasonable, and sensible construction.”’” Key Tronic Corp. v. Aetna (CIGNA) Fire Underwriters Ins. Co., 124 Wn.2d 618, 627, 881 P.2d 201 (1994) (quoting Queen City Farms, Inc. v. Cent. Nat'l Ins. Co. of Omaha, 126 Wn.2d 50, 65, 882 P.2d 703, 891 P.2d 718 (1994) (quoting Grange Ins. Co. v. Brosseau, 113 Wn.2d 91, 95, 776 P.2d 123 (1989))). Where a term is undefined, it is assigned its ordinary meaning. Peasley, 131 Wn.2d at 424. Ambiguities in the policy are construed against the drafter-insurer. Queen City Farms, 126 Wn.2d at 68 (citing Greer v. Nw. Nat’l Ins. Co., 109 Wn.2d 191, 201, 743 P.2d 1244 (1987)). Because “[e]xclusions from insurance coverage are contrary to the fundamental protective purpose of insurance,” we construe exclusions strictly against the insurer. State Farm Fire & Cas. Co. v. Ham & Rye, LLC, 142 Wn. App. 6, 13, 174 P.3d 1175 (2007) (citing Stuart v. Am. States Ins. Co., 134 Wn.2d 814, 818-19, 953 P.2d 462 (1998)). “[W]e will not extend [exclusions] beyond their clear and unequivocal meaning.” Id.
[513]*513¶21 The policy here was an all-risk policy with a resulting loss clause.6 Because an understanding of how all-risk coverage and resulting loss provisions operate is necessary to our analysis, we describe these concepts before analyzing the particular provisions at issue.
All-Risk Insurance Coverage
¶22 Property insurance policies generally fall into two categories: named-peril and all-risk. In insurance parlance, “ ‘ “perils” . . . refers to fortuitous, active, physical forces such as lightning, wind, and explosion, which bring about the loss.’ ” Garvey v. State Farm Fire & Cas. Co., 48 Cal. 3d 395, 406, 770 P.2d 704, 257 Cal. Rptr. 292 (1989) (quoting Michael E. Bragg, Concurrent Causation and the Art of Policy Drafting: New Perils for Property Insurers, 20 Forum 385, 386-87 (1985)).
f23 “ ‘Named perils’ or ‘specific perils’ policies provide coverage only for the specific risks enumerated in the policy and exclude all other risks.” Steven Plitt, Daniel Maldonado & Joshua D. Rogers, Introductory Concepts of the Risk; Public Policy Insurability, and Causation, in 7 Couch on Insurance 3d § 101:7, at 101-17 (2006). All-risk policies, on the other hand, “provide coverage for all risks unless the specific risk is excluded.” Id.; see also Findlay v. United Pac. Ins. Co., 129 Wn.2d 368, 378, 917 P.2d 116 (1996) (noting that in an all-risk policy, “any peril that is not specifically excluded in the policy is an insured peril”); Villella v. Pub. Emps. Mut. Ins. Co., 106 Wn.2d 806, 816, 725 P.2d 957 (1986) (noting that “[i]n the case of all risk homeowners insurance, the peril insured against would be any peril that is not specifically excluded”); McDonald v. State Farm Fire & Cas. Co., 119 Wn.2d 724, 731 n.5, 837 P.2d 1000 (1992) (describing all-risk insurance as “ ‘a prom[514]*514ise to pay upon the fortuitous and extraneous happening of loss or damage . . . from any cause whatsoever, . . . except when occasioned by the wilful or fraudulent act or acts of the insured’ ” (alterations in original) (quoting 2 Warren Freedman, Richards on the Law of Insurance § 212 (5th ed. 1952))).
¶24 All-risk policies generally allocate risk to the insurer, while specific peril policies place more risk on the insured. Gust K. Newberg Constr. Co. v. E.H. Crump & Co., 818 F.2d 1363, 1364 (7th Cir. 1987) (“[U]nder an ‘all risk’ policy, the insurer bears the risk that a catastrophe not mentioned in the policy will occur; in a ‘specified peril’ policy, the insured bears that risk.”); Frank Coluccio Constr. Co. v. King County, 136 Wn. App. 751, 767, 150 P.3d 1147 (2007) (noting that the purpose of all-risk builder’s insurance “is to shift the risk of loss away from the contractor and the owner and to place it upon an insurer”).
¶25 In both types of property insurance, coverage is commonly triggered — or excluded — when a specified peril “causes” a loss. See Garvey, 48 Cal. 3d at 406. That is, a property insurance policy might provide that it does not insure or cover “loss caused directly or indirectly by” an enumerated peril. E.g., Wright v. Safeco Ins. Co. of Am., 124 Wn. App. 263, 273, 109 P.3d 1 (2004).
Ensuing Loss
¶26 While coverage may be excluded when a certain peril causes a loss, a resulting or ensuing loss clause operates to carve out an exception to the policy exclusion. McDonald, 119 Wn.2d at 734; Capelouto v. Valley Forge Ins. Co., 98 Wn. App. 7, 16, 990 P.2d 414 (1999). For example, a policy could exclude losses “caused directly or indirectly” by the peril of “defective construction,” but then an ensuing loss provision might narrow the blanket exclusion by providing that “any ensuing loss not excluded is covered.” E.g., Wright, 124 Wn. App. at 273.
[515]*515f 27 In this way, ensuing loss clauses limit the scope of what is otherwise excluded under the policy. Such clauses ensure “that if one of the specified uncovered events takes place, any ensuing loss which is otherwise covered by the policy will remain covered. The uncovered event itself, however, is never covered.” McDonald, 119 Wn.2d at 734.
¶28 An example helps illustrate how the ensuing loss clause works. Suppose a contractor miswires a home’s electrical system, resulting in a fire and significant damage to the home. And suppose the homeowner’s policy excludes losses caused by faulty workmanship, but the exclusion contains an ensuing loss clause. In this situation, the ensuing loss clause would preserve coverage for damages caused by the fire. But it would not cover losses caused by the miswiring that the policy otherwise excludes. Nor would the ensuing loss clause provide coverage for the cost of correcting the faulty wiring.
¶29 Ensuing loss clauses may not cover losses that are otherwise excluded. For example, in Wright, 124 Wn. App. at 267, 276, the Court of Appeals held that an ensuing loss provision did not provide coverage when the insured’s condominium unit was damaged by mold after defective construction allowed water to seep through the walls. While the policy excluded losses caused by defective construction and mold, id. at 273, 276, the defective construction exclusion stated that “ ‘any ensuing loss not excluded or excepted in this policy is covered,’ ” id. at 273 (emphasis omitted). Although mold damage arguably would have been covered under the ensuing loss clause, the fact that the policy excluded all losses caused by mold was dispositive. Id. at 274-75. Accordingly, the Court of Appeals held that the ensuing loss clause did not operate to provide coverage for a specifically excluded loss. Id.
130 Similar reasoning was followed in Allianz Insurance Co. v. Impero, 654 F. Supp. 16 (E.D. Wash. 1986). The policy at issue excluded losses caused by defective construction but covered damage “ ‘resulting from’ ” faulty or defective [516]*516construction. Id. at 17. The insured sought coverage for the cost of repairing poorly constructed concrete walls. Id. The court held that this loss fell within the exclusion, and not the ensuing loss clause, because “[t]he defective concrete caused no damage to any other portion of the structure, other persons or property.” Id. at 18. The court suggested that if “the wall, as a result of the deficiencies in the concrete, [had] collapsed and caused damage to some other portion of the work, or to equipment of a subcontractor or some similar thing,” the outcome would be different. Id.
¶31 As these cases illustrate, the dispositive question in analyzing ensuing loss clauses is whether the loss that ensues from the excluded event is covered or excluded. If the ensuing loss is also an excluded peril or an excluded loss under the policy, there is no coverage. See, e.g., Wright, 124 Wn. App. at 274-75; McDonald, 119 Wn.2d at 734. But if the policy covers the peril or loss that results from the excluded event, then the ensuing loss clause provides coverage.
¶32 Because an all-risk policy covers any loss that the policy does not expressly exclude, it is important to read the ensuing loss clause in the context of the coverage that was contemplated by the parties. One court colorfully described how absurd results could otherwise follow:
As an “all-risk” policy, this insurance policy basically covers everything unless specifically excluded. That means the number of possibilities for last-in-time “but for” causes of damage are limited only by the imagination of the reader. . . . What if faulty construction allows humid summer air to enter the building, which rusts metal fixtures? But for the exposure to the summer air, no damage to the fixtures would have occurred. Yet the contract does not exclude damages caused by “air.” Coverage? What if a poorly constructed ceiling beam falls, smashing the floor below? But for the force of gravity, no damage to the floor would have occurred. Yet the contract does not exclude damages caused by “gravity.” Coverage?
TMW Enters., Inc. v. Fed. Ins. Co., 619 F.3d 574, 576-77 (6th Cir. 2010) (citation omitted). In sum, we look to the lan[517]*517guage of the policy to ensure that the parties contemplated coverage for the ensuing loss.
¶33 With these considerations in mind, we turn to the policy language at issue.
Vision’s Policy
¶34 Here, Vision purchased an “all-risk” policy: it covered losses to the building “caused by or resulting from” any peril unless the loss was excluded. Excluded losses included those “caused by or resulting from” defective design and faulty workmanship, but the faulty workmanship exclusion contained an ensuing loss exception. Vision sought coverage for what it claims was the ensuing loss from the collapse: the clean-up, repair, and reconstruction of the collapsed concrete floor. Thus, we first examine the policy language to determine whether coverage exists for the ensuing loss.
Ensuing Loss Clause
¶35 Under the ensuing loss clause, damages resulting from faulty workmanship are covered if they are caused by an otherwise covered event. Because Vision sought coverage for losses associated with the collapse of the floor resulting from faulty workmanship, we consider whether “collapse” was covered under the policy.
¶36 The all-risk policy did not exclude the peril of collapse. Moreover, it affirmatively appears the parties intended coverage for collapse. For example, one section in Vision’s policy excludes losses “caused by or resulting from” the peril of “[r]ain, snow, sleet, or ice.” CP at 5977. This exclusion is followed by a clause stating, “But we will pay for loss’... [claused by collapse of the building or structure resulting directly from the weight of rain, snow, sleet, or ice.” Id. (emphasis added). The risk of collapse is, of course, at its highest during the construction phase of a building project — and this is an all-risk builder’s policy. Philadelphia does not argue that collapse was a risk beyond the reasonable contemplation of the policy.
[518]*518¶37 Despite this, the Court of Appeals found no coverage under the ensuing loss clause, reasoning that “[i]f faulty workmanship was the initial excluded peril [,] then the simultaneous collapse of the shoring and concrete slab was the loss. Had the collapse triggered a secondary covered peril, such as a fire, then damage caused by the fire would be covered as a resulting loss.” Vision One, 158 Wn. App. at 108 n.3. This analysis fails to consider that collapse is a covered peril under the policy. Many events can be characterized as both a loss and a peril. Characterizing collapse as the loss, rather than the peril, rests on a semantic distinction without a difference and ignores the policy’s coverage for all risks, including those “[claused by collapse of the building.” CP at 5977.
¶38 Indeed, the situation here is no different than if the shoring had given way, scraped a wall as it fell, and sparked a fire. The faultily assembled shoring (excluded) would have caused a fire (covered), resulting in damage to the building — a loss admittedly covered under the ensuing loss clause. In fact, Philadelphia’s coverage expert acknowledged at trial that there is no meaningful difference between a fire loss and a collapse loss under the ensuing loss clause. VRP (Oct. 13, 2008) at 1242-43. We agree and hold that the collapse damages are a covered ensuing loss under the policy.
Causation and the Efficient Proximate Cause Rule
¶39 Although collapse damages are covered under the faulty workmanship resulting loss clause, Philadelphia contends we must remand for the jury to determine whether the damages were in fact caused by faulty workmanship. In Philadelphia’s view, the resulting loss clause applies only if the jury finds that faulty workmanship was the efficient proximate cause of the loss. If the efficient proximate cause was defective design — an excluded peril without an ensuing loss clause — Philadelphia contends coverage must be denied.
[519]*519¶40 Philadelphia’s argument persuaded the Court of Appeals, which held that the trial court was required to submit the causation issue to the jury under the efficient proximate cause rule. Vision One, 158 Wn. App. at 106. The court determined that Philadelphia was not precluded by its denial letter, which stated that the damage resulted from faulty workmanship and defective design, from raising efficient proximate cause because it “is a rule of insurance contract construction, not a new ground for denying coverage.” Id. at 103. Reasoning that the jury did not have an opportunity to apply the efficient proximate cause rule, the Court of Appeals remanded on the issue of causation.
¶41 This analysis misses the mark for several reasons. First, the Court of Appeals misapprehended the efficient proximate cause rule. The efficient proximate cause rule is a rule of law, but not as the Court of Appeals described. The efficient proximate cause rule applies only when two or more perils combine in sequence to cause a loss and a covered peril is the predominant or efficient cause of the loss. McDonald, 119 Wn.2d at 732; Kish v. Ins. Co. of N. Am., 125 Wn.2d 164, 170, 883 P.2d 308 (1994). In such a situation, the efficient proximate cause rule mandates coverage, even if an excluded event appears in the chain of causation that ultimately produces the loss. Safeco Ins. Co. of Am. v. Hirschmann, 112 Wn.2d 621, 628, 773 P.2d 413 (1989). The efficient proximate cause rule operates as an interpretive tool to establish coverage when a covered peril “sets other causes into motion which, in an unbroken sequence, produce the result for which recovery is sought.” McDonald, 119 Wn.2d at 731.
¶42 The opposite proposition, however, is not a rule of law. When an excluded peril sets in motion a causal chain that includes covered perils, the efficient proximate cause rule does not mandate exclusion of the loss. Key Tronic Corp., 124 Wn.2d at 626. We explained in Key Tronic Corp.:
[520]*520The insurer here reasons that the converse of this rule should apply, i.e., where an excluded risk sets in motion a causal chain, coverage should be precluded as to all the causal events in the chain. As Key Tronic aptly points out, however, the efficient proximate cause rule operates in favor of coverage. A converse rule would, of course, operate in favor of no coverage. Other than maintaining that the efficient proximate cause rule should apply to excluded acts which are the efficient proximate cause, and that it is “settled law”, the insurer offers no other reasoning why the rule should be applied to preclude coverage. Because policies should normally be construed in favor of coverage, because there is no settled law favoring this argument, contrary to the insurer’s claim, and because the insurer does not offer any further justification or authority supporting such a rule, we decline to adopt the rule urged by the insurer.
Id.
f 43 We have left open the possibility that an insurer may draft policy language to deny coverage when an excluded peril initiates an unbroken causal chain. Findlay, 129 Wn.2d at 376. But a policy provision is not the same as a controlling rule of insurance contract interpretation. A provision must be asserted as a basis for denying coverage, and during litigation insurers may be precluded from asserting new grounds for denying coverage. Hayden v. Mut. of Enumclaw Ins. Co., 141 Wn.2d 55, 63, 1 P.3d 1167 (2000). Here, the trial court ruled that Philadelphia could not assert grounds for denying Vision’s claim other than those raised in its denial letter. CP at 5723. This order was not appealed.
¶44 In accordance with the trial court’s order, coverage must be determined under the policy language Philadelphia relied upon when it denied coverage. Under the policy, loss or damage is considered to have been “caused” by an excluded event in two instances: (1) when the excluded event “[djirectly and solely” causes the loss or damage or (2) when the excluded event “[initiates a sequence of events that results in loss or damage ...” CP at 5971. But in its [521]*521denial letter, Philadelphia stated that the collapse was “a sole and direct result of the marginal shoring design and faulty installation of the shoring.” CP at 13136. This letter makes clear that Philadelphia relied on the first causation prong, and not the “sequence of events initiated by an excluded event” prong.
¶45 While this second prong (i.e., when the excluded event “initiates a sequence of events that results in loss or damage”) may permit the sort of inverse efficient proximate cause analysis we allowed for in Findlay, Philadelphia never invoked this clause in denying Vision’s claim. Causation is properly analyzed in accordance with the policy provision invoked in Philadelphia’s denial letter. Therefore, Philadelphia is precluded from arguing the second prong of causation, i.e., that defective design may have initiated a sequence of events resulting in loss or damage.
¶46 The Court of Appeals erred when it turned to the efficient proximate cause rule to override the trial court’s decision. The trial court correctly determined causation as a matter of law because Philadelphia consistently maintained that faulty workmanship and defective design combined to “directly and solely” cause the loss, and it offered no evidence that would allow a jury to find that the loss was caused directly and solely by faulty design. Because there is coverage under the faulty workmanship exclusion in light of the ensuing loss clause, as a matter of law Vision’s loss was not caused “directly and solely” by an excluded peril. The trial court correctly ruled that Vision’s loss from the collapse is covered.7
¶47 Even if we allowed Philadelphia to belatedly rely on the “sequence of events” causation clause in its policy, nothing in the record supports its application here. The language allows for a denial of coverage when an excluded peril “initiates a sequence of events” in a causal [522]*522chain, resulting in the loss. The record does not show the defective design of the shoring initiated a causal chain of events. To the contrary, the engineering reports Philadelphia relied upon indicated that both inadequate design and poor installation caused the shoring to fail. To the extent defective design and faulty workmanship combined to cause the loss, they acted as concurrent causes; there is no indication the faulty design caused the faulty workmanship. In short, the trial court’s ruling on causation was correct, and there is no issue of fact for the jury meriting remand. We therefore reverse the Court of Appeals.
Extra-Expense Endorsement
¶48 In its cross appeal, Vision contends the trial court erred in interpreting the policy’s extra-expense endorsement to preclude Vision from presenting evidence of millions of dollars of delay losses following the collapse. The extra-expense endorsement provided $1 million in coverage for certain losses “incur[red] as the result of the project being delayed,” when the delay is “directly caused by any of the Covered Causes of Loss.” CP at 5985. The endorsement provided a list of covered losses, including “[c]onstruction loan interest,” “[r]eal estate and property taxes,” and “[l]egal and accounting fees.” Id.
¶49 The trial court concluded that these soft costs listed in the extra-expense endorsement were limited to the $1 million amount provided in the endorsement because the general policy — which had a $12.5 million limit — did not cover these soft costs. Vision claims the trial court erred because the losses listed in the endorsement were included in the policy’s general grant of coverage. Vision argues the extra-expense endorsement was designed to provide an additional $1 million for the specified delay losses in the event the $12.5 million limit was exhausted.
f 50 Vision’s policy covered “direct physical ‘loss’ to Covered Property” caused by a covered peril. CP at 5973. The soft costs listed in the extra-expense endorsement were [523]*523neither “physical” losses nor losses to covered “property.” Nevertheless, Vision claims that because soft costs were not excluded under the policy, they were covered. Vision argues that the policy’s exclusion of losses “caused by or resulting from . . . [d] elay ... or any other consequential loss,” CP at 5977, applied only to physical losses, not financial losses.
¶51 Vision’s argument fails because it wrongly assumes the policy covered financial losses in the first instance. While Vision’s policy was an all-risk policy, coverage extended only to “physical” losses to covered “property.” CP at 5973. It is of no consequence that soft costs were not specifically enumerated under the delay-and-consequential-loss exclusion. Because the policy did not cover soft costs, there was no need to exclude them. While the extra-expense endorsement provided coverage for certain financial losses resulting from the delay, the endorsement limited coverage to $1 million. We affirm the trial court on this issue.
Attorney Fees
¶52 Vision requests attorney fees on appeal. Because it has demonstrated coverage under the policy, Vision is entitled to fees under Olympic Steamship Co. v. Centennial Insurance Co., 117 Wn.2d 37, 51-54, 811 P.2d 673 (1991).
CONCLUSION
¶53 The trial court correctly ruled that Vision is entitled to coverage under its builders’ risk policy with Philadelphia. While the policy excludes losses caused by faulty workmanship, the ensuing loss clause at issue covers the collapse damages. Accordingly, we reverse the Court of Appeals and reinstate the trial court’s judgment. We also award Olympic Steamship fees to Vision.
Madsen, C.J.; C. Johnson, Chambers, Owens, Fairhurst, and J.M. Johnson, JJ.; and Alexander and Korsmo, JJ. Pro Tem., concur.
Reconsideration denied July 31, 2012.