In the

    United States Court of Appeals
                For the Seventh Circuit
                    ____________________
No. 19-1426
LEXINGTON INSURANCE COMPANY and NATIONAL
UNION FIRE INSURANCE COMPANY OF PITTSBURGH, PA,
                                     Plaintiffs-Appellants,

                                v.

RLI INSURANCE COMPANY,
                                               Defendant-Appellee.
                    ____________________

        Appeal from the United States District Court for the
                      Central District of Illinois.
       No. 1:17-cv-01514-JBM-JEH — Joe Billy McDade, Judge.
                    ____________________

  ARGUED NOVEMBER 7, 2019 — DECIDED JANUARY 27, 2020
               ____________________

   Before HAMILTON, SCUDDER, and ST. EVE, Circuit Judges.
   HAMILTON, Circuit Judge. In this contract dispute, two in-
surers of New Prime, Inc., a trucking company, accuse a third
insurer of not paying its share toward two multimillion-dollar
personal injury settlements. Plaintiffs Lexington Insurance
Company and National Union Fire Insurance Company con-
tend that defendant RLI Insurance Company underpaid
2                                                              No. 19-1426

according to the policy it sold to New Prime, leaving National
Union to make up the difference.
    In the district court, Lexington and National Union sought
a declaratory judgment as to the meaning of the RLI Policy
and equitable contribution of $2.5 million from RLI toward
the settlements in question. Both sides moved for summary
judgment. Both based their motions on the language of the
RLI Policy and on extrinsic evidence of the parties’ intent. The
district court granted summary judgment to RLI, relying ex-
clusively on contract language that it found unambiguous.
We affirm. The text of the RLI Policy is not as clear to us as it
was to the district court, but undisputed extrinsic evidence
shows that RLI’s position is correct.
I. Factual and Procedural Background
    As a large commercial trucking company, New Prime
faces substantial risks of tort liability. In the relevant years,
New Prime managed and covered its own liability without
insurance for the first $3 million of exposure per occurrence.
But to protect itself from unusually large claims, New Prime
bought excess liability insurance from three different compa-
nies: RLI, Lexington, and National Union. (Lexington and Na-
tional Union are both wholly owned subsidiaries of the Amer-
ican International Group, Inc., and for most purposes we can
refer to them together as AIG.) In contracting with several dif-
ferent insurers, New Prime followed a common industry
practice to stack policies into sequential “layers” of excess in-
surance coverage.1 This case concerns the threshold of liabil-
ity at which RLI’s responsibility ended and AIG’s began.


1 See Scott M. Seaman & Charlene Kittredge, Excess Liability Insurance: Law

and Litigation, 32 Tort & Ins. L.J. 653, 654–55 (1997) (“An insured’s liability
No. 19-1426                                                              3

     The facts of the layers’ sequential ordering are undis-
puted. New Prime, through its insurance agent Cottingham &
Butler, contacted RLI in October 2011 and purchased a $2 mil-
lion policy largely overlapping with the calendar year 2012.
New Prime renewed the $2 million RLI Policy, with one rele-
vant change we discuss later, for the years 2013, 2014, and
2015. Not until May 1, 2014 did New Prime, this time through
the agents AmWINS and Regions, obtain a $5 million policy
from Lexington to sit above RLI’s layer. New Prime already
had a $25 million “umbrella” policy from National Union to
sit above Lexington’s layer, and New Prime renewed that pol-
icy for the year starting May 1, 2014 as well. Both AIG policies
were then renewed without changes for the year beginning on
May 1, 2015.2
    The two tragic accidents that led to this lawsuit occurred
in 2015, when New Prime was covered by the RLI, Lexington,
and National Union policies. On March 4, 2015, a New Prime
tractor-trailer drifted into the median of Interstate 80 in Mer-
cer County, Pennsylvania. The truck struck and severely in-
jured Daniel Montini, who was changing a flat tire. See Mem-
orandum Order, Montini v. New Prime, Inc., No. 2:15-cv-1591,
slip op. at 1 (W.D. Pa. Nov. 15, 2017). In February 2018, Mon-
tini settled his lawsuit against New Prime for $16 million. On
December 28, 2015, near Santa Rosa, New Mexico, a New


insurance program generally includes a layer of primary insurance or self-
insurance coverage followed by one or more layers of excess insurance.”).
2 “Umbrella insurance” is a generic term for excess insurance that does not

perfectly “follow form” with the underlying insurance. Umbrella cover-
age can be both broader and narrower than the lower layers, depending
on the specifics of the contracts. See generally Seaman & Kittredge, cited
above in note 1, at 660.
4                                                    No. 19-1426

Prime tractor-trailer rear-ended a sedan driven by Katherine
and Samuel Herrera. Both were killed. See Complaint, Tafoya
v. New Prime, Inc., No. D412CV201600190 (N.M. Dist. May 19,
2016), 

. In March 2018, the Herreras’ estates
settled their claims against New Prime for $20 million.
    The dispute here is over how much RLI needed to contrib-
ute first to the Herrera settlement and then to the later Montini
settlement, which were so large as to trigger the excess insur-
ance policies. The parties agree that, starting from the first
dollar, New Prime itself was required to cover $3 million of
costs or losses for each occurrence because of the so-called
“Self-Insured Retention” built into the RLI Policy. In effect,
the Self-Insured Retention made New Prime its own primary
insurer up to $3 million per occurrence, with both RLI and
AIG providing forms of excess insurance. See, e.g., Kajima
Const. Servs., Inc. v. St. Paul Fire & Marine Ins. Co., 

, 313 (Ill. 2007) (“Excess insurance coverage attaches only
after a predetermined amount of primary insurance or self-
insured retention has been exhausted.” (quotation omitted)).
    The RLI Policy provided the next layer of coverage but
came with a feature called the “Aggregate Corridor Deducti-
ble” or “ACD,” which is the central focus of this lawsuit. Read
alone, the main text of the RLI Policy would have provided
for the first $2 million of coverage immediately above New
Prime’s $3 million Self-Insured Retention. An endorsement to
the Policy, however, added the Aggregate Corridor Deducti-
ble and described its function (emphasis added):
       The Insured [i.e., New Prime] shall respond to,
       investigate, adjust, defend, and dispose of by
       payment or otherwise all losses and claims for
       losses covered by the Policy for which the total
No. 19-1426                                                 5

      claim is greater than the $3,000,000 Self Insured
      Retention (SIR) until the Aggregate Corridor De-
      ductible of $2,500,000 has been satisfied. Once the
      Aggregate Corridor Deductible has been ex-
      hausted by payment for one or more losses &
      “costs”, the Insured is only responsible for losses
      and “costs” up to [the] Per Occurrence Self In-
      sured Retention.
The endorsement also specified that the ACD amounts to
“$2,500,000 excess of the $3,000,000 Self Insured Retention.”
Although not a model of clarity, this endorsement obligated
New Prime to pay out an additional $2.5 million above its
Self-Insured Retention of $3 million per occurrence before RLI
began to pay. But while the Self-Insured Retention applied to
each covered incident, the $2.5 million ACD applied only
once per year, so New Prime needed to pay that additional
amount only once per policy year. On these basics RLI and
AIG agree.
   The dispute, however, is whether New Prime’s payments
toward the Aggregate Corridor Deductible diminished the
amount that RLI owed on any claims. RLI argues that the
ACD sat within RLI’s $2 million layer, leaving RLI with no re-
sponsibility for making any payment on any claim until New
Prime had both (a) paid $3 million per occurrence and (b)
paid the year’s ACD total on losses between $3 million and
$5 million per occurrence. If that is correct, then New Prime
and RLI would together owe at most $5 million on any claim:
the $3 million Self-Insured Retention plus the $2 million RLI
Policy. In contrast, AIG argues that the ACD sat below RLI’s
$2 million layer. On this view, RLI had to provide coverage
whenever the loss exceeded the sum of the Self-Insured
6                                                   No. 19-1426

Retention and remaining ACD balance. In other words, AIG
asserts, “the ACD operates … as an additional $2.5 million
self-insured retention applying per policy period.” If that
were correct, then AIG’s duty to pay would not have been
triggered until New Prime and RLI had together paid
$7.5 million for the first big occurrence(s) of the policy year.
The charts below illustrate the parties’ respective views on
how to allocate the first $10 million of the Herrera and Montini
settlements, showing a dispute over $2.5 million.
      RLI’s Position                      AIG’s Position

        AIG Coverage                        AIG Coverage

        RLI Coverage                        RLI Coverage

        New Prime ACD                       New Prime ACD

        New Prime SIR                       New Prime SIR


                                          $2.5
      $5.0         $5.0                                $5.0
                                          $2.0
      $0.0
      $2.0         $1.5                   $2.5         $2.0
                   $0.5                                $0.0
      $3.0         $3.0                   $3.0         $3.0


    Herrera     Montini                Herrera      Montini


    At the time of the Herrera and Montini settlements, RLI in-
sisted on its view of the Aggregate Corridor Deductible. It
paid none of the Herrera settlement and only $1.5 million of
the Montini settlement. AIG reserved its rights to recoup the
alleged deficit of $2.5 million from RLI. Because the settle-
ments exhausted Lexington’s policy layer on any view of the
No. 19-1426                                                       7

RLI Policy, it is National Union that seeks equitable contribu-
tion through this lawsuit.
    The parties filed cross-motions for summary judgment in
the district court. Both RLI and AIG attached to their motions
extrinsic evidence, including email correspondence, under-
writing files, and employee affidavits. Both sides referred to
extrinsic evidence and disputed its significance in their sum-
mary judgment briefing. The district court declined to con-
sider the evidence. The court found that the RLI Policy pro-
vided unequivocally that payments toward the Aggregate
Corridor Deductible erode RLI’s policy layer. The court en-
tered summary judgment for RLI on that basis, and AIG ap-
pealed.
II. Principles Governing Illinois Contract Interpretation
    The parties agree on appeal that Illinois law governs this
contract dispute. In interpreting any insurance policy, the
“primary function” of an Illinois court “is to ascertain and
give effect to the intention of the parties, as expressed in the
policy language.” Thounsavath v. State Farm Mut. Auto. Ins. Co.,

, 1244 (Ill. 2018). This analysis ends with the
text of the agreement if it is unambiguous: “A court will first
look to the language of the contract itself to determine the par-
ties’ intent. … If the words in the contract are clear and unam-
biguous, they must be given their plain, ordinary and popular
meaning.” Thompson v. Gordon, 

, 47 (Ill. 2011).
On the other hand, if the contract language is ambiguous, Il-
linois courts may consider extrinsic evidence of the parties’
intent. See id.; Gallagher v. Lenart, 

, 58 (Ill. 2007).
On appeal, both AIG and RLI maintain that the RLI Policy was
unambiguous in their favor; RLI also argues that the extrinsic
8                                                     No. 19-1426

evidence provides an alternate basis for affirming the judg-
ment in its favor.
    Illinois courts treat contracts as ambiguous where the lan-
guage of the contract is “susceptible to more than one mean-
ing,” 

, or “obscure in meaning
through indefiniteness of expression.” Central Illinois Light Co.
v. Home Ins. Co., 

, 213 (Ill. 2004). “All the provi-
sions of the insurance contract, rather than an isolated part,
should be read together to interpret it and to determine
whether an ambiguity exists.” Rich v. Principal Life Ins. Co., 

, 1090 (Ill. 2007), quoting United States Fire Ins. Co.
v. Schnackenberg, 

, 1205 (Ill. 1981). The mere
fact of disagreement between the parties does not render lan-
guage ambiguous, of course. 

. Con-
versely, however, and equally obviously, “a contract is not
necessarily unambiguous when, as here, each party insists
that the language unambiguously supports its position.” Cen-
tral Illinois 

.
     In practice, Illinois courts do not hesitate to order consid-
eration of extrinsic evidence if contract language alone cannot
resolve a dispute. E.g., Shapich v. CIBC Bank USA, 

, 99–100 (Ill. App. 2018) (reversing grant of summary judg-
ment and remanding to trial court because both parties’ inter-
pretations contradicted portions of text); Morningside North
Apartments I, LLC v. 1000 N. LaSalle, LLC, 

, 419
(Ill. App. 2017) (same). With these principles in mind, we first
consider the text of the RLI Policy.
III. Textual Analysis
   The district court’s construction of the RLI Policy was a
legal finding, so we review it de novo. See, e.g., Soarus L.L.C. v.
No. 19-1426                                                     9

Bolson Materials Int’l Corp., 

, 1011 (7th Cir. 2018).
We agree with much of the court’s analysis, but we ultimately
cannot agree that the contract language unambiguously spec-
ified how the Aggregate Corridor Deductible operated.
    The basic problem is that the RLI Policy failed to define
the custom-tailored Aggregate Corridor Deductible feature or
to describe its mechanics with precision. On the key question
in this case—whether payments toward the ACD erode the
policy limit—the contract was silent. A lengthy definitions
section included such diverse terms as “bodily injury,” “dam-
ages,” “nuclear material,” “we,” and “you.” Yet the word “de-
ductible” appeared nowhere in the main text of the agree-
ment. Without a policy definition, three endorsements to the
Policy created the ACD and described some aspects of its op-
eration. These provisions did not explicitly state whether the
ACD eroded policy limits. We find that neither the reasoning
of the district court nor the arguments of the parties fully dis-
pel this ambiguity.
   A. Inconclusive Usage and Custom Arguments
    We first address a genre of argument on which both sides
rely heavily. AIG contends that the Aggregate Corridor De-
ductible actually functioned as a “self-insured retention.” RLI
insists that it operated as a “deductible,” as its name would
suggest. With extensive citation to non-Illinois cases and sec-
ondary literature, the parties expound on these terms’ mean-
ings in insurance law, which they believe clarify the role of
the ACD. Both sides, for example, cite Judge Hellerstein’s
opinion in the World Trade Center litigation, which explained
the general distinction:
10                                                             No. 19-1426

        A [self-insured retention] differs from a deduct-
        ible in that a SIR is an amount that an insured
        retains and covers before insurance coverage
        begins to apply. Once a SIR is satisfied, the in-
        surer is then liable for amounts exceeding the
        retention, less any agreed deductible. … In con-
        trast, a deductible is an amount that an insurer
        subtracts from a policy amount, reducing the
        amount of insurance.
In re Sept. 11th Liability Ins. Coverage Cases, 

,
124 n.7 (S.D.N.Y. 2004), citing 2 Barry R. Ostrager & Thomas
R. Newman, Handbook on Insurance Coverage Disputes
§ 13.13[a] (12th ed. 2004).3
    AIG and RLI both seek to label the ACD as one or the other
and thus to determine how the RLI Policy worked. The name
for such reasoning about contractual language is usage and
custom. Courts have long recognized that technical trade us-
ages can sometimes clarify contract terms. See Shreffler v.
Nadelhoffer, 

, 633 (Ill. 1890) (directing courts “to col-
lect the real intention of the parties from the terms used in the

3 AIG’s brief omitted this excerpt’s second sentence, which tends to favor
RLI’s position in this lawsuit, and, more troubling, did not even use an
ellipsis or any other signal that text was omitted. See Appellant’s Br. at 21.
AIG’s counsel initially said he thought the omitted sentence “wasn’t rele-
vant” to this case. He then retreated to the assertion that the omission was
not intended to mislead the court, and he apologized to the court. Oral
Arg. 35:20–36:30. We remind all counsel that they must not “knowingly
misrepresent, mischaracterize, misquote, or miscite facts or authorities in
any oral or written communication to the court.” Lawyers’ Duties to the
Court, No. 5, Standards for Professional Conduct Within the Seventh Fed-
eral Judicial Circuit; see also Illinois Rule of Professional Conduct 3.3(a)(2)
& cmt. [4].
No. 19-1426                                                    11

contract, taking them in their plain, ordinary, and popular
sense, unless by the known usage of the trade they have ac-
quired a peculiar sense”); see generally 12 Richard A. Lord,
Williston on Contracts § 34:5 (4th ed. 2019) (“It is currently the
widely accepted rule that custom and usage may be proved
to show the intention of the parties to a written contract or
other instrument in the use of phrases of a peculiar technical
meaning which, when unexplained, are susceptible of two or
more plain and reasonable constructions.”).
    Along these lines, AIG and RLI both seek to import what
they say are established industry understandings into the RLI
Policy. At oral argument, counsel for AIG asserted that the
ACD “bears every single characteristic of a self-insured reten-
tion recognized by the cases”—in function if not in name.
Counsel for RLI responded that it was an “industry practice”
and “accepted legal principle” that anything labeled a “de-
ductible” erodes policy limits unless the policy states other-
wise. Counsel went on to assert that New Prime and RLI “un-
derstood … what the word deductible meant; they didn’t
need to define it.” These are assertions that usage and custom
should guide our decision.
     A threshold question, then, is whether an Illinois court
would consider trade usage in this situation. Illinois law ap-
pears somewhat unsettled on just when parties may introduce
such evidence. Compare Illinois Ins. Guar. Fund v. Nwidor, 

, 1046 (Ill. App. 2018) (“Usage or custom is admis-
sible to explain or make clear what a contract means but not
to contradict a meaning obvious on the face of the instru-
ment.”), with Amalgamated Transit Worker’s Union, Local 241 v.
Pace Suburban Bus Div. of Reg’l Transp. Auth., 

, 40
(Ill. App. 2011) (“[E]ngrafted on every written contract are the
12                                                    No. 19-1426

customs, practices and definitions which are commonly un-
derstood and accepted by the parties.”). That said, the situa-
tion here blurs the line between trade usage and everyday tex-
tual interpretation; AIG and RLI resort not to extrinsic factual
evidence but to case law and treatises. We will assume for pre-
sent purposes that an Illinois court would consider such ma-
terials in a case like this. See Matter of Envirodyne Indus., Inc.,

, 305 (7th Cir. 1994) (“It would be passing odd to
forbid people to look up words in dictionaries, or to consult
explanatory commentaries that, like trade usage, are in the na-
ture of specialized dictionaries.”).
    In the end, though, the parties’ arguments from industry
usage are inconclusive. The debate revolves around the fact
that the RLI Policy gave New Prime primary authority to “re-
spond to, investigate, adjust, defend, and dispose of” claims
falling within the Aggregate Corridor Deductible. AIG con-
tends that a true deductible policy would task the insurer, not
the insured—RLI, not New Prime—with claims-processing
duties: only a self-insured retention would lodge them with
the insured. Authority supports this distinction as a matter of
industry practice. See, e.g., Monroe Guar. Ins. Co. v. Langreck,

, 495 (Ind. App. 2004) (“in a policy with a de-
ductible, the insurer retains complete control of claims han-
dling; in a policy with a retained amount, the insurer has no
claims handling responsibility”); 3 Jeffrey E. Thomas et al.,
New Appleman on Insurance Law Library Edition
§ 16.09[3][b][i] (2019) (“A ‘self-insured retention’ (or ‘SIR’) is
an insurance arrangement whereby the insured takes all re-
sponsibility for dealing with claims up to a certain amount of
loss.”). RLI responds that the Policy apportioned duties as it
did because it provided only excess insurance, with New
Prime serving as its own primary insurer. This view also finds
No. 19-1426                                                       13

support in case law and secondary literature. See, e.g., Royal
Ins. Co. v. Process Design Assocs., Inc., 

, 1242 (Ill.
App. 1991) (“the primary insurer, and not the excess carrier,
has the duty to defend its insured”); Seaman & Kittredge,
cited above at n.1, at 662 (“In contrast to the primary insurer,
the excess insurer rarely undertakes to defend the insured.”).
    This debate shows only that the usages of the insurance
industry do not resolve this case. As a deductible attached to
excess insurance, the Aggregate Corridor Deductible was a
customized term that did not mimic either a standard “de-
ductible” or a standard “self-insured retention” in every re-
spect. The parties were of course free to contract for non-
standard terms to suit their particular needs. See Nation Oil
Co. v. R. C. Davoust Co., 

, 265 (Ill. App. 1964)
(“parties to a contract may express intent contrary to a custom
of the trade”); see also Sulser v. Country Mut. Ins. Co., 

, 431 (Ill. 1992) (“Parties to a contract may agree to
any terms they choose unless their agreement is contrary to
public policy.”). Since New Prime and RLI agreed on such a
customized policy feature here, one that included at least
some features of both standard deductible terms and self-in-
sured retentions, received trade usage simply does not pro-
vide a reliable guide to their agreement. Instead, we move on
to other potential evidence of their mutual intent as expressed
in the language of the RLI Policy.
   B. Other Textual Arguments
   The parties also base arguments solely on the text of the
RLI Policy. We address these points both to give them due
consideration and to provide background for the extrinsic ev-
idence that we find decisive here. On balance, the text offers
support for RLI’s view that payments toward the Aggregate
14                                                            No. 19-1426

Corridor Deductible eroded policy limits, but not unambigu-
ously so. We start with the terms most favorable to RLI.
    First, RLI makes the simple point that the Policy used dis-
tinct terms for Self-Insured Retention and Aggregate Corridor
Deductible. The Policy defined both “Retained Limit” and
“Self-Insured Retention” without reference to the ACD.4 It
then obligated RLI to indemnify New Prime for losses “in ex-
cess of” the “Self-Insured Retention,” otherwise known as the
“retained limit,” again without reference to the ACD. The dis-
trict court reasoned that, as a result, the Self-Insured Reten-
tion and ACD were “entirely separate,” and the ACD neces-
sarily sat within RLI’s policy layer. This argument has merit
but we do not view it as conclusive. It ultimately depends too
much on the labels alone to decide how the ACD feature
works, and we rejected that path above because of other fea-
tures of the policy. Cf. Learning Curve Intʹl, Inc. v. Seyfarth
Shaw, LLP, 

, 1079 (Ill. App. 2009) (“We deter-
mine the character of a contract from its substantive effects,
not from the labels that parties prefer to place on its provi-
sions.”).
    Second, RLI points to the “Defense Endorsement,” which
set forth the division of labor between New Prime and RLI in
defending lawsuits and processing claims. The Endorsement
explained that New Prime “shall have the sole and unre-
stricted right to settle or pay any loss for which [RLI] has no

4 These two terms turn out to be synonymous: “Retained Limit” was de-
fined in the RLI Policy to “consist of a ‘self-insured retention’ or required
primary insurance, as indicated in the Declarations,” and the Declarations
indicated that no primary insurance was required under this policy. By
process of elimination, the Retained Limit was simply the Self-Insured Re-
tention.
No. 19-1426                                                              15

potential responsibility … as a result of the application of the
SIR and Corridor Deductible.” The district court reasoned,
and RLI argues, that unless ACD payments eroded policy lim-
its, RLI had at least potential responsibility for every filed
claim, rendering New Prime’s right to defend a nullity. AIG
responds that this formalistic definition of “potential” ignores
numerous claims that “pose nothing but the most tenuous
theoretical possibility of implicating RLI’s coverage.” That
contention is at least plausible since RLI’s coverage did not
even begin until New Prime itself had spent at least $5.5 mil-
lion in the policy year. The text alone does not resolve this de-
bate about the sense of the word “potential.” As we will see
later, extrinsic evidence does.
    Third, the district court placed “significant” weight on the
RLI Policy’s use of the word “aggregate” to describe the
$2 million indemnity limit, reasoning that New Prime’s ACD
payments and RLI’s indemnities sum up to “aggregate” cov-
erage. Therefore, this argument goes, ACD payments must
have reduced RLI’s share. We do not find this logic convinc-
ing. RLI itself did not assert this meaning of the word “aggre-
gate” in summary judgment briefing. The district court intro-
duced the argument, and RLI has endorsed it on appeal. As
AIG points out, however, the district court’s sense of “aggre-
gate” clashes with other parts of the policy, such as the Sepa-
ration of Insureds Condition. This argument also does not set-
tle the function of the ACD.5



5 On this point, RLI protests that the Separation of Insureds Condition
“does not use the term ‘aggregate’ anywhere,” Appellee’s Br. at 24, but it
did: “nothing contained herein shall operate to increase Our total aggregate
limit of indemnity in excess of Self-Insured Retention” (emphasis added).
16                                                   No. 19-1426

    Finally, on the other side, AIG’s strongest textual argu-
ment is that the RLI Policy contemplated that a single occur-
rence could exhaust the ACD. Recall that, if ACD payments
eroded the RLI Policy limit, then New Prime could spend at
most the $2 million limit toward the ACD on a single occur-
rence and thus could not exhaust the entire $2.5 million ACD.
Yet the Aggregate Corridor Deductible Endorsement ex-
pressly allowed that the ACD could be “exhausted by pay-
ment for one or more losses & ‘costs’” (emphasis added). An-
other provision directed New Prime to pay all costs until the
total on “one or more claims together equals the amount of the
Annual Aggregate Corridor Deductible” (emphasis added).
These two passages are not consistent with RLI’s interpreta-
tion of how the ACD functions because payments toward one
occurrence could not possibly exhaust the ACD of $2.5 mil-
lion.
    The district court held that the references to “one or more”
did not introduce ambiguity into the RLI Policy because this
“boiler-plate language” could not “render the policy as a
whole unclear.” If we were otherwise convinced that the text
unequivocally provided for ACD payments to erode the pol-
icy limit, we might agree that an “isolated part” of the contract
should not “determine whether an ambiguity exists.” Rich v.
Principal Life Ins. Co., 

, 1090 (Ill. 2007). But we
are not convinced of the premise, especially after setting aside
the inconclusive arguments regarding usage and custom. The
RLI Policy left the key term in this dispute undefined, and
both sides’ proffered interpretations of the ACD would in ef-
fect nullify some language in the contract, which is ordinarily
a result to be avoided in legal textual analysis. We therefore
conclude that, as applied to the dispute here, the custom-tai-
lored ACD feature of the RLI Policy was “obscure in meaning
No. 19-1426                                                    17

through indefiniteness of expression” and hence ambiguous.
Central Illinois Light Co. v. Home Ins. Co., 

, 213
(Ill. 2004).
IV. Extrinsic Evidence
    Undisputed evidence of the negotiations between New
Prime and RLI, and between New Prime and AIG, resolves
the ambiguity in RLI’s favor. As noted, Illinois courts may
consider extrinsic evidence upon a finding of ambiguity. See
Gallagher v. Lenart, 

, 58 (Ill. 2007). Under federal
procedural law, if uncontested facts clarify the meaning of a
contract, an appellate court may decide the issue as a matter
of law at summary judgment. See Citadel Grp. Ltd. v. Washing-
ton Reg’l Med. Ctr., 

, 587 (7th Cir. 2012) (“Even
when a contract is ambiguous, as long as the extrinsic evi-
dence bearing on the interpretation is undisputed and leads
to only one reasonable interpretation, we can decide the mat-
ter on summary judgment.”); see generally Fidelity & Deposit
Co. of Maryland v. Rotec Indus., Inc., 

, 949 (7th Cir.
2004) (“in all cases in federal court, including diversity cases,
the allocation of responsibility between judge (‘law’) and jury
(‘fact’) is governed by federal rather than state law”).
   Although the district court here ruled only on textual
grounds, the parties submitted extensive evidence and brief-
ing regarding the history of the insurance contracts at issue.
We may affirm summary judgment “on any ground that finds
support in the record,” so long as that ground was “ade-
quately presented in the trial court so that the non-moving
party had an opportunity to submit affidavits or other
18                                                              No. 19-1426

evidence and contest the issue.” Box v. A & P Tea Co., 

, 1376 (7th Cir. 1985).6
    Summary judgment for RLI was proper because emails
and underwriting files show that New Prime, RLI, and AIG
itself all intended the combined liability of New Prime and
RLI to be capped at $5 million per occurrence, so that AIG’s
liability would begin at $5 million per occurrence, not at $7 or
$7.5 million. As a matter of Illinois law and common sense,
the parties’ statements during negotiations and their conduct
afterward carry more weight than legal interpretations of-
fered in the run-up to litigation. See Szafranski v. Dunston, 

, 1157 (Ill. App. 2015) (“The intended meaning of
ambiguous contract language may be derived from the cir-
cumstances surrounding the formation of a contract or from
the conduct of the parties subsequent to its formation.”). We
first consider the record of negotiations prior to the Montini
accident in March 2015—not only between New Prime and
RLI but also between New Prime and AIG. Then we turn

6 Dicta in some of our decisions have incorrectly suggested an additional
requirement for alternate grounds for affirming summary judgment, that
“the district court adequately considered them.” Scheidler v. Indiana, 

, 540 (7th Cir. 2019); see also Nationwide Agribusiness Ins. Co. v.
Dugan, 

, 450 (7th Cir. 2015); Gerhartz v. Richert, 

,
685 (7th Cir. 2015); Costello v. Grundon, 

, 637 (7th Cir. 2011);
Best v. City of Portland, 

, 702 (7th Cir. 2009). But see, e.g., Levy
v. Marion Cty. Sheriff, 

, 1009 (7th Cir. 2019) (not requiring dis-
trict court to have considered the issue). The correct rule, as stated in 

, is that we may affirm summary judgment for reasons not
considered by the district court as long as the parties had a fair oppor-
tunity to present their arguments and evidence. See Singleton v. Wulff, 

, 120–21 (1976) (federal appellate courts usually do not decide an
issue not decided in the district court but have discretion to do so if parties
had fair opportunity to be heard on the issue).
No. 19-1426                                                  19

briefly to statements made during the defense of the claims at
issue in this lawsuit.
   A. Negotiations Between New Prime and RLI
    To start with identified textual ambiguities, the “one or
more” references that are so critical to AIG’s textual argu-
ments turn out to be mere vestiges of an earlier version of the
RLI Policy. The record contains the relevant portions of policy
in effect for 2012, the first year that New Prime and RLI con-
tracted for coverage. During that year, the Aggregate Corri-
dor Deductible was only $2 million—not $2.5 million—so
New Prime would have been able to exhaust it through losses
on a single claim even if ACD payments erode the policy. The
ACD remained at $2 million for 2013. New Prime and RLI
then agreed to increase it to $2.5 million for 2014. RLI
acknowledges that it and New Prime should have revised
“one or more claims” to “two or more claims” to reflect the
changed arithmetic. But in interpreting the ambiguity, we will
focus on New Prime and RLI’s objectively expressed mutual
intent, not just their drafting oversights. Cf. Regency Commer-
cial Assocs., LLC v. Lopax, Inc., 

, 316 (Ill. App.
2007) (in case of ambiguity, “a court may consider prelimi-
nary negotiations between the parties in order to determine
the meaning of contract provisions and the intent of the par-
ties”). We suspect any lawyer or judge who has edited a legal
document has had the experience of overlooking in the edit-
ing process a needed conforming change like this. AIG has
presented no contrary evidence to create an issue of material
fact on this point.
    Extrinsic evidence also clarifies the purpose of the Defense
Endorsement and its use of the word “potential,” which also
figures prominently in AIG’s textual arguments. On
20                                                             No. 19-1426

December 13, 2011, in the final days of the negotiations for the
first policy year, a New Prime employee emailed the broker
at Cottingham & Butler to explain why New Prime needed
the Aggregate Corridor Deductible. New Prime was con-
cerned that RLI could force it to settle claims “using our SIR
dollars,” in other words below $3 million where RLI had no
skin in the game.7 New Prime hence requested a “$2 M. an-
nual aggregate deductible” so that the “combination of SIR
plus the annual aggregate deductible equals the $5 M. per oc-
currence limit.” Although we do not have RLI’s response to
New Prime’s proposal, we know that the RLI Policy ulti-
mately incorporated the ACD as described. In this manner,
the ACD created a class of claims for which RLI had “no po-
tential responsibility,” in the sense of even theoretical possi-
bility. The Defense Endorsement granted New Prime “the
sole and unrestricted right to settle or pay” those claims. Of
course, this plan would work only if New Prime’s payments
toward the ACD did in fact erode the RLI Policy limit.
   Record evidence confirms that the 2013 version of the Pol-
icy carried forward this understanding. On November 30,
2012, RLI responded in writing to queries from Cottingham &
Butler regarding the renewal of New Prime’s coverage. This
document described the policy as “Primary Fleet Auto



7 The Third Circuit has observed that the structure of excess insurance cre-

ates the inherent conflict of interest New Prime identified: “the excess car-
rier [here, RLI] wishes the primary insurer [New Prime] to dispose of the
case within its limits and is not unduly impressed with the primary in-
surer’s desire to save some or all of its policy limits by a favorable verdict
at trial.” Puritan Ins. Co. v. Canadian Universal Ins. Co., 

, 78 (3d
Cir. 1985).
No. 19-1426                                                                21

Liability at $5,000,000 CSL per occurrence.”8 The responses
went on to explain that the ACD “was installed [in] Decem-
ber, 2011 to make-work an agreement with the underwriter
that PRIME would have sole control of defense and settlement
of all claims within the above layer.” Far from disputing the
accuracy of this document, AIG cites two phrases from it to
try to support its own position: “$2,000,000 Limit excess a
$3,000,000 SIR + $2,000,000 Corridor Deductible” and “policy
will remain on SIR, not a Deductible.” But these descriptions
merely reiterated basic aspects of the RLI Policy; they did not
contradict the clear evidence that New Prime and RLI under-
stood their collective liability to be capped at $5 million per
occurrence.
   The record also contains evidence, albeit slightly less one-
sided, regarding the 2015 RLI Policy at issue in this case. In
the lead-up to renewal, RLI prepared a proposal dated De-
cember 4, 2014. The proposal said that the Aggregate Corridor
Deductible was “in Layer above $3,000,000 SIR” and that it
“applies to the $2,000,000 Limit” (emphasis added). These
phrases clearly indicate that RLI and New Prime intended the
ACD to sit within RLI’s policy layer. AIG has not disputed the
authenticity of the December 4 proposal. In an email two
weeks later, on December 18, a Cottingham & Butler broker
asked RLI to “bind” the coverage “with the $2,500,000 corri-
dor per the quote attached.” RLI asserted in the district court
that the referenced “quote” was the December 4 insurance
proposal, and AIG did not dispute that point.


8 In the insurance business, “CSL” is an acronym for “combined single
limit.” E.g., Alshwaiyat v. American Serv. Ins. Co., 

, 185 (Ill.
App. 2013).
22                                                           No. 19-1426

    Nevertheless, AIG argues that RLI’s reply the next day to
Cottingham & Butler creates a material fact dispute to pre-
clude summary judgment. In a one-sentence summary of the
policy, RLI described the indemnity limit as “above” the
ACD, a word choice that conflicts with the earlier insurance
proposal. Not long afterward, however, RLI returned a “Bind-
ing Confirmation” to New Prime that repeated the “in Layer
above” and “applies to” language from the original quote.
This exchange does not create a genuine dispute of material
fact. RLI’s brief reply email was at best ambiguous, and the
more formal policy documents consistently made clear that
ACD payments would erode policy coverage.
    A final, pre-dispute account of how RLI understood the
Policy comes from a December 20, 2014 memo in the under-
writing file. The memo described the terms under which
Swiss Re, a reinsurance company, offered to reinsure the RLI
Policy assuming “a $2.5MM Corridor Deductible within the
$2MM layer above the SIR” (emphasis added). Absent evi-
dence to the contrary, we can assume that RLI was not lying
to Swiss Re, especially given the duty of “utmost good faith”
that governs reinsurance transactions. Amerisure Mut. Ins. Co.
v. Glob. Reinsurance Corp. of Am., 

, 749 (Ill. App.
2010). AIG does not contend otherwise. RLI has consistently
understood, and expressed its understanding, that New
Prime’s ACD payments reduce its own responsibility for
losses, and New Prime did not disagree before this dispute
arose.9


9One unusual feature of this case, as compared to most insurance cover-
age disputes, is that the insured and insurer agree about how to apply the
disputed RLI Policy. The disagreement comes from excess insurers who
were not parties to that policy. The several excess insurance contracts were
No. 19-1426                                                            23

    B. Negotiations Between New Prime and AIG
    The record also contains evidence that AIG knew at the
time it contracted with New Prime that its responsibility for
losses would begin at $5 million per occurrence. On April 2,
2014, Todd Brejcha, a broker with New Prime’s agent Am-
WINS, emailed Lexington to seek a quote for an excess insur-
ance layer of “$5MM xs $5MM P.” An attached “Transporta-
tion Risk Summary” described the “U/L Program” as “$5MM
CSL–RLI subject to a $3MM SIR.” Translating the industry jar-
gon, Brejcha was seeking an offer to provide $5 million of cov-
erage in excess of a $5 million primary policy. The Risk Sum-
mary explained that the underlying insurance program was
RLI’s $5 million combined single limit policy, $3 million of
which belonged to New Prime’s Self-Insured Retention. In
other words, Brejcha told Lexington that New Prime sought
excess coverage to begin at a $5 million per occurrence thresh-
old.
    Two weeks later, Brejcha emailed Lexington again and
sent a copy of the entire 2014 RLI Policy, presumably after
Lexington expressed interest in selling coverage to New
Prime. Brejcha described the RLI Policy as “the main $2MM
xs $3MM primary policy,” thus reiterating that Lexington’s
responsibility would begin at $5 million. Four days later, on
April 21, 2014, an AIG broker replied to Brejcha with a “$5M
xs of $5M Quote” from Lexington. The attached “Quote Con-
firmation” listed the underlying policies as AIG understood
them: a $2 million RLI combined single limit above a


all designed to fit together, however, and there has been no argument here
to the effect that the AIG excess insurers are not entitled to contest the
effect of the RLI Policy.
24                                                 No. 19-1426

$3 million self-insured retention. The Quote Confirmation did
not mention the Aggregate Corridor Deductible, even though
Lexington had received a copy of the RLI Policy and so knew
about the ACD. A week later, on April 28, 2014, Brejcha
emailed the AIG broker to “bind the $5MM xs $5MM per your
attached quote.”
    The only reasonable inference from this back-and-forth is
that AIG did not believe the ACD affected the threshold at
which its layer began—namely $5 million per occurrence.
AIG thus understood that ACD payments eroded the RLI Pol-
icy. In the district court, AIG insisted that “no such inference
exists given the plain language of the only document the
Court can consider absent finding an ambiguity–the RLI Pol-
icy.” We have found ambiguity, however. The extrinsic evi-
dence of dealing between New Prime and AIG shows beyond
reasonable dispute that AIG knew exactly what it bargained
for in April 2014, and that’s exactly where the district court’s
judgment left it.
     C. Later Events
    Although the evidence from before the Herrera and Mon-
tini accidents decides this case in RLI’s favor, we must still
consider whether later events introduce a factual dispute that
precludes summary judgment. RLI can point to many addi-
tional emails among New Prime, its brokers, and RLI from
April 2016 onward that endorsed RLI’s interpretation of the
ACD. Since the evidence from before the Herrera and Montini
accidents suffices on its own, however, we see no reason to
rely on after-the-fact correspondence that could be criticized
as affected by strategic posturing in the dispute then brewing
with AIG. We likewise decline to rely on the affidavits RLI
No. 19-1426                                                  25

submitted to the district court to the extent that they are un-
corroborated by documentary evidence.
    We do need, however, to examine in detail a chain of cor-
respondence from April 24, 2015. At 10:42 AM on that date,
Michelle Mertz, the Risk Manager at New Prime, emailed
Todd Brejcha at the AmWINS brokerage regarding the Mon-
tini claim. Mertz’s initial view on how New Prime’s insurers
should share the losses tracked AIG’s later position in this
lawsuit: “Prime would pay the first 5.5 million ($3m per oc-
currence, $2.5m aggregate corridor) [and] RLI would then
pay their $2m. So, as of now, $7.5m would have to be paid on
this claim before it got to Lexington’s layer.” This email rep-
resents the first and only time either party to the RLI Policy
endorsed AIG’s interpretation of how the ACD functions. No-
tably, Mertz had become New Prime’s Risk Manager in 2013,
so she was not present when the RLI Policy was initially ne-
gotiated.
    Less than half an hour later, Brejcha emailed the other bro-
kers—Tim Larocca at AmWINS and Jack Welbourn at Re-
gions—to express his disagreement with Mertz’s view that
“RLI will participate above the corridor deductible.” Larocca
then asked to see the RLI Policy himself because he recalled
that “this wording is very confusing.” At 11:35 AM, Welbourn
wrote that, after reading the policy, he remained uncertain: “it
really does not spell out the actual limit per claim.” About
forty-five minutes later, though, Welbourn emailed Mertz,
copying the others, to explain that he “thought the limit was
$5,000,000 and not $7,5000,000 [sic],” but that “the wording in
the policy is a bit ambiguous.” He asked Mertz to request clar-
ification from RLI “so in the event of a loss we do not have a
carrier dispute.”
26                                                    No. 19-1426

    But as the need to write this opinion indicates, that ship
had sailed. Mertz had already told Charlie Creamer, a “com-
plex director” for AIG, that Lexington’s responsibility would
not begin until $7.5 million of losses. At 12:33 PM, sixteen
minutes after she received the email from Welbourn, Mertz
emailed Creamer to walk back her earlier position: “My de-
scription of the underlying limits is incorrect. I incorrectly
stated that RLI would pay $2m on this claim over $5.5m. The
RLI policy has a $5m limit so Lexington would have any
amount over $5m.”
    The question is whether Mertz’s temporary view during
the morning of April 24, 2015 creates a genuine dispute as to
a material fact such that a reasonable jury could return a ver-
dict for AIG. See, e.g., Medical Protective Co. of Fort Wayne v.
American Int’l Specialty Lines Ins. Co., 

, 445 (7th Cir.
2018). We think not. Mertz’s confusion, as well as the brokers’
statements that the language in the RLI Policy was difficult to
parse, tends to confirm what we have already held: the Pol-
icy’s language was ambiguous in this respect. Mertz’s errone-
ous statements regarding the legal effect of language she had
not negotiated did not address or rebut the objective evidence
from 2011 to 2014 summarized above.
    To put it another way, Illinois courts clarify ambiguous
contract language based on general circumstances around
contract formation but only actual conduct subsequent to for-
mation. See Szafranski v. Dunston, 

, 1157 (Ill.
App. 2015). The April 24, 2015 statements from Mertz and the
brokers were not conduct. All of New Prime’s concrete ac-
tions, in contrast, were consistent with RLI’s position in this
lawsuit: Although New Prime originally advanced nearly
$10 million in defense costs for the Herrera lawsuit, Mertz
No. 19-1426                                                  27

testified in the district court that it ultimately paid only the
$5 million RLI Policy limit. AIG’s position would have
obliged New Prime to pay $5.5 million of Herrera costs. New
Prime’s conduct therefore supports RLI’s interpretation of the
ACD. The April 24 emails do not present a genuine issue of
material of fact that a jury would need to decide.
    To sum up, the language of the RLI Policy is ambiguous
as applied to this dispute, but the extrinsic evidence compels
summary judgment in favor of RLI. On this ground, the judg-
ment of the district court is
                                                   AFFIRMED.