Practical Tips · Case Study · June 2026
Insurers wave Shirley v. Allstate as a privacy wall. It isn’t. It is a lesson in asking for the wrong thing — the names, not the files — and losing the whole case a month later.
With Tilem v. Travelers, Shirley v. Allstate is one of the two decisions insurers cite most to shut down discovery of other insureds’ files and identities. Read closely, neither is the wall it is made out to be. Shirley is worth studying because it failed for a specific, avoidable reason — and because the two orders that make up the case, a month apart, show exactly how a discovery misstep becomes a lost case.
After the 2017 Lilac Fire, a San Diego couple filed a smoke-damage claim with their homeowner’s carrier. The insurer denied it on inspection findings of no covered damage, and the couple sued for bad faith, alleging the investigation had been narrowed on purpose — tested only for soot, ash, and char — to manufacture the denial. To prove the narrowing was a practice and not a one-off, they needed to see how the insurer had handled the other Lilac Fire claims. So they served two interrogatories asking the insurer to identify — by name and address — every insured (about seventy-two) who had filed a Lilac Fire claim, so they could send each a court-approved letter seeking consent to look at the file. On July 16, 2019, the magistrate denied the request as a “quintessential fishing expedition.” Shirley v. Allstate Ins. Co., No. 18CV994 AJB (BGS), 2019 WL 3208000 (S.D. Cal. July 16, 2019).
Shirley did not hold that the files are off-limits. It held that this plaintiff asked for the wrong thing — the names instead of the files — and asked for them first.
This is the part insurers leave out. The court ran California’s three-part privacy test under Hill v. National Collegiate Athletic Assn. and found the threatened invasion was not “serious” — privacy, on its own terms, did not bar the request. The court denied it anyway, on relevance and proportionality: the wildfire alone was too thin a thread to connect seventy-two strangers’ files to this couple’s claim, and the bad-faith practices the plaintiffs pleaded were so general they “could apply to every claim ever handled by Allstate.” The privacy law was a sideshow; the request was lost on how it was framed.
| What sank the request | What it actually shows |
|---|---|
| Names, not files. The plaintiffs asked for a roster of seventy-two identities as the predicate to a consent-letter campaign — never, directly, for the redacted files themselves. | Even total victory would have produced a list of strangers, not the pattern evidence, and not before summary judgment was ripe. The files — sought in redacted form — were the thing they needed and never requested. |
| A self-imposed opt-in. The plaintiffs affirmatively asked the court to approve an affirmative-consent (opt-in) letter. | That choice was theirs, not the court’s. It advertised a contact-and-solicitation purpose that colored the relevance ruling — and was unnecessary, because redacted files require no contact at all. |
| The practice pleaded too generally. The court found the listed bad-faith practices “so vague and broad they could apply to every claim ever handled by Allstate.” | Yet the case had a concrete, nameable practice — the vendor tested only for soot, ash, and char. A request keyed to that protocol supplies the similarity nexus the court found missing. They had the nexus; they did not plead to it. |
| No procedural protection. The pattern discovery was still pending when the insurer’s summary-judgment motion ripened, and nothing in the record shows a request to defer the merits ruling until that discovery was decided. | Rule 56(d) exists for exactly this situation. Without it, a discovery loss becomes a merits loss: judgment enters on a record the discovery ruling had emptied. |
Twenty-seven days later, a different judge granted the insurer summary judgment on everything. Four expert reports had found no contamination, so under the genuine-dispute doctrine the denial was reasonable as a matter of law, and the couple had “no evidence” of a biased investigation — only, in the court’s words, “a lot of rhetorical questions” with “no answers.” The pattern evidence that would have answered those questions was exactly what the July discovery ruling denied them. The Ninth Circuit affirmed, and the judgment became final — but in an unpublished, non-precedential disposition that affirmed a summary judgment on an empty record, not a rule that other-insured discovery must come out this way.
Read the deep-dive → The checklist → The bias-evaluation service →
See the issue this case turns on in The One Discovery You Should Almost Never Ask For, the six moves that keep a request clean in The Discovery to Avoid Asking For, and the other most-cited denial in The Case Insurers Cite Most to Deny Discovery.
Distilled from the project’s own analysis and case-wiki entry for Shirley v. Allstate Ins. Co., No. 18CV994 AJB (BGS), 2019 WL 3208000 (S.D. Cal. July 16, 2019) (discovery order), and 392 F. Supp. 3d 1185 (S.D. Cal. Aug. 12, 2019) (summary judgment), aff’d No. 19-56066 (9th Cir. Oct. 9, 2020) (mem.). The seminal authority is Colonial Life & Accident Ins. Co. v. Superior Court, 31 Cal.3d 785 (1982); the privacy framework is Hill v. National Collegiate Athletic Assn., 7 Cal.4th 1 (1994). The statutory-exception authority is reserved for the subscriber deep-dive. Educational and informational only; not legal advice.