The Discovery Objections Series · Part 4 · Case Study · July 2026
The claimants had won the argument that pattern discovery survives into bad faith. Then they asked the court’s permission to send ten thousand consent letters — conceding nothing, protecting no one, and demanding the one thing their case never needed.
Part 4’s case study is Dobro v. Allstate Insurance Co., No. 16cv1197-AJB (BLM), 2016 WL 4595149 (S.D. Cal. Sept. 2, 2016), and it earns its place as the negative image of the concession package. Where Part 4 teaches that the privacy objection is beaten by the requesting party who protects privacy more carefully than the insurer does, the Dobro plaintiffs did the reverse at every turn. They conceded nothing. They offered no protective order to speak of. They demanded the identities of more than ten thousand nonparty insureds — the single most sensitive thing in the case — for a theory that never needed a single name. And they lost every request. Read as a checklist of concessions not made, Dobro is the most useful cautionary tale in the series.
A San Diego couple’s roof was mid-reconstruction when a November 2015 rainstorm arrived; the roofer had failed to tarp the incomplete sections, and water poured through. Allstate denied the claim in eighteen days, citing the policy’s “Planning, Construction or Maintenance” exclusion for faulty workmanship. The denial had a problem the plaintiffs’ lawyers understood well: the Ninth Circuit had held, a quarter century earlier and against Allstate itself, that under California law the ambiguous term “faulty workmanship” in this posture means a flawed product, not a flawed process — so a roofer’s failure to protect the premises did not exclude the resulting rain damage. The bad-faith theory followed naturally: this was not one mistaken denial but a practice of misreading the same exclusion the same way.
To prove the pattern, the plaintiffs served interrogatories seeking the names and addresses of other insureds — every Allstate customer since 1991 whose rain claim had been denied under the same exclusion, plus variations keyed to the couple’s adjuster. In the meet-and-confer they narrowed sensibly: claims after 2010, six Southern California counties, roof leaks, mid-repair, contractor-negligence denials. Allstate searched what it said it could — the first three limitations only — and reported 10,082 claims. The plaintiffs then moved to compel, in a motion captioned — literally — “Motion to Compel Defendant’s Responses to Colonial Life Discovery.” The plan: send a court-approved consent letter to all 10,082 insureds, and review the files of whoever wrote back.
They lost every request.
As in Part 3’s case study, the claimants won the argument the insurer usually makes everything turn on. Allstate argued that Colonial Life — the 1982 California Supreme Court decision blessing pattern discovery of other insureds’ claims — died with the private statutory cause of action it arose under. The court rejected that gambit and held, on the federal line of authority, that “the reasoning in Colonial Life applies whether bad faith is alleged under a statute or under a common law theory.” Pattern discovery of similar claims, properly limited, is relevant to bad faith and punitive damages. The court even mapped the boundary: limited by all six meet-and-confer factors, the interrogatory “may appropriately seek relevant information and may be proportional to the needs of the case.”
Preserve that. A federal court, on a contested motion, holding that similar-claims discovery survives into common-law bad faith — with a blueprint for the scope that would pass. If the order ended there, Dobro would be a claimant’s citation. It did not end there.
| The names | The theory needed patterns, not people. A practice of misreading an exclusion is proved by denial letters and claim dispositions — what was claimed, what was denied, on what stated ground. Identities are machinery for a consent procedure, not evidence. Yet the plaintiffs sought “contact information” for 10,082 households and put the most sensitive item in the case at the center of their motion. Conceding in the requests that no names were sought would have cost the theory nothing and stripped the objection of its only sympathetic fact. Instead they handed the court a request that looked, from a distance, less like discovery than like a lever: settle now, or answer to ten thousand strangers. The court was “not willing to allow Plaintiffs to contact more than 10,000 third-party insureds when the vast majority of the claims made by those insureds are irrelevant.” |
|---|---|
| The protective order and its terms | The plaintiffs reached for the most restrictive privacy mechanism on the shelf — opt-in consent letters — instead of the least restrictive one that would have carried the evidence: a stipulated protective order under which the denial letters and claim data are produced with the court’s own order supplying the protection. An order loaded with the two terms that matter — nonparty identifiers redacted in any filing rather than sealed, and no contact with insureds absent independent discovery or a later order — would have answered the solicitation fear that sank the motion, because there would have been no contact and no ten-thousand-letter mailing to fear. They proposed contact as the mechanism, so the court evaluated contact, and contact failed on scale. |
| That the statute answers itself | Insurance Code section 791.13 prohibits disclosure of insureds’ personal information — and then supplies its own exceptions, including one for disclosures made in response to a facially valid judicial order or subpoena. Nine years before Dobro, a court in the same district had found “no support” for the position that the statute bars court-ordered production; the year before, another had walked through the subpoena and permitted-by-law exceptions in detail. But the Dobro order quotes the statute “in relevant part” — the prohibition and the written-authorization exception only — and concludes that discovery “is conditioned upon obtaining the written consent of the third-party insureds.” The judicial-order exception appears nowhere. The claimants were in no position to point this out, because their own motion assumed the same consent-only world — they had asked to send letters, so the court evaluated letters. |
| That production under the order answers the burden | Allstate’s claims manager declared that its systems could not search for the last three limitations, so isolating the relevant claims would require reviewing all 10,082 by hand. The court relied on the declaration, and it went unrebutted: “Plaintiffs do not submit any information contradicting Defendant’s evidence.” No methodology deposition, no demand for denial-code and cause-of-loss field inventories, and no self-induced-burden argument — even as the court quoted, approvingly, an older case’s observation that there is “no business reason” for an insurer to build systems accommodating unknown future plaintiffs. Part 3 supplied the answer; on this record, nobody gave it. And the whole burden fight was one the concession package would have narrowed: a bounded, unredacted production of denial letters and claim data under a protective order does not require the mass hand-review the consent-letter theory invited. |
Two features keep Dobro instructive rather than merely fatal. The denial was without prejudice. And the court volunteered the next move: the plaintiffs “are entitled to conduct discovery to investigate the statements contained in Mr. Seaman’s declaration and/or to determine the capabilities of Defendant’s computer system” — depose the declarant, examine the databases, find the search terms that isolate the relevant claims, come back. That is Part 3’s methodology deposition, prescribed by the court that just accepted the declaration it would test.
Stack the concessions and the motion transforms. Concede the names: state in the requests that no identities are sought, only patterns. Offer the protective order first, with its two terms — identifiers redacted in filings rather than sealed, no contact with insureds absent independent discovery or a later order — so the solicitation fear that sank the motion never arises. Take the documents unredacted under that order: the denial letters and claim dispositions for the bounded set, produced pursuant to court order under the statute’s own judicial-order exception, with the whole statute quoted so the exception the order omitted is on the page. Propose the sample the court can say yes to: the most recent fifty (or hundred) denials under the exclusion within the six factors, rather than ten thousand households. Test the declaration before the motion, per the roadmap the court ended up writing anyway. And hold the consent letter in reserve for the one thing it is actually for — after the production reveals which handful of claims warrant follow-up, a targeted, court-authorized letter to those few insureds, on the crystallized grounds the case law recognizes.
The claimants’ instinct — Colonial Life blesses pattern discovery — was right, and the court agreed. What failed was the absence of concessions: a consent procedure where a protective order belonged, a demand for names where a disclaimer belonged, aimed at a population no one had bounded, defended against a statute no one quoted in full.
Concede the names before the insurer makes them the issue — if the pattern proves up without identities, say so in the requests and the privacy objection loses its only sympathetic fact. Offer the protective order first, loaded with the public-filing and anti-contact terms, so the solicitation fear has no target. Quote the whole statute — the omission tell can come from the bench, and only the full text answers it. Take production unredacted under the order rather than fighting a redaction burden you can concede away. And read a denial without prejudice plus a judicial roadmap for what it is — an invitation.
Free edition · The story
The full narrative: the pattern theory the court accepted, the relevance ruling in the claimants’ favor, the four concessions they never made, and the rewrite that turns every request into one a court could grant.
Paid edition · Ground by ground
The practitioner’s dissection: the requests verbatim, each objection and its disposition, the statutory misreading against the text and the authorities that refute it, the Sakata consent-mechanics sidebar, and the full concession-package rewrite.
← Part 4 overview The checklist → The bias-evaluation service →
The doctrine this case tests is laid out in Defeating Privacy and Privilege Objections Through Strategic Concessions; the Part 3 counterpart on the burden declaration it never tested is The Claimant Won the Argument, then Lost the Motion; and the discipline for the names it should never have sought is in The One Discovery You Should Almost Never Ask For.
Distilled from the project’s own reading of Dobro v. Allstate Ins. Co., No. 16cv1197-AJB (BLM), 2016 WL 4595149 (S.D. Cal. Sept. 2, 2016), and California Insurance Code section 791.13 and its exceptions. Source text in sources_md_conversions/cases/; case-wiki entry in wiki/cases/. The ground-by-ground rewrite is reserved for the subscriber edition. Educational and informational only; not legal advice.