A Northern District of California court agreed that transaction-level data the FTC wanted from Uber was relevant, then denied the motion to compel because it came six weeks before the close of fact discovery and the burden of producing billions of transactions in that window was no longer proportional. On the same day, the same court found those same six weeks long enough for five more depositions. The lesson: when timing is the issue, the dated record you build and the size of what you ask for decide the outcome as much as relevance does. Reading time approximately 11 minutes.
Meet and Confer Podcast · Kelly Twigger · 🎧 Listen · 📄 Case in Minerva26: FTC v. Uber Technologies, Inc., · No. 4:25-cv-03477-JST (TSH) · Decision: Discovery Order re: ECF No. 275 (ECF No. 281) — motion to compel RFP 61 denied; administrative motion to seal (ECF No. 276) granted · Date: October 15, 2026 · Judge: United States Magistrate Judge Thomas S. Hixson · Court: United States District Court, Northern District of California
Read the full order on Minerva26 →
Listen on Meet and Confer → | Read the transcript →
By Kelly Twigger
Podcast | Transcript
Why Timing Keeps Deciding Discovery Disputes
Timing has been one of the core themes of Case of the Week for a long time, and the short version is to start early. Litigators know there are always a million things to do in a case, and it’s next to impossible to follow up on every issue as quickly as you need to. That includes analyzing the data you get, understanding what your expert is telling you and fitting it into the bigger picture, and responding to a letter from opposing counsel. It’s an incredibly difficult job, and courts are making it harder by allowing for no leeway within fact discovery cutoffs.
Here is a situation every litigator eventually faces. Late in discovery, you find something in the other side’s production that changes what you need, and you learn that the data that would prove your point exists. The court agrees the data is relevant, but you still don’t get it, because of when and how you asked.
That is what happened in FTC v. Uber, a case the Federal Trade Commission brought against Uber, which 21 states and the District of Columbia later joined. The Court agreed that the transaction data the FTC wanted was relevant. It then denied the motion to compel because the motion came six weeks before the close of fact discovery, and producing billions of transactions in that window was no longer proportional. On the same day, the same court found those same six weeks long enough for five more depositions. Proportionality is not only about how important the case is. It is also about whether the work can actually get done in the time that is left.
That adds something new to the timing theme. You can be inside the deadlines and have relevant evidence, and still lose because the burden of what you asked for depends on how much time is left to do it. And while this is high-stakes government enforcement litigation, the lesson applies just as much to smaller cases where the budget is tight and every request has to count.
Case of the Week
What The Case Is About
The FTC sued Uber in April 2025 over Uber One, Uber’s subscription program. The plaintiffs allege that Uber enrolled people without their consent, made it hard to cancel, and misrepresented how much members save, in violation of federal law, including the Restore Online Shoppers’ Confidence Act (ROSCA), and state consumer protection laws. For much of this period, Uber advertised that members saved about $25 a month on average, so whether members actually saved that money is central to the case.
Request for Production 61 asked for transaction-level data on a 10% random sample of all U.S. consumers who used Uber between January 2021 and August 2026, members and non-members alike: the time of each ride or Eats order down to the second, pickup and drop-off coordinates, distance, surge pricing, every component of the price, and any Uber One discount, plus a data dictionary and proof the sample was random. The goal was a difference-in-differences analysis comparing what members actually paid against comparable non-members. That is a legitimate way to test a savings claim, and it requires exactly this kind of data.
The Timeline in Six Dates
The full timeline is in the episode and the transcript. These are the dates that matter for the lesson.
August 2025. District Judge Jon Tigar’s scheduling order tells the parties to take all necessary steps to conduct and compel discovery so they can finish on time. Fact discovery is set to close in June 2026.
March 2026. After a hearing on the FTC’s Request 33, Magistrate Judge Hixson denies the general pricing parts but compels the parts about how Uber One membership affects price. The FTC asked about pricing early, and the Court narrowed it.
Mid-June to mid-July 2026. According to the FTC, Uber produces its first custodial document in mid-June and finishes in mid-July, when fact discovery is set to close in about two months. The cutoff later moves to November 16 on the plaintiffs’ unopposed motion.
August 24, 2026. Uber’s interrogatory answers explain that until July 2025, its advertised savings figure included amounts saved from promotions, and that it then changed the method to exclude promotions available to all users. My read is that the promotions data is the kind of field that would have been part of the original pricing ask. The FTC also relies on a sealed internal Uber document it says shows members paid higher fees than non-members, which Uber says the FTC misread.
September 23, 2026. Uber’s written response refuses to produce anything for Request 61. Around the same time, Uber refuses to designate 30(b)(6) witnesses on how it prices for members and how it prepared the databases it produced.
October 2, 2026. Nine days after Uber’s refusal, with six weeks left in fact discovery, the parties file their joint letter and the FTC asks for production within two weeks. According to Uber, the FTC’s September 29 draft of that letter was the first time it raised the internal fee document with Uber.
A Page and a Quarter for the Hardest Issue
Magistrate Judge Hixson resolves discovery disputes on joint letter briefs. His standing order sets a five-page limit and limits attachments to things like discovery requests and responses, deposition testimony, and meet and confer correspondence. This letter ran about 15 pages and covered five categories of disputes, several of them already agreed or close to it. Request 61, the most complex issue in the letter, got about a page and a quarter from each side. In that space, the FTC had to explain relevance, why the request differed from what the Court had already limited in Request 33, why the burden was reasonable, and why it was asking now. Uber only had to raise doubt.
This judge gives parties room when they ask for it. In this case alone, he suspended page limits for a March joint letter and an April status report, and in Kadrey v. Meta he approved a single 30-page combined letter in place of five separate ones. He also held ten discovery hearings in this case, including more than an hour on the Request 33 dispute. Request 61 was decided on the papers, with no hearing.
The Ruling: Proportionality Is About Time, Not Just Size
Uber told the Court that Request 61 covered billions of transactions, each with dozens of data points, and that even the FTC’s 10% sample would still run into the billions. Uber didn’t support that estimate with a declaration, and the Court didn’t require one. It found the estimate likely correct just from reading the request and the level of detail it asked for.
The Court did not deny the request because the data was irrelevant or because the case didn’t justify it. It found the data relevant, and useful for answering the central question of whether Uber One members actually save money. It denied the request because of when it came. The Court said that if the FTC had moved earlier in fact discovery, it might have granted the motion, but that pulling this much data together in the final weeks was a burden that could easily have been avoided. That burden was not proportional to the needs of the case, which is the standard under Rule 26(b)(1).
That is the crux of this case. Proportionality is usually argued as a question of size: how much data, how much it costs, and how important the issues are. Here, the deciding factor was whether the producing party could actually deliver the discovery in the time that remained. The same-day deposition ruling makes the point. Five more depositions fit inside six weeks, but billions of transactions did not. When you ask for structured data at this scale, the clock is part of the burden calculation, and the later you ask, the heavier that burden becomes.
What the Order Doesn’t Say
The Court measured the delay from when the motion was filed. But the FTC filed nine days after Uber refused in writing, so the only way to file earlier was to ask earlier, and the order doesn’t say when that should have been. It also doesn’t address the FTC’s argument that the request followed up on newly discovered evidence, or the FTC’s account of how late Uber’s documents arrived. This is a complex case with hundreds of docket entries, and there may be factors the order doesn’t mention. Another judge might weigh a motion filed inside the discovery period differently, but you have to plan as if yours will see it the way Magistrate Judge Hixson did.
So the real question is not whether the FTC moved quickly once Uber said no. It is what it takes to get a court to see the whole timeline, and to say yes, when timing is the issue.
Where This Sits in the Timing Line
In Staggers v. Medtronic, the Court held the defendant to notice from its 2016 requests for production because the same request that gave notice in 2022 also gave notice in 2016. In Schulte v. LinkedIn, also from the Northern District of California, the plaintiffs lost a challenge to search strings because they had no factual showing of deficiency and didn’t raise concerns when the strings were disclosed. All three cases turn on when a party acted relative to when it knew what it needed. The difference is the mechanism: Staggers fixed notice from the requests themselves, Schulte failed for lack of facts, and FTC v. Uber won relevance but lost proportionality because burden was measured against the time left.
What to Do This Week
Put the dated timeline where the judge decides the issue. When you are asking late because the other side produced late, the court has to see that, with dates, in the argument on that request. Here, the pieces were in the record, with Uber’s August 24 answers as an exhibit and the late custodial production in a footnote in the deposition section, but they were not pulled together in the Request 61 argument. Five dated lines would do it: when production finished, when you learned the new fact, when you served the request, when the other side refused, and when you filed. Document, document, document.
Make the ask fit the time left. Late evidence can justify a late request, but the court has to be able to grant it in the time remaining. Consider a phased proposal: a small pilot sample, say 1%, over a narrower window, with only the fields that drive price and the data dictionary, followed by your expert’s report to the court within a set number of days on whether more is needed. That gives the judge a decision point instead of an all-or-nothing choice, and because this Court had already found the data relevant, it would have given the Court a way to say yes. Pair it with a narrow request to let that one production happen after the cutoff, and show good cause under Rule 16(b)(4), which turns on your diligence.
Treat burden as a fact, and get the facts. Neither side put facts about the extraction itself in front of the Court. In my experience, data like this at large technology companies lives in platforms built to be queried, and drawing a sample can be simple or very complex; complexity rises when you have to cross-reference databases, such as matching each transaction to membership status on that date. The FTC’s 30(b)(6) notice already had a topic on how Uber prepared its databases, but that topic was not one of the disputes taken to the Court. Done early enough, a short 30(b)(6) deposition on how the pricing data is stored and what it takes to pull a sample would have put sworn testimony in the record, and deposition testimony is something this judge lets you attach to a joint letter. If one of the FTC’s remaining depositions provides that information, it may be enough to renew the motion. To know what you want, you have to know the technology you want the data from.
Raise new evidence in writing as soon as you find it, even if the other side won’t engage. Uber used the timing of the FTC’s first mention of the fee document to call the request a pretext. Put it in writing as soon as you find it, with the Bates number. If opposing counsel won’t engage, delays, or refuses to explain the document, that letter is for the judge as much as for them, and meet and confer correspondence is something this judge lets you attach. I recommend sending that letter within a week of finding the document and serving the follow-up request within two weeks. Use it to lay out the full timeline so you can refer to it in the limited space of the letter brief, staying within what your judge allows to be attached.
In letter-brief courts, treat every move as critical. Know your judge’s letter-brief rules before the dispute exists, because they tell you what record you need to build. Ask for a dedicated letter or more briefing on the issue that matters most, and ask for a hearing, where you can walk the court through the timeline you couldn’t fit on the page. Triage your issues: put agreed items in a stipulation and spend your pages on the fight that moves the case. A simple issue, like the depositions here, can win in a short letter, but a complex one needs room to tell the story.
Start early, and re-plan the day the court narrows you. When the Court cut back Request 33 in March, that was the moment to plan the next request with the expert: what data the analysis actually needs, how to tie it to Uber One rather than pricing in general, and a sample request served while there was still time to produce it. Structured data takes longer to get than almost anything else in discovery, so it belongs at the front of your plan.
If you are the responding party, don’t count on the request proving your burden for you. Here, the Court found Uber’s estimate of billions of transactions likely correct from the face of a very broad request. Against a narrow, phased request, you will need a declaration from someone who knows the systems, and it has to match any testimony the other side has already taken.
What to Watch Next
On October 16, the day after the ruling, the FTC filed a motion to extend the time to complete discovery and the case schedule. The FTC could also ask District Judge Tigar to review the order under Rule 72(a), which allows 14 days after service, so about October 29. In the Northern District of California, that is a five-page motion for relief under Civil Local Rule 72-2, reviewed for whether the order is clearly erroneous or contrary to law, generally on the record the magistrate judge had. As of October 20, no objection had been filed. My read is that a new, phased request built on new facts may be a more practical path than review, because it is a different and much smaller ask, and more time would make it more realistic. I am also watching how other courts treat a request that was late because the other side’s production was late. This order did not reach that question, and I expect we will see courts address it squarely.
Listen to the full episode
This week’s Case of the Week segment of the Meet and Confer podcast walks through the full timeline behind Request 61 in FTC v. Uber, the letter-brief constraints that shaped the fight, the Court’s proportionality ruling, and the practical moves that can get a court to say yes to structured data when timing is the issue. Listen on Meet and Confer →
See Minerva26 in action
Minerva26 is the discovery intelligence platform that connects case law, rules, and real-world workflows. We curate and tag close to 50,000 discovery decisions by issue, including proportionality and sampling, so when you are sizing a structured data request or building the record for a late motion, the rulings you need are already organized for you. Book a 30-minute demo
Related on Minerva26: U.S. ex rel. Staggers v. Medtronic · Schulte v. LinkedIn · Kadrey v. Meta
This decision is available on the Minerva26 platform with full issue tagging. If you’re a litigator navigating discovery strategy and want to stay ahead of decisions like this one, visit Minerva26.com to learn more or schedule a demo. Every decision covered on Case of the Week is searchable by issue, jurisdiction, and judge.
Kelly Twigger is CEO and founder of Minerva26 and Principal at ESI Attorneys. She has been a discovery strategist and practicing attorney for nearly 30 years. Case of the Week is a segment of the Meet and Confer podcast, breaking down one recent ESI discovery decision each week into practical strategy you can use.
- 📩 Get the newsletter
- 🎧 Listen to Meet and Confer
- 🔁 Share this post on LinkedIn
