I'm working through a legal strategy memo that needs to balance the missing documents issue carefully—avoiding overreach while recommending a practical path forward. The approach combines a motion to defer the certification briefing with a targeted production request under the rules, paired with either a conditional opposition or extension request, rather than pushing for a broad stay or jumping straight to merits briefing. I'm also mapping out the key representatives involved, starting with Milo Chen on the mobile side.

For Chen specifically, I need to assess his standing and typicality claims. He paid a fee and presumably can't get it back, which supports standing, but I need actual evidence of payment. On typicality, he can only represent the mobile subclass—not desktop users with unchecked boxes, legacy email-only customers, or those contacted by phone, since those involve materially different opt-in mechanisms. His memory gaps about whether the box was preselected or whether he clicked the link weaken his personal testimony under the reliance rules, but StreamLedger's own enrollment records may settle the question objectively. If their database shows Chen enrolled through the mobile path and that path involved a uniform representation across all 21,300 mobile customers, then reliance can be inferred from that uniformity rather than depending on his recollection—that's the honest framing here, and it matters for the strength of the claim.

For Tara Vale and the call-center cohort, her situation is much weaker for class certification. She claims an oral misrepresentation ("fee mandatory"), but she can't identify the date, representative, or which of the six scripts she heard, so there's no way to map her individual experience to a specific script or establish what was actually said. Oral misrepresentations are inherently individualized, and if her alleged statement contradicts all the scripts on file, that's a major problem for typicality across the 9,200 call-center customers. Her claim introduces too many individual issues and likely fails the predominance test for any call-center subclass.

The evidentiary gaps cut both ways though—her phone and account records plus StreamLedger's call metadata could narrow the date range, but the sample-only recording retention creates mutual problems. I need to be careful here: arguing that she can't prove which script she heard invites an adverse-inference argument about StreamLedger's own retention policy, so I shouldn't lean too hard on that gap as a clean defense win.

For Jae Morgan, the "CB-R" notation is the key issue. If it means completed chargeback, Morgan has no unreimbursed loss and therefore no standing under the rules. His card and bank statements should resolve this definitively—they'll show whether he had a net $7.95 debit or credit. This is straightforward discovery that can eliminate him cleanly, but I should also cross-check StreamLedger's internal reconciliation with the processor rather than just relying on Morgan's failure to clarify.

Now I'm distinguishing between class-wide facts and individualized ones. The fee amount, whether priority processing actually delivered the claimed weekday improvement, the internal decision to charge, and the uniform refund policies are all common questions provable through common evidence. But exposure to deceptive marketing varies significantly across the four different user paths—some saw an unchecked box with free alternative disclosed, others a preselected box with a link, and the landing page exposure differs too.

Materiality and reliance also split by path since Rule 5 requires a materially uniform representation across the class. For the contract claim, the unchecked box path involves separate knowing selection, the preselected path has no separate selection, the legacy path raises terms incorporation questions, and the call path depends on oral assent records. Whether the free option was disclosed and how prominently it appeared is path-dependent, as is whether the customer filed on a weekday or weekend—which determines whether they actually benefited from the priority processing and affects causation and damages calculations.

On predominance, the consumer statute requires proving exposure, causation, and unreimbursed loss across the class. But StreamLedger's own records show the desktop path expressly discloses the fee and free alternative with an unchecked box, so the "uniformly hidden" allegation fails factually and predominance collapses for the entire class. At the path level, the desktop subclass likely fails on the merits since there's arguably no deception and knowing selection defeats reliance anyway, while the mobile subclass presents the strongest uniformity argument—a single preselected screen with a linked free option raises the core question of whether preselection combined with a link to the free tier constitutes deception.

The legacy subclass also shows relative uniformity since disclosure appears only in linked terms, raising the separate question of whether that incorporation method is adequate. The call subclass fails uniformity entirely given six different scripts plus sample recordings and oral variation. For contract claims, everything depends on which terms were presented and assented to, and this varies by path, plus the terms themselves likely shifted over the 36-month period so I need to verify versions. The breach theory also matters—if the terms disclosed the fee, there's no breach; if the fee was charged without any disclosed term, that's unjust enrichment or breach. Contract certification probably won't work class-wide but might be manageable within a legacy or mobile subclass if there's a single terms version and single interface version, which means I need to pull interface change logs from our own discovery. I also need to reframe the "no benefit" theory since the weekday/weekend split undermines any class-wide no-benefit claim.

The damages model breaks down because a full-refund theory assumes the service was worthless, but weekday users actually got an 18-hour benefit, so full restitution doesn't fit. I need to deduct the 5,400 refunds and 1,180 chargebacks, though I should check for overlap since some customers might have both. The 7,900 employer-card payments create bilateral uncertainty about who actually bore the loss—plaintiffs have to show unreimbursed personal loss, which likely requires individualized inquiry for that 12.7% of the class.

The stronger framing treats employer payment as a standing/injury issue rather than just a damages quantum problem, since unreimbursed economic loss is the key element. This might be handled through a claims process, but it also suggests the class definition is overbroad because it sweeps in people with no injury—those who got refunds or chargebacks. A price-premium model would need to account for the free alternative and fit whatever theory gets certified, but plaintiffs haven't offered one yet.

On the class structure side, there are genuinely common issues worth separating: whether the backend queue actually provided differentiated processing by day of week, the standardized disclosures for each path and whether they were materially deceptive as a matter of law, and whether terms were incorporated by reference in the legacy landing page. A workable subclass split would be desktop, mobile, and legacy—each with its own representative and assent record—though StreamLedger argues no adequate representative exists for desktop or legacy among the current plaintiffs.

The strategic move is to concede that a narrower class limited to queue performance might be certifiable, but flag the manageability problem: it wouldn't resolve liability without individualized findings on exposure, assent, and loss, so it has limited practical value. This gives the court an off-ramp we can accept rather than fighting everything.

Now I'm looking at which named-plaintiff materials actually matter for each element—confirmation emails, enrollment records, and screenshots showing which enrollment path each plaintiff took. Refund and chargeback records show whether the dispute was resolved and if the customer accepted a settlement. Employer reimbursement documentation establishes who ultimately bore the cost, which affects both injury status and class membership. Communications around enrollment—emails, chats, call recordings—help identify which script was used and whether the fee was presented as optional or mandatory. Usage patterns tied to weekday versus weekend activity determine whether they received the 18-hour benefit, which connects to causation and damages calculations. These documents may not always be conclusive, so I need to be careful about what they actually demonstrate.

The real leverage here is that certification under Rule 23 requires evidence, and the defendant controls these records. The argument isn't about predicting the outcome—it's about showing that these categories are decisionally significant under Rules 7 and 8. My strategy is to file a targeted motion to defer certification briefing for 45 to 60 days while compelling production of these seven categories, with a chart mapping each document type to the specific rule requirements. I'll file that motion within about a week, but I'm also preparing a full opposition brief due on day 21 as a backup—no point gambling on the deferral being granted.

On the denial question, there's a real tension I need to resolve: I can argue the record is adequate to show plaintiffs' failure to meet their own burden (which supports immediate denial without prejudice), but simultaneously argue that targeted discovery is necessary if the court wants to give them a chance to cure. The key is framing it as plaintiffs' failure of proof on the present record—their burden, their problem—while the discovery is conditional, not a prerequisite to denial.

For the motion strategy itself, I should lead with denial without prejudice for failure of proof, then offer in the alternative that if the court is inclined toward a narrower subclass or wants to let them supplement, we'd need discovery on the representatives' individual standing before any certification moves forward. That's coherent and doesn't undercut the primary position.

I also need to avoid asking for a broad six-month stay—Rule 8 won't tolerate unfocused delay, and pushing for that could weaken the credible narrow request. Keep it tight at 45 to 60 days. And I shouldn't try to brief the merits now; merits arguments only matter insofar as they affect certification.

On standing, I'll raise Morgan's standing issues in the opposition itself and reserve a Rule 12(b)(1) challenge after we get the records. For the June 1 agreement breach, I'll make a modest sanctions request—fee-shifting or an order precluding reliance on undisclosed documents and declarations, plus barring new plaintiff declarations that contradict what wasn't disclosed. And I need to connect each variation they claim to a specific certification element using a matrix, not just say "every customer is different."

For our opposition, I'm building an affirmative record with declarations showing all four paths by version, click-logs, terms versions, refund and chargeback reconciliation data, and queue performance broken down by weekday versus weekend—not just averages, since the distribution reveals that some weekday users got no benefit at all. This cuts both ways: it undermines their uniform-benefit theory but also shows individualized causation, which I'll frame as evidence that impact varied by user.

On class definition, the 5,400 refunded members and 1,180 chargebacks are uninjured and create a rule 4 problem on loss element and ascertainable injury; plus the class includes employer payers who weren't directly charged. For numerosity and commonality, I'm conceding the 62,400 class size and at least one common question—queue performance and whether the fee page existed—to build credibility, then focusing the real fight on typicality, adequacy, predominance, and superiority. On superiority, the total exposure is roughly $496,000 across 62,400 people, which individually would be small claims, so superiority normally favors certification—but I need to argue it fails here because subdivided proceedings and individualized inquiries into assent and loss make it unmanageable compared to a refund or claims program alternative.

StreamLedger already refunded 5,400 people, so I can point to an administrative refund program as a feasible alternative under rule 2, which undercuts superiority. I'll pair that with a narrow subclass fallback since courts often find class superior for small claims anyway. For the 21-day sequence, I'm laying out concrete steps: days 1–2 involve internal preservation and reconciliation, retaining the four interface versions, pulling Morgan's ledger and processor records, assembling declarations, and sending a meet-and-confer letter demanding production within 5 days with a short opposition extension offered in exchange. Day 3 is sending a rule 37-style deficiency letter identifying the seven categories of missing documents.

Days 4–6 involve a confer call, and if there's no production by day 7, I file a motion to defer certification briefing and compel targeted production with a chart, counsel declaration, proposed 45-day schedule, and expedited hearing request, plus a narrow alternative allowing 14 days for production. Days 7–14 I'm building the opposition record regardless: engineering declaration on queue data patterns, product declaration with archived screenshots per path and version, payments declaration reconciling refunds and chargebacks, call-center declaration on the six scripts and retention practices, and custodian declaration on Morgan's ambiguity and processor records. Days 14–18 I draft the opposition and potentially depose the named plaintiffs.

Chen, Vale, and Morgan have already testified in part, so I'll seek limited further depositions after document production and reserve that option. By day 19–21 I file the opposition preserving the deferral request, or if the court granted deferral, I file a notice instead. If the court denies deferral, the opposition must stand alone without contingencies. I need to include Rule 23 evidentiary objections to their marketing page