Reverse trial in-app A/B — proposal v2 (revision)

Showing no day-0 modal at all outperforms the welcome modal — testing whether msg 1 has any value for users on signup.

50/50 split: 50% of users see the control welcome modal, 50% see nothing on signup. We measure downstream paid conversion per arm. If the no-message arm matches or beats control, msg 1 can be retired.

Applying the winning Sales Outreach voice (no-pressure, honest, "stuck on anything") to the Glen pop-up lifts bookings vs the control's discovery-call framing.

Mirrors the winning Sales Outreach email tone (tpl 342): "Stuck on anything? Or want a hand getting set up? Happy to help — no pressure." Same Calendly destination.

Replacing 'Upgrade' with 'Claim offer' on the day-2 discount banner lifts CTR because 'claim' reframes the action from commercial commitment to capturing a benefit.

Same body copy as control ("Get 50% off for your first 6 months"). Only the CTA verb changes — minimal test surface, clean single-variable comparison.

Same 'Upgrade' → 'Claim offer' CTA test at day 3 (no-reminders branch). Tests whether the lift holds across trial days + targeting branches.

Identical to msg 3 — control body kept, only CTA changes. Replicating the test across day-3 no-reminders branch builds confidence in any observed lift.

Treating a partially-completed task as "one step to finish" outperforms re-explaining the value because the user has already shown intent by viewing the page.

Branch fires only if reminders_page_viewed == true. Control explains what reminders are — but the user is past that point. Variant skips the explanation, asks for the last step.

Third replication of the 'Upgrade' → 'Claim offer' CTA test, at day 4. Triple-replicated test across days 2, 3, 4 = ~3× sample size on the same hypothesis.

Identical structure to msg 3 + 4. Control body kept, CTA flipped. Three placements of the same test give stronger statistical power on the CTA lever.

Reframing day 7 from "what you'll lose" to "what's earned its place" outperforms because the email sequence already handles loss messaging — variant adds a positive in-app counterweight.

Day 7. Headline drops the "Day 7 of 14" formality for "Halfway there". Body asks which features have earned a place — same 4-item list as control, but framed positively. CTA stays "Keep Advanced".

Replacing the control's vague "one of the big unlocks" intro with a concrete outcome (cards charged in-session, settle in 2 days, no chasing) lifts bookings because clinicians convert on specific benefits not category labels.

Eyebrow ($1,000 FREE) + h2 (15 min) + p2 ($1k hook) + CTA stay identical to control. Only p1 changes — from vague "big unlocks" to specific cash-flow benefit. Single-variable test.

Stating time-remaining without loss-listing outperforms because the email sequence already covered the loss; variant adds a calmer in-app counterweight.

Day 12. Loss emails fully landed. "You'll miss out on…" reads heavy by now. Variant just states the timeline + a calm reassurance: workspace stays as-is when you keep Advanced.

A simple time-statement banner ("ends tomorrow") outperforms a discount banner because clinicians prefer a date over a deal at end-of-trial.

Day 13. Email "ends tomorrow" lands today/tomorrow. Variant in-app matches the email's calm date-framing — drops the price overlay, matches the email's reality.

Removing the "remind me tomorrow" snooze + calmer factual framing outperforms because the snooze contradicts the day-14 deadline and the loss-list theatre fatigues by now.

Day 14 — trial actually ends today. Control's "remind me tomorrow" CTA is operationally false (tomorrow = Free). Variant removes the snooze; single calm decision.