🕰️ HISTORICAL DOCUMENT — June 2026 Stripe payment analysis. This is kept for the record and for its reasoning, not as a current read. It diagnosed the May → June payment decline across all rails; its conclusion — that the cause was retention, not acquisition — still stands and is why retention is the company's binding constraint. For today's numbers go to the cockpit. Republished under the cockpit domain 11 Aug 2026 so it has one permanent home instead of a file:// path on one Mac.

TMA Payment Drop — Full Diagnosis v3

Stripe payments + Google Play + Meta ads + Stripe Recoveries · Feb–Jun 2026 · for Aga · 1 Jul 2026
The answer. Your Stripe cash fell from ~£1,370/wk to ~£950/wk in June, but the true combined-rail drop is smaller — about £331/week — and ~80% of it is renewals: your active subscriber base is shrinking because churn is outrunning new adds. New customer acquisition held basically flat (£257→£235/wk) once you count Google Play — the Stripe "collapse" was mostly sign-ups moving to Google in-app billing. Recovery did not break, ProfitWell is innocent, checkout & Radar are fine. Two real things to fix: churn (the base) and Meta ad efficiency (cost-per-lead more than doubled in late June).
True combined drop
£331/wk
the real number
…that is renewals
80%
shrinking base
New biz (both rails)
£257→£235
≈ flat
Meta CPL late Jun
£2.57→£6.13
+139%

1 · Stripe weekly take — the drop is real, but Stripe can't see the whole business

17882/213312/98162/1614442/2317833/215223/99813/1610823/2315313/3017704/614504/1310484/2012354/2715325/412975/1121435/1826235/2513756/19506/88046/156286/229966/29
Baseline  May sale  June  — low week £628 (22 Jun). ("£57/wk" earlier = new-sign-ups only, mislabelled — corrected.)

2 · New business didn't collapse — it moved to Google Play

June new-customer cash: Stripe + Google Play per week

baseline £257/wk538Jun 1322Jun 8144Jun 15117Jun 2257Jun 29
Stripe new  Google Play  — dashed = Feb–Apr baseline (£257/wk)

You launched Google in-app billing, so new Android sign-ups now bill through Google (£433 in June), not Stripe. Netting both rails, June new business was £235/wk vs a £257 baseline — essentially flat (−8%). The Stripe-only −42% was ~80% channel shift, not lost revenue. You were right that Google's total looks small — but it almost exactly offsets the Stripe new-business dip.

3 · The real driver: a shrinking renewal base (≈80% of the true drop)

Segment (per week)BaselineJuneChangeVerdict
Renewals£1004£741−£263Real — fewer subs to renew (29→23/wk); success rate held ~60%
New (Stripe+Google)£257£235≈ £-21Held channel shift

The renewal fall is not payments breaking — it's the base eroding: churn (~12%/mo) is running ahead of net new subscribers, so each week there are fewer active subs to bill. That compounds. This is the actual problem to solve.

4 · Recovery did NOT break in June (proven)

Recovery rate by month the invoice first failed

55Aug61Sep40Oct33Nov38Dec64Jan50Feb33Mar60Apr44May24Jun
33–64% every month for a year — noisy, no break. Dashed = 46% completed avg. Faint June bar = 8/17 invoices still mid-retry.

June's low reading is a retry-window artifact — 8 of 17 June failures are still retrying into July. Recovery emails + retries both fired all along; cancelling ProfitWell (23 Jun) left no dent. Chronic truth: only 46% of failed invoices ever recover and ~47% of failures are "insufficient funds" — a steady leak, not a June event.

5 · Meta ads — spend rose, leads fell, cost-per-lead doubled

Meta cost-per-lead by week (June)

£2.74Jun 1-7£2.57Jun 8-14£2.64Jun 15-21£6.12Jun 22-28£5.26Jun 29-30

June Meta: £2074 spend, 602 leads, blended CPL £3.44. But efficiency collapsed late June: you nearly doubled daily spend (to £80–120) while leads fell — CPL jumped from ~£2.57 to £6.13 in the last 8 days. That's creative fatigue / audience saturation. It's why late-June new business softened even with Google counted — and it's burning money at 2.4× the cost.

6 · Cleared (with evidence)

SuspectVerdictEvidence
Broken checkoutNoTrial conversion 40.5%, steady
Radar over-blockingNo0.09% block rate, 1 block, 0% fraud
Cancelling ProfitWellNoRecovery flat before/after 23 Jun; ~£100/mo half-broken tool
"Recovery broke in June"No8/17 June invoices still mid-retry (artifact)
Organic SEO declineMinorOrganic traffic −16% (SEMrush) — a headwind, small slice

7 · Do this (ranked)

  1. Protect & rebuild the renewal base — the ~80% driver. Attack churn: cancellation-save flow, win-back & reactivation (much already built in AC), and re-engagement. A shrinking base is the whole ballgame.
  2. Fix Meta ad efficiency. Refresh creative, check audience/frequency — CPL at £6+ late June means you're paying 2.4× for the same lead. Don't just spend more into fatigue.
  3. Build a combined-rail revenue view (Stripe + Google Play + iOS). Watching Stripe alone caused this scare — it can't see Google.
  4. Lift chronic recovery (~46%): payday-timed Smart Retries + Card Account Updater (free). Not urgent, but real money.
  5. Leave checkout, Radar, ProfitWell alone. Cleared.

8 · Still would sharpen it

Method: Stripe GBP = converted amount (settled) / 0.755×USD (attempted). "Baseline" = weekly avg 2 Feb–27 Apr (excludes May sale). Google Play & Meta assumed June 1–30 as supplied. Recovery "completed" excludes invoices still in retry window. v3 supersedes v1/v2 — adds Google Play + Meta, nets the channel shift. Generated from Stripeunified_payments.csv, Recoveries_2025-08-01_to_2026-06-30.csv, and supplied Google/Meta daily.