Toby — State of Business

Weekly nightly report · 2026-05-18 · data snapshot 2026-05-11 22:32 UTC · sources: prod Postgres (heartbeat / Stripe replica) + BigQuery Amplitude. Note: DB clock lags wall-clock by ~7d; numbers below are most recent snapshot available.

TL;DR

Flat week, every dial still flashing the same warnings. Net MRR $40,356 (−$34 WoW, essentially noise). Paid subs 7,835 (−1). Signup velocity 720/wk, still at −60.8% YoY — the seven-week downtrend is now flat-to-down. Retention save flow remains silent (0 logged offers in the last 30d; only 7 in the 12-week window vs 122 cancels in 30d). Feb-2027 mega-cliff barely moved: 2,354 legacy subs / $12.45K MRR due to renew that one month (−17 subs WoW; cohort is leaking, not collapsing). Trial zombies untouched (8,833, all trial_end > 180d ago).

KPI grid — week of 2026-05-04 → 2026-05-11

Net MRR (active)

$40,356

−$34 WoW · flat

Paid subs (active)

7,835

−1 WoW

Paid seats

8,600

−3 WoW

WAU (heartbeat 7d)

59,642

+6 WoW · flat

MAU (heartbeat 30d)

69,885

+12 WoW · flat

WAU (Amplitude 7d)

60,289

vs DB 59,642 · agrees

MAU (Amplitude 30d)

71,562

vs DB 69,885

Signups (7d)

720

YoY −60.8% (vs 1,836)

Signups (30d)

3,290

~110/day

New active subs (7d)

−2 WoW · weakest week of qtr

New active subs (30d)

unchanged WoW

Cancellations (7d)

+2 WoW

Cancellations (30d)

122

unchanged WoW

Queued to cancel

294

+1 WoW

Trialing (zombies)

8,833

all trial_end > 180d ago

Trialing w/ payment

unchanged WoW · the "real" trials

Total users

1,046,874

+31 WoW

Dormant (last_active >90d)

305,651

+56 WoW

Ghosts (never active)

654,085

−1 WoW · 62% of all users

CWS rating

4.2 ★

stale (fetch blocked); 3,300 ratings last seen

MRR composition

Cohort	Plan	Interval	Subs	Seats	MRR ($)	% of MRR
Legacy ThankYou	team_productivity	year	6,857	7,442	$33,489.00	83.0%
Non-legacy	team_productivity	month	687	728	$4,204.00	10.4%
Legacy ThankYou	team_team	year	67	178	$1,424.00	3.5%
Non-legacy	team_productivity	year	198	212	$954.00	2.4%
Non-legacy	team_team	year	12	25	$150.00	0.4%
Non-legacy	team_team	month	14	15	$135.00	0.3%
Total			7,835	8,600	$40,356.00	100%

Concentration risk: 86.5% of MRR sits in the legacy ThankYou cohort (single coupon nBvJx1dq). Non-legacy + true-pricing revenue is only $5,443, growing at ~$0/wk. Reconciliation note: this week's MRR breakdown re-classifies more revenue as "legacy" than last week's snapshot (34.9K vs 26.9K) — same query, same source; the WoW comparison should be read on the total line ($40,390.5 → $40,356), not the per-cohort split.

Renewal cliff — next 24 months (legacy ThankYou yearly)

Renewal month	Subs	Seats	MRR at risk ($)	Severity
May 2026	153	169	$764.00	routine
Jun 2026	208	218	$1,009.00	routine
Jul 2026	179	185	$832.50	routine
Aug 2026	179	190	$862.00	routine
Sep 2026	146	178	$804.50	routine
Oct 2026	1,063	1,135	$5,209.00	cliff (mid)
Nov 2026	400	427	$1,981.00	elevated
Dec 2026	196	201	$929.00	routine
Jan 2027	1,018	1,084	$4,909.50	cliff (mid)
Feb 2027	2,354	2,700	$12,447.50	MEGA-CLIFF
Mar 2027	542	614	$2,791.00	elevated
Apr 2027	302	315	$1,442.00	routine
May 2027	80	92	$414.00	routine

Feb 2027 is the single largest revenue event in the company's runway. $12,447.50 of MRR (31% of all MRR; 26% of all paid subs) reaches its first renewal that month — and it's the cohort with the highest-probability churn (legacy $4.50/seat/yr → about to be re-priced). WoW: cohort shrank by 17 subs / $81 MRR (early canceller leakage). Net $22.6K MRR / 55% of total reaches a renewal decision in the Oct-26 / Jan-27 / Feb-27 window.

Action lag warning: we are now ~8.5 months from the Feb-27 wave. No offer-ladder UX, no instrumented save flow, no pre-cliff messaging campaign is in production. Q3 2026 is the latest reasonable runway to build that.

Signup trend — last 7 complete weeks

Week starting	Signups	Trend
Mar 23 2026	796
Mar 30 2026	862
Apr 6 2026	1,024
Apr 13 2026	862
Apr 20 2026	769
Apr 27 2026	709
May 4 2026	725

Last 7 days (5/5–5/11): 720 — slight tick down from prior week. YoY same window: 720 vs 1,836 = −60.8%. Acquisition has been below 800/wk for 4 consecutive weeks; the previous 5-month trough (Mar 30 = 862, the Apr 6 = 1,024 spike now looks like an outlier, not a recovery).

Retention save flow — still silent

Window	Offers issued	Accepted	Implied save rate
Last 7 days	0	0	—
Last 30 days	0	0	vs 122 cancellations
Last 12 weeks	7	7	100% accept; cohort too small to read
All-time	16	—	table has ~zero traffic

Either the save-flow UX isn't triggering, the events aren't reaching retention_offers, or both. At 122 cancels / 30d we should be seeing at least tens of save-flow rows weekly. Currently we'd save 0 even if the offer was perfect — there's no surface to accept.

Cohort health roll-up

Slice	Users	Share
Total users	1,046,874	100%
Active in last 7d	59,642	5.7%
Active in last 30d	69,885	6.7%
Active in last 90d	87,138	8.3%
Dormant (active ever, >90d ago)	305,651	29.2%
Ghost (never recorded as active)	654,085	62.5%

The 62.5% ghost share is essentially a proxy — users.last_active only fires from extension heartbeats. Without the BigQuery core_api.events ↔ users join, we can't confirm whether ghosts are truly inactive or whether the heartbeat path simply never fires for some install profiles. Either way, the 6.7% MAU/total ratio is the real funnel signal.

WoW summary vs 2026-05-11 baseline

Metric	2026-05-11	2026-05-18	Δ
Net MRR	$40,390.50	$40,356.00	−$34.50
Paid active subs	7,836	7,835	−1
Paid active seats	8,603	8,600	−3
WAU (DB)	59,636	59,642	+6
MAU (DB)	69,873	69,885	+12
Signups 7d	722	720	−2
Cancellations 7d	32	34	+2
Cancellations 30d	122	122	±0
Queued to cancel	293	294	+1
New active subs 7d	10	8	−2
Trialing zombies	8,833	8,833	±0
Feb-27 cliff MRR	$12,528.50	$12,447.50	−$81 (early leakage)
Total users	1,046,843	1,046,874	+31

Standing blockers — unchanged this week

Stripe Sigma integration still pointed at TEST mode → all sub/MRR numbers come from the Postgres replica, not Stripe live.
8,833 trialing zombies with trial_end 2023–2024; Stripe→Postgres sync should have transitioned these to canceled or active long ago.
retention_offers only logs ~7 rows/12w against 122 cancels/30d — the save flow either isn't triggering or isn't writing rows.
cards table still has no index on created_at, so card-creation activity remains un-reportable at scale.
BigQuery core_api.events ↔ users join not yet wired; ghost-rate is still a static proxy, not a true weekly delta.
data-sources.md not yet committed to axiomzen/toby — every nightly re-discovers schema instead of trusting a single source of truth.

Recommended actions

Resurrect retention save flow this sprint. 122 cancels/30d & 294 queued is plenty of inventory; the dial is at zero offers because there's no surface, not because the offer is bad.
Start the Feb-27 cliff response now. 8.5 months out is barely enough lead time to: (a) design the offer ladder for legacy ThankYou → new pricing, (b) instrument the funnel, (c) A/B the email cadence. If we wait for Q4 2026 we land in production 1 month before the wave.
Diagnose the signup decline. −60.8% YoY for the 4th consecutive week is not seasonality. The "blank extension page / frustrating onboarding" CWS complaints last seen in late April point at a SW boot regression — pull a release-diff between mid-2025 and current to bisect.
Force-finalize the 8,833 zombie trials. Either re-poll Stripe per subscription_id or run a one-shot Stripe Sigma reconciliation to bulk-transition them to canceled — they pollute every funnel calc.
Wire Sigma to LIVE mode on the integration; until then we cannot triangulate Postgres replica against billing source-of-truth for MRR/churn magnitudes.

Sources: prod Postgres be55a66b-c905-4759-9ce1-a97785bb69e6 · BigQuery ec765049-99ac-44e9-8c95-0ac80f284054 (toby_amplitude.EVENTS_306007). Snapshot timestamp 2026-05-11 22:32 UTC. Generated by aios agent toby-state-of-business---nightly-report, run b9bfe5ec-9a61-4206-89bf-3257224a21dc.