Email Verification Rollout
Updated 2026-08-02 18:37 UTC · window: last 35 days · source: prod follower (read-only) · regenerates every 30 min
prod: unknown staging: unknown
Observe is live. 390 accounts currently pending, 9 open address proposals (both within the 35-day window).
537
manual signups / day (7d avg)
75.3%
verify-through <24h (7d)
24.4%
confirmed delivered (35d)
1.4%
bounced — watch for reputation (35d)
Daily signups & verification funnel
| Day | All signups | Manual (email) | Verif. required |
Verified (rate) | <24h rate | Manual trials ≤7d (rate) |
Trials: verified / unverified | All trials ≤7d |
| 2026-08-02 * |
1781 |
425 |
17 |
12 (70.6%) |
70.6% |
29 (6.8%) |
1 / 0 |
197 |
| 2026-08-01 |
2300 |
415 |
437 |
358 (81.9%) |
81.9% |
45 (10.8%) |
45 / 0 |
285 |
| 2026-07-31 |
2603 |
517 |
531 |
405 (76.3%) |
76.1% |
46 (8.9%) |
48 / 0 |
323 |
| 2026-07-30 |
2712 |
534 |
553 |
389 (70.3%) |
69.8% |
40 (7.5%) |
44 / 0 |
350 |
| 2026-07-29 |
2436 |
495 |
57 |
41 (71.9%) |
70.2% |
45 (9.1%) |
8 / 0 |
313 |
| 2026-07-28 |
2570 |
526 |
· |
· |
· |
46 (8.7%) |
· |
329 |
| 2026-07-27 |
2833 |
623 |
· |
· |
· |
41 (6.6%) |
· |
329 |
| 2026-07-26 |
2823 |
647 |
· |
· |
· |
52 (8.0%) |
· |
349 |
| 2026-07-25 |
2844 |
619 |
· |
· |
· |
56 (9.0%) |
· |
381 |
| 2026-07-24 |
3231 |
764 |
· |
· |
· |
68 (8.9%) |
· |
396 |
| 2026-07-23 |
3174 |
615 |
· |
· |
· |
56 (9.1%) |
· |
375 |
| 2026-07-22 |
3544 |
634 |
· |
· |
· |
72 (11.4%) |
· |
436 |
| 2026-07-21 |
2878 |
464 |
· |
· |
· |
54 (11.6%) |
· |
379 |
| 2026-07-20 |
2657 |
442 |
· |
· |
· |
50 (11.3%) |
· |
316 |
| 2026-07-19 |
2723 |
492 |
· |
· |
· |
64 (13.0%) |
· |
379 |
| 2026-07-18 |
2863 |
501 |
· |
· |
· |
59 (11.8%) |
· |
386 |
| 2026-07-17 |
2548 |
466 |
· |
· |
· |
47 (10.1%) |
· |
334 |
| 2026-07-16 |
2536 |
505 |
· |
· |
· |
67 (13.3%) |
· |
363 |
| 2026-07-15 |
2688 |
521 |
· |
· |
· |
52 (10.0%) |
· |
363 |
| 2026-07-14 |
2562 |
511 |
· |
· |
· |
54 (10.6%) |
· |
368 |
| 2026-07-13 |
2657 |
506 |
· |
· |
· |
49 (9.7%) |
· |
348 |
| 2026-07-12 |
2544 |
468 |
· |
· |
· |
58 (12.4%) |
· |
345 |
| 2026-07-11 |
2484 |
503 |
· |
· |
· |
63 (12.5%) |
· |
381 |
| 2026-07-10 |
2365 |
482 |
· |
· |
· |
60 (12.4%) |
· |
341 |
| 2026-07-09 |
2273 |
439 |
· |
· |
· |
51 (11.6%) |
· |
352 |
| 2026-07-08 |
2381 |
479 |
· |
· |
· |
49 (10.2%) |
· |
357 |
| 2026-07-07 |
2338 |
425 |
· |
· |
· |
51 (12.0%) |
· |
362 |
| 2026-07-06 |
2421 |
414 |
· |
· |
· |
47 (11.4%) |
· |
353 |
| 2026-07-05 |
2634 |
437 |
· |
· |
· |
37 (8.5%) |
· |
351 |
| 2026-07-04 |
3307 |
527 |
· |
· |
· |
76 (14.4%) |
· |
633 |
| 2026-07-03 |
2128 |
386 |
· |
· |
· |
59 (15.3%) |
· |
359 |
| 2026-07-02 |
2219 |
405 |
· |
· |
· |
44 (10.9%) |
· |
321 |
| 2026-07-01 |
2451 |
429 |
· |
· |
· |
42 (9.8%) |
· |
297 |
| 2026-06-30 |
2184 |
454 |
· |
· |
· |
47 (10.4%) |
· |
278 |
| 2026-06-29 |
2363 |
483 |
· |
· |
· |
36 (7.5%) |
· |
297 |
| 2026-06-28 |
522 |
105 |
· |
· |
· |
12 (11.4%) |
· |
75 |
* today is a partial day. Greyed trial cells are cohorts younger than 7 days — their trial window hasn't closed.
Verification email delivery (from token rows)
| Day | Emails | Delivered (rate) | Bad address | Bounced | By status |
|---|
| 2026-08-02 | 25 | 22 (88.0%) | 2 | 1 | bounced: 1, delivered: 22, undeliverable: 2 |
| 2026-08-01 | 478 | 416 (87.0%) | 2 | 24 | accepted: 36, bounced: 24, delivered: 416, undeliverable: 2 |
| 2026-07-31 | 614 | 1 (0.2%) | 0 | 0 | accepted: 613, complained: 1 |
| 2026-07-30 | 613 | 0 (0.0%) | 0 | 0 | accepted: 613 |
| 2026-07-29 | 67 | 0 (0.0%) | 0 | 0 | accepted: 67 |
How to read this
- Manual (email) is the proxy for email+password signups: email set, no Apple/Google/Facebook id, no phone. Pre-observe this is the baseline the funnel will be judged against.
- Working as intended: manual signups may drop, but the losses concentrate in Bad address, verify-through of delivered signups stays high, and All trials stays flat vs baseline.
- Discouraging real users: signups whose mail was delivered and still never verify accumulate, and cohort trial rates fall below the baseline manual trial rate.
- Bounced is not the same as a bad address: a bounce can be a full mailbox or our sending reputation being blocked, so a rising Bounced column is an operational warning, not successful filtering.
- Delivered vs accepted:
accepted only means Mailgun took the message. delivered, bounced, and deferred come from Mailgun's delivery webhook; undeliverable is set before sending when DNS says the domain takes no mail. Rows still sitting at accepted have an outcome we have not heard back on yet.
- Mixpanel/AppsFlyer registration events are deferred until verification on updated app builds — the DB counts here are the honest signup denominator.
- Trial columns use
first_trial_or_purchase_date within 7 days of signup.