Every CMN lead Sun 6 Sep – Sat 19 Sep,
checked against every 6AB, 5AB, N1N and R row that person still has anywhere back to
Tue 10 Mar. Reaching a stage is meant to consume the one before it,
so every row found here is one that should have disappeared and did not.
Newest lead in the feed: 2026-09-18 18:50:39.
1 of 81 leftovers sat in a different city from their CMN.
Original code→ chase →
R→ re-SMS, sends a picture →
N1N→ free for a chat? →
CMN
— each step is meant to CONSUME the one before it
CMN in the window
222
people · 222 rows
Shifted cleanly
163
73.4% — nothing left behind
Still carrying an old stage
59
26.6% · 81 rows that should have gone
The shift itself failed
1
old stage within 14 days of the CMN
Two different faults — do not add them together.
1 people (1 rows) carry a stage row from within 14 days of
their CMN: that is the shift failing to consume the previous stage.
59 people (80 rows) carry one older than that — an earlier journey
that died and was never cleaned up. Different causes, different fixes.
What is left behind
A person can be counted under more than one stage, so these do not sum to the total.
| Stage still live | People | Of all CMN |
Rows within 14d | Rows older |
|---|
| 6AB | 33 14.9% | 222 | 1 | 42 |
| 5AB | 10 4.5% | 222 | 0 | 13 |
| N1N | 14 6.3% | 222 | 0 | 15 |
| R | 10 4.5% | 222 | 0 | 10 |
How long before the CMN was that row created?
A same-day leftover is unambiguous: the lead converted and the previous stage was simply never cleared.
| Gap | 6AB | 5AB | N1N | R |
Rows |
|---|
| 8–14 days | 1 | 0 | 0 | 0 | 1 1.2% |
| 15–30 days | 5 | 0 | 2 | 0 | 7 8.6% |
| 31+ days | 37 | 13 | 13 | 10 | 73 90.1% |
Which combination is left
| Stages still live for that person | People |
|---|
| 6AB | 26 44.1% |
| 5AB | 10 16.9% |
| N1N | 9 15.3% |
| R | 7 11.9% |
| 6AB + N1N | 4 6.8% |
| 6AB + R | 2 3.4% |
| 6AB + N1N + R | 1 1.7% |
Download the cleanup list — every leftover row (CSV)R is the biggest single fault, and it is not subtle.
Of the 0 leftover rows dated the
SAME DAY as the CMN, 0 are R codes. The lead converts and its R row is simply
never cleared — same person, same day, both rows still live.
R rows still live
0
within 14 days of the CMN
Of those, same day
0
— — no ambiguity at all
Distinct R codes affected
0
it is not one broken feed
The R codes not being cleared
Every one of these is {CITY}{SOURCE}R — the chase stage of a lead that has since converted.
| R code | City | Rows still live |
|---|
Original code→ chase →
R→ re-SMS, sends a picture →
N1N→ free for a chat? →
CMN
— each step is meant to CONSUME the one before it
Push gap: Sun 6 Sep, Sat 12 Sep, Sun 13 Sep
carried almost no recycled leads at all against a median day of 372, while total lead flow held up.
That is the recycled push stopping, not a quiet day.
Leads in, by stage, per day
Recycled counts every stage row plus CMN once each. The last day in range is always part-built.
| Date | All leads | Recycled |
6AB | 5AB | N1N | R |
CMN |
|---|
| Sun 6 Sep | 1,062 | 17 1.6% | 6 | 2 | 2 | 4 | 2 |
| Mon 7 Sep | 1,254 | 131 10.4% | 9 | 45 | 5 | 64 | 6 |
| Tue 8 Sep | 1,413 | 250 17.7% | 21 | 55 | 10 | 154 | 9 |
| Wed 9 Sep | 1,280 | 290 22.7% | 25 | 55 | 7 | 185 | 18 |
| Thu 10 Sep | 1,712 | 649 37.9% | 91 | 46 | 15 | 482 | 14 |
| Fri 11 Sep | 1,496 | 372 24.9% | 8 | 31 | 3 | 320 | 8 |
| Sat 12 Sep | 1,184 | 35 3.0% | 7 | 4 | 3 | 20 | 0 |
| Sun 13 Sep | 1,170 | 23 2.0% | 6 | 1 | 0 | 15 | 1 |
| Mon 14 Sep | 2,031 | 944 46.5% | 402 | 18 | 216 | 243 | 65 |
| Tue 15 Sep | 1,635 | 405 24.8% | 187 | 41 | 76 | 70 | 31 |
| Wed 16 Sep | 1,483 | 401 27.0% | 202 | 50 | 63 | 49 | 37 |
| Thu 17 Sep | 1,478 | 446 30.2% | 220 | 56 | 72 | 80 | 18 |
| Fri 18 Sep | 1,477 | 384 26.0% | 166 | 57 | 68 | 79 | 13 |
| Sat 19 Sep | 156 | 0 0.0% | 0 | 0 | 0 | 0 | 0 |
| TOTAL | 18,831 | 4,347 23.1% | 1,350 | 461 | 540 | 1,765 | 222 |
By city
| City | CMN people | Still carrying an old stage |
Shift failed (≤14d) |
|---|
| New York | 58 | 17 29.3% | 0 0.0% |
| Houston | 35 | 10 28.6% | 0 0.0% |
| Chicago | 32 | 6 18.8% | 1 3.1% |
| Florida | 30 | 8 26.7% | 0 0.0% |
| Boston | 19 | 5 26.3% | 0 0.0% |
| Nashville | 18 | 5 27.8% | 0 0.0% |
| Dallas | 17 | 5 29.4% | 0 0.0% |
| Orlando | 13 | 3 23.1% | 0 0.0% |
Yes — and that is what makes the leftovers a fault rather than the design.
A CRM pull shows a code as it stands now, so an in-place shift is invisible unless the same row was
observed earlier. Across 3,516 rows with an earlier observation on file, 36 were
caught mid-shift: the row's own code was rewritten to CMN and the old stage genuinely disappeared,
exactly as intended. 0 ever moved the other way.
Rows observed twice
3,516
the only rows that CAN show a shift
Shifted in place to CMN
36
the mechanism working correctly
Shifted OUT of CMN
0
CMN is an end state
Every shift caught in the act
| Was | Became | Rows |
|---|
| NY6AB | NYCMN | 13 |
| CHI6AB | CHICMN | 4 |
| HOU6AB | HOUCMN | 3 |
| FL5AB | FLCMN | 3 |
| FL6AB | FLCMN | 3 |
| BOS6AB | BOSCMN | 2 |
| CHI5AB | CHICMN | 2 |
| HOU5AB | HOUCMN | 2 |
| NA6AB | NACMN | 1 |
| DAL6AB | DALCMN | 1 |
| ORL6AB | ORLCMN | 1 |
| ORL5AB | ORLCMN | 1 |
What this count can and cannot see
A shift that happened before the row was ever observed is invisible and always will be —
the ledger only started recording on 25 Aug 2026, seeded from /recycled-codes' chunk cache. So the
shift count is a floor. It grows on its own as the ledger accumulates, which is why this report
runs daily: a night not run is evidence that cannot be recovered later.