Missed chases — the recovery list

Recurring only. Every row here comes from the per-day chase files, which are collapseToCustomers(day).filter(bucket === 'Finance' && !cleared) — and Finance means the Square location name starts with Recurring. A card failing at the studio counter is a sale that did not go through, not arrears, and it is not on this page. Checked: 0 of the 7,556 source rows were studio sales, deposits or office declines, and 0 of the 101 rows in the download are anything other than Finance arrears (0 owe nothing).

What went wrong. A day's hand-over file used to cut everyone the ledger had ever recorded, and the ledger has no time limit — so the customers who fail every cycle were handed over once and then dropped forever. On 31 August that cut 684 of 877 people and Hamad received 193 names. The cut rule was fixed on 3 Sep 2026; this page is the backlog it left behind.

How a person counts as handed over. They appear in the laptop's daily CSV, or Render's copy of it, or a weekly window file the ledger actually records as sent. Matching is Square customer id, then last-10-digit phone, then lowercased email — the same three keys the ledger uses. Where two copies of a day disagree, the larger hand-over wins, so this list under-claims rather than sending anyone a second call.

Download the recovery list — 101 people, one row each

One row per person, biggest debt first, with Days missed and Times missed added to the usual chase columns. Somebody who missed on three different days is one row, not three.
11 column(s) are blank for some people — BIN 101, Issuer 101, Early access 101, Card type 101, Prepaid 101, Payday stated 101, Payday kind 101, Payday stated on 101, Payday quote 101, Charge day 101, Charge day % 101 rows. Those columns were added to the chase file after the older per-day lists were written, so a person whose latest miss is on an older day has nothing to put there. The columns are kept rather than dropped so this file never stops matching the chase file Hamad already works; rebuilding a day fills them in.

Of the 7,556 people who missed a recurring payment across the days we can check, 7,429 were handed over and 127 were not.

The whole point of the fix: these are the repeat missers the old rule was designed to suppress.

To close that gap the per-day uncut list has to be rebuilt for those days: node scripts/reports/report_square_declines.js 2026-09-18 2026-09-18 --split-days, then re-run this report. Nothing here reconstructs those names from context — the gap is stated, not filled.

The per-day files are one route to "who missed a recurring payment that day". The rolling declines.json pull is an independent one. It counts every recurring decliner, including people who have since cleared, so it should be the larger of the two. A day with people in the file but not in the pull would mean the two disagree, and that is what the last column watches for.