Leasing AI page: the homepage's figures, nothing unsourced

/platform/leasing-ai reads the same stat records as the homepage (<1 min, 41%, and the 92% Sean kept), and loses the figures nothing backs.

11 Oct 2026 · branch copy/leasing-ai-shared-stats · captured against propflow-stage · at 2e3ef049481+local

Data note. Every name and balance below comes from propflow-stage, which is anonymized by construction (ADR-0097/0110 — phones in the impossible +1000 NPA, emails @example.test). The identical capture against prod would carry real tenants' names and balances and is not safe to attach anywhere, which is why the generator refuses to run against one.

1 · Before (main): the velocity chips, desktop

Today's page, captured from a commit just behind main (the leasing page is the same on main). It hardcodes its own 47s, 38% and 92%, so it already disagrees with the homepage, and it adds figures with nothing behind them: Industry median: 10h 28m, +38pts vs team-only coverage, +44pts vs voicemail-first and 0 Missed inbound.

Asserted in the captured DOM — Industry median: 10h 28m · +38pts · Missed inbound, any channel

Before (main): the velocity chips, desktop
Provenance. Captured from commit b83e6b8f747, read from the worktree the server was started from (/Users/miniclaw/code/PropFlow/wt-site-stats-base) — the server could not report its own commit, so this is the weaker of the two sources. That commit is already merged into origin/main.
Sections 2–5 and notes

2 · After: the velocity chips, desktop

Three chips, each read from the homepage's own record in claims.ts: <1 min median first reply, 41% of tours booked after hours, and the 92%, which stays only because Sean ruled its slot stays until he replaces it. The made-up comparisons and the 0 missed claim are gone (4 of 185 prospect-started conversations went unanswered).

Asserted in the captured DOM — Median first reply. · Tours booked after hours · Leads qualified on first touch. The chips' labels come from the shared records; the figures count up on screen.

After: the velocity chips, desktop

3 · Before (main): the response-time chart

This chart's series is typed into the code by hand, yet the caption calls it the portfolio-median first-reply time across pilot properties, past 90 days, down 75%. Our real monthly medians went the other way, 9s in April to 23s in October, so there is no honest version of this picture. The section goes, with its unsourced over 24 hours heading and a Zillow figure the linked page would not let us check.

Asserted in the captured DOM — The national average inquiry response is · pilot properties

Before (main): the response-time chart
Provenance. Captured from commit b83e6b8f747, read from the worktree the server was started from (/Users/miniclaw/code/PropFlow/wt-site-stats-base) — the server could not report its own commit, so this is the weaker of the two sources. That commit is already merged into origin/main.

4 · After: velocity runs straight into the bilingual demo

The whole page after the change. The hero now says median first reply under a minute ... Nights and weekends included (no holiday), and the velocity section is followed directly by the bilingual demo.

The chips count up the first time they scroll into view, so this capture scrolls past them first.

Asserted in the captured DOM — median first reply under a minute · Nights and weekends included. · Speed is the

After: velocity runs straight into the bilingual demo

5 · After: the velocity chips, phone (390px)

One column on a phone.

Asserted in the captured DOM — Median first reply. · Tours booked after hours

After: the velocity chips, phone (390px)

Not on any public page

PersonaSwitcher, QuoteSpotlight and ROIBreakdownGrid carry 47s only inside their component-library previews, which sit behind staff sign-in. The public pages that use ROIBreakdownGrid pass their own cards, and none shows the reply-time, after-hours-tours or qualified-leads figures. (/solutions/multifamily prints a different -38% vacancy loss, which is its own unsourced number.)

PropFlow Docs