Auto-book or review queue: how the decision is made

Most submitted inventories are fine. A few are thin, contradictory or quietly missing the hard parts. Reviewing all of them to catch the few is how a booking pipeline turns back into a phone queue — which is the thing you installed it to stop.

Scoring lets you spend attention only where it changes the outcome. The clean submissions book while your office is closed. The questionable ones wait for a person, and arrive already labelled with what is questionable about them.

• Completeness — how much of the home the session actually covers, and whether obvious categories are missing.
• Media coverage — how much of the inventory is backed by photographs or video rather than typed in unaided.
• Internal consistency — whether the inventory, the qualification answers and the stated home size agree with each other.
• Route sanity — whether the pickup and delivery addresses resolve to a lane that makes sense.

Media coverage is a real factor, so a text-only submission scores lower than the same inventory backed by photographs. That is by design, not a defect: a list nobody photographed is a list nobody has checked.

The scoring weights, the base scores and the auto-book threshold are configuration, not constants. They live in the admin configuration screens and can be changed without a release.

The usual pattern is to start conservative — a high threshold, most sessions reviewed — read a few weeks of the queue to see which holds were unnecessary, then lower it until the queue only contains submissions you genuinely wanted to see.

A held session arrives with its score, the reason it was held, the full inventory and the quote that was generated. From there it can be approved as it stands, adjusted and re-priced, or declined. Whichever happens is recorded against the reservation.