Backlog

Raised, not sequenced. Every row says what would promote it. A row leaves by being sequenced into planning.md with an issue, or by being deleted when it stops being true.

What is raised?

The alpha merchant cohort from her existing contacts has no issue at all.

• ALPHA.md line 99, P1. The plan calls it the cheapest source of merchants, and it is the only row in that group tracked by nothing.
• What would promote it: her word on who is actually in the cohort. Only her conversations can produce that list, so it cannot be worked ahead of her.

#867 does not link its own design spec.

• The safety-measurement design has existed since 2026-08-16 at docs/planning/specs/2026-08-16-alpha-safety-measurement-design.md, and #867: feat(safety): how we measure whether Alpha participants felt safe, without averaging groups together does not cite it.
• The cost is measurable: on 2026-08-23 the morning survey carried the question forward as undecided, and she had to remember the spec from memory.
• What would promote it: nothing. It is a one-line edit to the issue body and it is outward-facing, so it needs her word rather than a slot.

The in-app redemption P0 is now tracked, and still unanswered.

• ALPHA.md line 38 previously read In progress and linked two closed issues, one of which was about replacing mock numbers on the merchant dashboard, a different subject.
• Applied 2026-08-24: the row reads Needs operator input, the off-topic link is gone, and #965: business(alpha): decide what redeeming at the counter means, for the no-POS attribution path now carries the question. #704 stays as the record that the claim and redeem flow shipped.
• What would promote it: her answer to that question. The issue exists now; what the row MEANS is still a product call and nothing was decided for her.

Two alpha P1 rows have no issue either.

• "Ad Network as a first-class surface" (ALPHA.md line 46) and "Venues search and filters" (line 47). Both Issues cells are empty.
• Neither was filed on 2026-08-24, because the session's goal line covered P0 rows only and both rows are also placement candidates (see prototype-vs-alpha-placement.md).
• What would promote it: deciding where each row lives first. Filing an issue against a row that is about to move is how a link ends up pointing at the wrong page.

Three Prototype rows are also listed as "Not scheduled" in the launch README.

• Team-built apps platform (#827, #849), Intranet search (#800) and Go links (#811) appear in PROTOTYPE.md Phase 4 and in the README.md Not-scheduled table, which the same README defines as "wanted, with no release attached yet". A row cannot be both.
• The apps platform also carries two names: "Team-built apps platform" on one page, "App creation portal" on the other.
• What would promote it: nothing structural. It is a three-row deletion from README.md. The 2026-08-24 lane door was scoped to ALPHA.md and PROTOTYPE.md only, so README.md was still out of reach; it needs her word naming that file.

The launch README's "Needs review" section is stale.

• It still says nine rows have no issue behind them, naming penetration testing, Google Places, safety-comfort measurement, pilot venue agreements, repeatable pitch, sales and onboarding process, target launch events, Google Workspace and Slack, and hire a second engineer.
• All nine now carry issues (#865 through #873), verified against the rows on 2026-08-24.
• What would promote it: nothing. It is a one-section rewrite. Same blocker as the row above: the 2026-08-24 lane door did not name README.md.

Three projects map to no launch-plan row, and two of them probably should.

• admin-auth-one-source (#886) and admin-portal-visible-defects (#882) cite issues that appear in neither ALPHA.md nor PROTOTYPE.md, re-verified 2026-08-27. A fail-open on admin authorization and a defect visible in a demo do not look off the alpha path.
• #881 from the same project IS on PROTOTYPE.md, as the "Profile completion metric" row, so only two of the three are unplaced.
• What would promote it: her word on which row #886 and #882 belong to. #881 needs nothing, because it is already placed. project-creation forbids matching a project to a row by guess, and that rule is the reason five links were wrong before.

The Why column is wrong on a lot of rows, and "the experiment" is the tell.

• Her words, 2026-08-24: "I think a lot of the 'WHY' values are incredibly wrong for a lot of these line items. For example, what experiment? There is not experiment."
• Three rows lean on that framing: line 40 ("How we watch the experiment while it runs"), line 63 ("If Alpha users carry iPhones, push is part of the experiment"), and line 109 ("experiment design"). Alpha is a launch with invited users, not a study, so the word imports a premise she has not made.
• This is wider than three greps. The Why column was written to justify rows rather than to say why they are on the plan, so the audit is a read of every Why, not a find and replace on one noun.
• Why it matters more than a wording nit: ranking reads this file, and a Why that argues for a row it should not argue for is how a row keeps its priority after the reason expired.
• What would promote it: it folds into the row-by-row pass this project already owns, so it needs a slot rather than a decision. Worth doing in the same sitting as the stale-status pass, because both require reading every row anyway.