The period proof: AI visibility measured at the start and the end of the window (before / after), the moves shipped between them, the scans that ran, prepared_count (site moves prepared right now and waiting for their live URL, the same count the in-app Scoreboard reads; not windowed; null when unreadable), and the receipts with their verified state and tracked-link clicks. like_for_like is the honesty field and must be read before reporting any change: true means both scans asked the same tracked questions, false means the set changed inside the window or the answer source changed, so the numbers are a new baseline rather than a movement, and null means it cannot be proven either way. Composed from the same functions the day-14 recap and the Feed proof strip use. Defaults to 14 days; ?window_days is clamped to 7 to 90 rather than rejected.