How the briefing chooses its items

Who decides what is in the briefing?

Rules do. Each morning, after the analysis, rules look through every account you monitor for candidates and put them in order (below). A language model then picks at most five of them and words them, keeping the rules' order unless a candidate's facts clearly justify a move. It cannot add an item the rules did not find, and each item's At stake figure is the rules' own.

What can become an item?

• Anomaly: a new departure from a campaign's or the account's own normal, of the kinds worth your morning (below).
• Budget pacing: the account's last 7 fully synced days are off pace, by the Accounts page's own rule.
• Search-term backlog: terms waiting for you in triage that cost money in the last 30 days.
• AI Max: over the last 30 days, a campaign's AI Max traffic has spent the account's cost per conversion on competitor terms, or has spent twice the account's cost per conversion and has no conversion, costs at least 1.5 times as much per conversion as the campaign's keyword traffic, or returns two thirds of the keyword traffic's ROAS or less.
• Tracking: a conversion action that has stopped recording, or fallen to a quarter or less of its usual week, while its campaigns keep spending.
• Auto-applied: Google applied one of its own recommendations to the account.
• Budget-limited: a campaign losing 10% or more of its impression share to budget, by Google's own 14-day figure, while spending at least 90% of its budget.
• Actions: strong new suggestions waiting in Actions, one item per account.

Which anomalies make it in?

The ones that are news. A rise in clicks, impressions or CTR never does. A fall in them does from Medium severity on its first day. A day's conversions or conversion rate needs a move of at least two conversions from normal. A departure that runs for days comes in on its first day, again when it grows more severe, and on its fifth day running (its eighth, for a figure judged on a week). When a campaign shows several related departures on one day, they come in as one.

What does At stake measure?

Money in the account's own currency, mostly a week's worth: the spend a backlog of negatives wastes, the spend lost to budget, a week of spend on the campaigns a tracking break leaves bidding blind. For an anomaly, it is the money its departure moved over the days it has run, up to 7; for pacing, the spend off the budget line. A move in your favour, such as more conversions or a lower cost per conversion, counts at half, because a move against you matters more.

How are items ordered?

First, anything that means the account's figures cannot be trusted, or that Google changed it without anyone approving: a tracking break, a risky auto-applied change. Then money, largest first, with accounts in different currencies compared at the day's exchange rate; that comparison is never shown, and every figure stays in its own currency. Housekeeping comes last, such as a backlog that wastes nothing yet. Up to 20 candidates go forward, and the briefing shows at most five.

Why doesn't the same item come back every day?

An item already in your briefing is held back before it can return: a search-term backlog, an AI Max item or the same tracking break for 7 days, a budget-limited campaign for 28 days. Each auto-applied change and each suggestion in Actions is briefed only once. A quiet morning is a short briefing, and that is the point.