Act: the overview of your actions

Act is the menu destination where measurement turns into work. It opens on the Overview tab, which answers two questions in a few seconds: what should be done now, and did what was done work? Everything is read from the project's action log: nothing is estimated, nothing is called externally.

The Overview cards, from top to bottom

• Where your actions stand: the four steps of the cycle — To do (waiting), In progress (started), Published (live, awaiting reading), Verified (read again at the next reading) — with their count and a stacked bar. Click a step to open the matching list in the Actions tab. The line below gives the share of actions published or verified; dismissed actions are counted separately, outside the cycle.
• Do first: up to three actions to do, the strongest by estimated impact (the same order as “This week's action”). Each one shows its source, the SEO or GEO Score brick it targets and its impact, with the How to fix it link.
• Measured effect: for each verified action, the change in the targeted brick between the reading before publication and the next reading, as bars (0 = brick unchanged; change in score points). The chips count actions up, stable and down. Below, published actions awaiting a reading, with the date of the next possible reading.
• Pace: the number of actions published over the last 4 weeks, compared with the 4 before, and a week-by-week histogram (12 weeks at most, the current week in violet). If nothing has been published for 4 weeks, the card recalls the date of the last publication.
• Where actions come from: open actions (to do and in progress) by source — recommendations, Search Console opportunities, content queue, briefs, schemas, rank against a competitor, actions entered by hand.
• Go further: one tile per Act tab (Actions, Impact proof, Content queue, Briefs, Schema Injector, Generators, GEO Lab), with its count when the log knows it.

When an action becomes “verified”

• When it moves to published, the targeted brick is frozen: that is the “before” score.
• At the next Sunday scan, at least 7 days after publication, the brick is read again: that is the “after” score. The action is then verified — the Actions tab calls it “Measured”.
• An action with no targeted brick is never measured; a brick not measured at the last reading (missing source) waits for the source to return. The card says so, without showing a zero.

The Actions tab

The Actions tab gathers, in a single log, everything that is waiting or done, wherever it comes from. It has two sub-tabs: This week's actions and Recommendations.

• At the top: the same clickable pipeline, bars by source and the estimated impact of open actions (high, medium, low, unknown).
• This week's action: the strongest action to do, with How to fix it, Mark in progress and Dismiss.
• Five status tabs: To do · In progress · Published · Measured · Dismissed.
• Buttons say who acts: Mark in progress and Mark published record what you have done; Back to to-do, Back to in progress and Pick up again undo a change. A measured action has no button left.
• Add a manual action: for work done outside Kaiku (technical fix, rebuilt page, link campaign…), with the targeted brick, the URL and the status — “Already published” freezes the brick right away.
• Changing an action's status here also changes it at its source (recommendation, content queue, brief…).

“How to fix it”: the guide for each action

Click an action's title, or the How to fix it link, to open a panel on the right.

• The problem in one sentence, with the known facts and the URL concerned.
• Why it matters: the targeted brick, its weight in the SEO or GEO Score, what it measures and its latest score — “no reading” when it is not measured.
• The steps: what the source recommends, then the method specific to the brick, filled in with your keyword, your page, the competitor and the engine concerned.
• The Kaiku tools that do the work: one click closes the panel and opens the right tool (brief, schema, citability, positions…).
• How we will know it's fixed: the brick frozen at publication, read again at the next reading.
• Detailed AI plan: on request only, written in the project's language, generated once then kept on the action, to be checked before acting. A viewer can read an existing plan, not generate one.

If the Overview is empty

Without a project, the page offers to create one. With a new project, the cards say what is missing: “Nothing to do right now” when everything is in progress or published, “No verified action yet” until a published action has been read again, “No action published yet” for the pace. Recommendations, the content queue, briefs and schemas feed the log as soon as they exist.