The What to do now section answers the question that work with a report starts from: “out of everything visible in the numbers, what do I take first”. KO runs the project’s data through its algorithms, keeps what the volume supports, and lays it out by priority.
This is not a newsletter of advice and not an autopilot: KO shows a finding and its breakdown, and a person makes the decision.
What you see here
At the top is the selection line: how many algorithms were checked, how many signals were found and how many actions were selected from them. Next to it is Knowledge as a percentage: how far the project’s data goes towards answering questions of this kind. If the funnel is not fully mapped, a funnel incomplete note stands here too — some conclusions cannot be built on such data.
Below, the actions are laid out in priority bands: Critical · High · Medium · Low. Inside a band are cards. A card holds:
- Category and status — the top line.
- The type of result: a confirmed bottleneck, a probable bottleneck, a measurement problem, a forecast or an opportunity. Next to it is the confidence as a percentage; hover it to see what it was built from.
- Title and description — what exactly was found.
- Business impact and cost of ignoring — what changes if you act, and what happens if you do not.
- Dimension, effort (low · medium · high) and when it was found.
On the right is the signals rail: what KO noticed in the data but did not raise to a separate action. A signal can be hidden — it moves to the Hidden block and comes back when the picture changes.
How to move an action
The card carries status buttons: In progress, Done, Not relevant. Only the transitions the engine considers allowed from the current status are shown, so nothing can be marked done twice.
The status is shared across the project: everyone with access sees it. That is the point of the section — so that a team does not work through the same finding three times.
Not relevant is a full answer, not a refusal. KO counts rejected actions: they show which findings do not land in this project.
The decision breakdown
The Open breakdown button (or a click on the card) opens the right-hand panel. It holds what the card does not:
- Why this first — what the priority is based on.
- What we know — facts from your data, with numbers.
- What we assume — hypotheses that do not follow from the facts but explain them.
- Forecast — what changes if you act.
- What to do now — the concrete steps.
- How to verify — the metric, the target value and the period that will show in a week or two whether it worked.
- What the confidence is made of — the contribution of each signal with its weight.
- Related insights — other findings on the same subject.
At the bottom of the panel — when the finding appeared and how many times it has been shown.
How to read it
Confidence is not the probability of success. It answers a different question: how solid the finding itself is — whether the volume is enough, whether the effect is stable, whether this is a one-off. An action with 60% confidence may bring more money than one with 95% — the priority weighs both.
Priority is calculated from money and effort together, not from the size of the deviation: a finding on a small channel will not become critical even if the drop there is manifold.
A measurement problem is an action too. If KO sees that part of the spend is untagged or that site events do not stitch to leads, it raises that to the top: while the data lies, the other conclusions are not worth reading.
The wording is written by the engine, not by a template. The numbers in the text come from your data over the report’s window; change the period and the set of actions may change.
If the list is empty
- No urgent actions — the algorithms ran but found nothing worth your time. Next to it stands how many algorithms were checked and how many signals were found: an empty list after dozens of checked algorithms is an answer, not a failure.
- Hidden by the filter — some actions are not shown because of the report’s conditions. Drop the filter in the toolbar.
- Knowledge is low — there is not much project data yet. What raises it most is a mapped funnel (the Fields area of the project settings) and a connected spend source.
- Nothing loaded — the request failed; press Retry.
What’s next to this
- How a report works — the report’s period, filter and signals, from which the section takes its defaults.
- The Breakdown section — where to see the numbers the breakdown refers to.