Recommendations.

Recommendations is the action inbox: what to change next, generated from your own collection data rather than from generic best practice.

What it is

The rest of the product tells you what is true. Recommendations proposes what to do about it. Each item is derived from your collected answers, so it is grounded in what the AI platforms actually said about you, not in a checklist that would read the same for any brand.

What a recommendation contains

Each item carries a category, a priority, and a status you can move as you work it. Read the priority as a suggestion about sequencing rather than a ruling: you know things about your own roadmap that the collection data does not.

Working the inbox

Treat it as an inbox rather than a report. Items get actioned or dismissed, and dismissing something you have judged irrelevant is a legitimate outcome that keeps the list honest.

Recommendations overlap on purpose with Sources and GEO Audit, which are the two surfaces most of the fixes lead back to. The inbox is the front door; those pages hold the detail.

Measuring the result

Because collections run every week against the same prompts, a change you make is measurable. Ship the fix, then read the next few runs. This is the loop the product is built around, and it is the reason the cadence matters more than any single number.