Recommendations
Every completed AI Visibility check ends with a prioritized list of recommendations — specific, evidence-backed fixes for the gaps the check found, not generic advice. Each one shows why it was suggested (the evidence behind it), how much it's expected to help, and how much work it takes.
Kinds of recommendations
| Recommendation | When you'll see it | What happens when you act on it |
|---|---|---|
| Create content | You appeared in less than a third of engines' answers for one of your questions | Generates a draft (a blog post or a web page, depending on the question) aimed squarely at that question |
| Contest a topic | A named competitor was mentioned more often than you were overall | Generates draft content aimed at the specific question(s) that competitor is winning |
| Canonical business profile page | Less than a quarter of citations across your answers pointed to something you own | Generates a single authoritative page about your business, with structured data included — see Setup notes |
| Repurpose to social | Very few citations were social despite you having active social profiles | Generates a social post repurposing content you already have |
| Address sentiment (manual) | AI answers about you lean negative on average | No generate button — this points at a real service or reputation issue that needs a human look, not new content |
| Fix listings (manual) | Several directory citations showed up with none pointing to something you own — usually a sign your business name/address/phone isn't consistent across directories | No generate button today — acting on this from inside the app is planned for a later update; for now it just flags the issue |
The two manual items are called out honestly rather than hidden or faked with a button that doesn't do anything — if the app can't act on something yet, the recommendation says so plainly instead of implying it can.
Generate → its own session → review → publish
Clicking Generate opens a brand-new session for that recommendation — the same way starting work on a campaign-plan item does — and the assistant gets to work in it right away. A row shows Open session as soon as that session exists, so you can watch it happen or come back to it later; the result is a draft (never something published or sent), which you review the same as any other draft — ask for changes, edit it yourself, or approve it to publish. Nothing goes live on its own; see Using AI sessions for how drafts work in general.
If there are several generate-able recommendations open at once, Generate all drafts kicks all of them off together — each one still gets its own separate session, so you get that many "Open session" links, not one shared thread.
You can also Dismiss a recommendation you don't want to act on — it stays visible on the report, marked as dismissed, so you keep a record of what you decided to skip and why.
Re-running later shows the trend
Recommendations are tied to the check that produced them. When you run another check later — after publishing something, fixing an issue, or just checking in again — you'll see a fresh set of recommendations reflecting where things stand now, and your visibility trend will show whether your score and sentiment moved in the meantime. That's the real test of whether a recommendation was worth acting on.
FAQ
Q: Does clicking Generate publish something automatically? A: No. Generate always produces a draft for you to review first, exactly like asking the assistant for an email, a social post, or any other piece of content. Publishing is always a separate step you take yourself.
Q: Why can't I generate a fix for the manual recommendations? A: Address sentiment and Fix listings both point at things the app doesn't have an automated path for yet — a service issue behind a low sentiment score, or a business-listing inconsistency across directories. They're still worth acting on; there's just no in-app "generate" for them today. Support for these is planned for a later update.
Q: What do impact and effort mean? A: They're a quick read on priority — impact estimates how much a fix could move your visibility, and effort estimates how much work it takes. Higher impact, lower effort recommendations are generally worth doing first.
Q: If I dismiss a recommendation, can I get it back? A: It stays visible on that run's report marked as dismissed, so you can still see it — dismissing just removes it from your open to-do list. If the same gap still exists after a later check, a fresh recommendation for it will appear on that check's report.