Get Recommendations

Summary#

Under every tool result is a panel offering to read that result and write you a prioritized action plan: what to fix first, what to do next, and the quick wins. It is one button, it takes about ten seconds, and it turns a page of measurement into a list of things to do. It is available on the Pro plan and above, and it draws on its own monthly allowance, separate from your credits.

Purpose#

A measurement tells you what is true. It does not tell you what to do about it, and for most people that second step is the hard one. A backlink profile with a spam score of 41 and eighteen toxic domains is only useful if you know whether to disavow, to outreach, or to ignore it this quarter.

The recommendation panel closes that gap without you having to interpret the report yourself. It reads the actual findings from the run you just did, weighs them by impact and effort, and writes the plan in plain language with the steps spelled out. Every recommendation links back to the tool that would let you act on it.

Requirements#

• A finished tool result on screen. The panel reads that result; it does not run anything new.
• The Pro plan or above. Free and Starter accounts see an upgrade panel instead.
• Remaining recommendation allowance for the month, and at least 1 credit.

Permissions#

Anyone in the workspace can use it; there is no separate role for it. The minimum plan and the per-plan allowances are configurable by Metric Vault, so an account with a custom agreement may differ from this table. Your own limit is enforced at the moment you click.

Navigation Path#

Dashboard → any tool → run it → Get recommendations

There are two identical buttons. One is in the panel below the result. The other is at the right-hand end of the result's category tab bar, labelled Get recommendations. Both do exactly the same thing.

Step-by-Step Guide#

1. Run any tool and wait for the result. See How to run a tool.
2. Look below the result. The panel is headed AI Action Plan, with the small line "Built from your result and the live SERP", then "Turn this result into a step-by-step plan." Underneath it previews what it found: how many things there are to improve, and chips reading "N to fix first", "N to do next" and "N quick wins".
3. Click Get recommendations. The note beneath the button says "Takes about 10 seconds", which is accurate for most results.
4. Watch the progress. The panel switches to a four-step loader with headings such as "Building your plan" and "Working through your result", and the steps name what is happening: reading your result, weighing the key signals, finding the biggest opportunities, writing your action plan. Several tools have their own wording; Gap Finder, for example, reads "Comparing you against the competitor".
5. Read the plan. You get a The bottom line verdict, a stat row, and then the recommendations in three groups.

| Group | Sub-label | Meaning | |---|---|---| | Fix first | high impact, do these now | The findings with the largest effect. | | Do next | worth the effort | Solid gains that are not urgent. | | Quick wins | easy momentum | Small effort, small but immediate gain. |

Each card carries a rank number, a title, the area it belongs to, what it is, a Why it matters paragraph, a Steps to take list, and Effort and Impact shown as three-dot meters.

1. Act on one. Each card ends with Fix this in <area>. Clicking it navigates to the right tool and fills in the target for you.
2. Regenerate if you want a fresh pass. The footer has Regenerate plan. That is a new request and it costs another recommendation and another credit.

Which tools have it#

The panel appears wherever there is something to recommend. Around thirty tools have dedicated rule sets that read their specific findings, including the technical tools (robots.txt, sitemap, redirects, hreflang, headers, schema, mixed content, tech stack), the content tools (duplicate content, internal linking, content decay), the research tools (topic research, keyword magic, top pages, position tracker, audience overlap), the AI-visibility tools (prompt research, AI competitor research, brand performance, questions, prompt tracking, AI Overview and citation trackers), and the promotion tools (PPC research, advertising research, AdClarity, Shopping Ads, social tracker, social analytics, influencer analysis, media monitoring, narrative driver).

Other screens attach their own panel inline, including the Backlink Hygiene Toolkit, Forum & Reddit Visibility and the Competitor Intelligence workflow.

If a rule set examines a result and finds nothing worth acting on, no button is drawn. That is the correct outcome, not a failure: there is nothing to fix.

Screenshots#

Troubleshooting#

FAQs#

Does it re-run my tool? No. It reads the result already on screen. That is why it is fast, and why the plan is only as current as the run it is reading.

Why did it charge a credit as well as a recommendation? The recommendation is itself an AI request, so it is metered twice: once against the recommendation allowance that controls access to the feature, and once as a 1-credit run like any other AI tool.

What happens if the AI call fails after I clicked? The recommendation is counted before the request is sent, so a failed attempt can still consume one. This is deliberate, so that several clicks at once cannot slip past the limit. If you believe you were charged for something that never produced a plan, contact support with the time of the attempt. See Contacting support.

Where can I see how many recommendations I have used? There is no counter in the interface today. The allowance is enforced server side, and you find out you have reached it from the gate panel.

Can an administrator change the minimum plan or the allowances? Yes. Both the minimum plan and each plan's monthly number can be overridden at the platform level, and the change takes effect within about a minute. See Runtime configuration.

Does the plan get saved? The recommendation run is recorded like any other run, so it appears in your activity. The panel itself is rebuilt each time you open the result. For the Domain Overview plan, a plan generated for a domain is reused for the rest of that UTC day and shown across both the Quick Preview and the full Domain Overview without charging again; Regenerate plan forces a fresh, billed pass.