Blog / Tools & automation /

Search Console recommendations: worth acting on?

Search Console has started suggesting things. A little card pops up with a recommendation, and the obvious question is: do I just do it?

The feature began appearing in August 2024, rolling out gradually rather than reaching every property at once. A recommendation can save time by pointing at an opportunity in your data. But it needs context. It’s an invitation to investigate, not an order from Google.

If you see recommendations, first understand what’s being suggested and which data it relies on. Then decide whether it fits the site and its goals. If you don’t see the feature, don’t change settings or assume something’s wrong. The rollout and interface vary.

What a recommendation is—and is not

A recommendation is an idea surfaced from what Google knows about a Search Console property. The specific cards can change over time. Read the description in the product rather than assuming every property gets the same advice, or that a recommendation signals a ranking problem.

It isn’t a manual action, a promise of more traffic or a full list of what the site should do next. Search Console only works with the data it has. It doesn’t know your business constraints, product priorities, legal requirements or content decisions. The owner still has to decide whether a proposed action is useful and safe.

A quick way to assess one

I start with the page or query involved. Is it a genuine opportunity for the business? Check the date range and the underlying report. If it suggests improving a page, read the page and look at the searches linked to it. If it relates to a search appearance or feature, verify the relevant information is accurate and on the page.

Then ask what the change costs and what it might affect. Expanding a page can create low-value content if you’ve nothing useful to add. A technical change may touch templates or many URLs. If an action could change indexing, structured data or navigation, test it safely and keep a way to undo it.

Prioritise by expected value and effort, not by how prominent the card is. A small issue on an important service page can matter more than a broad suggestion with little business impact. I’m also happy to decide not to act and write down why. That’s a perfectly valid outcome when a recommendation doesn’t fit.

Check the result after acting

If you make a change, write down what changed, which pages were affected and when. Watch the relevant Search Console report for errors or shifts, but don’t judge the result after a day or credit the edit for every movement. Demand, seasonality, other site changes and search systems all affect performance.

Where possible, make one focused change at a time. Update several templates, rewrite content and rework internal links all at once, and you’ll never know which change mattered. A recommendation may have helped you notice an issue, but the evidence of a result has to come from the site and the outcome you care about.

Keep your own priorities

A clean Search Console account is nice. It isn’t a business plan. The site may need a clearer offer, a better checkout flow, accurate product information or content that answers a question, and none of those may ever show up as a recommendation. Treat the product as one input in a review that also covers customer questions, technical checks and business goals.

For teams, give each accepted task an owner and keep a short log. It avoids a common mess: one person acts on a recommendation, another makes a related change, and later nobody remembers why. Clear ownership makes work easier to test, review and reverse.

Watch for a mismatch between data and advice

Recommendations can rest on a narrow slice of what Search Console sees. A suggestion about one page doesn’t mean the rest of the site was assessed, and no suggestion doesn’t prove a page is healthy. The product is useful because it can find something worth checking, but the person responsible for the website has the wider context.

Before accepting a task, ask whether the page is still current, whether it has an owner and whether the change is already planned. If the page is being retired, improving its metadata is wasted effort. If the recommendation points to a feature the business can’t maintain, adding it creates future support work. Check permissions before giving a contractor access, too. Search Console access should go only to people who need it, and changes should be documented. That small bit of process stops an automated prompt becoming an unreviewed site-wide change.

When the suggestion is not actionable

Sometimes you can’t act straight away. The page may be waiting on approved product information, the team may be planning a redesign, or the account owner may lack access to the system where the change belongs. Record the dependency and decide who resolves it. Don’t ask anyone to rush a change just to clear a card from the screen.

If a recommendation looks wrong, keep what you observed and check the underlying page and report. The data may be delayed, grouped differently than you expect or sampled. Ask whether it still holds once you account for the site’s business model and publishing process. Feedback options help Google understand the feature, but they don’t replace your own diagnosis. Deferring, declining or asking for technical advice are all informed decisions.

If a recommendation is showing in your account, open the underlying report before you act. What evidence would make you confident it suits this site?