What a Good SEO Recommendation Should Show You — Guide

A good SEO recommendation should show the evidence behind the decision, the page or scope affected, the reason the action matters, the owner, the verification step and the limitation. It should help you decide what to do next without pretending that a recommendation guarantees rankings, traffic, leads or revenue.

“Improve SEO” is not a recommendation. “Review the title and opening answer on this service page because the selected Search Console window shows impressions but weak selection; compare the page and query again after the review” is closer. It names the evidence and the next question while leaving the outcome honest.

Direct answer

Before accepting an SEO recommendation, look for these fields:

• Target: Which URL, template, query group or business path is affected?
• Observation: What was directly measured or seen?
• Date and scope: When and under what conditions was it measured?
• Reason: Why might the observation matter to the page or customer decision?
• Action: What should be inspected, changed, consolidated or measured?
• Owner: Who can make or approve the decision?
• Verification: What will be checked afterward?
• Boundary: What does the evidence not prove?

A recommendation may be sensible even when the evidence is incomplete, but the assumption should be visible. A recommendation is not stronger because it uses more technical terms. It is stronger when another reader can follow the evidence to the decision.

Core explanation

The first quality test is traceability. Can you trace the recommendation to a page, report row, rendered condition or dated measurement? If the report says “add more content” but does not identify the page, question or missing answer, it is too broad to verify. If it says “fix technical SEO” without naming the condition, it is a category rather than an action.

The second test is fit. The action should fit the evidence. A page with an access problem may need a URL or indexability review before a rewrite. A page with impressions and weak clicks may need a title, query-fit or result-promise review. A page that is difficult to reach may need a relevant internal link. The recommendation should not leap to a new page simply because a keyword exists.

The third test is consequence. Explain what the finding could affect without overstating it. A missing link can create a discovery question. An unclear heading can make the page’s purpose less obvious. A template defect can repeat across URLs. These are useful possibilities; they are not proof of a lost customer or ranking position.

The fourth test is ownership. Some actions require an editor, some a developer, some a business owner and some a measurement owner. A recommendation should not assign a technical change to someone without the authority or access to make it. It should also make clear when a business decision is needed, such as choosing between two page owners.

The fifth test is verification. A change should have a later check appropriate to its evidence. A rendered-page change can be checked on the page. A link change can be checked by following the route and reviewing the public output. A Search Console comparison needs a defined date window. A call or form question needs an authorized business measurement.

The sixth test is limitation. The recommendation should say what it cannot establish. Search Console can show a selection signal, not a sale. A lab performance result can show a test condition, not a field ranking result. A public listing can show visible context, not private conversion data. The limit keeps the action useful without turning it into a promise.

Good recommendations also distinguish fix, investigate, refresh, consolidate, measure and defer. These verbs are not interchangeable. Sometimes the evidence supports a correction. Sometimes it supports learning more before changing the page. A recommendation that always says “fix” may be hiding uncertainty.

Worked example

Consider a hypothetical recommendation: “This page needs a complete rewrite because SEO is poor.” It has no URL, date, observation or reason. It might be correct that the page needs work, but the statement cannot be tested.

Rewrite it as an evidence record. “On the service page inspected on 12 October 2026, the title names the service but the opening section leads with general company history. The page is linked from the homepage, but the next action is difficult to find on mobile. Review whether the page should own the service intent, make the answer visible earlier and test the contact path. This inspection does not prove a ranking loss or a lost enquiry.”

The revised recommendation shows the URL category, date, observations, possible relevance, action and limitation. It does not claim that a rewrite will create a particular result. It also leaves room for a business owner to decide whether the page is the right owner before changing the copy.

Now add first-party search context if it is available and authorized: “The selected Search Console window recorded impressions for the page but few clicks.” This adds a visibility and selection observation. It still does not prove that the opening copy caused the CTR, because query mix, position, result features and other conditions may matter. The correct action is to compare the page, query and result promise, not to declare a cause.

That is the standard to expect. The recommendation should show enough evidence for a reader to understand the decision and enough limitation for a reader not to overread it.

Practical inspection method

Use this review before approving a recommendation or implementation queue.

1. Circle the target. Identify the URL, template, query group or customer path.
2. Underline the observation. Mark what is directly supported by the report, page or authorized measurement.
3. Check the date and scope. Confirm the reporting window, property, URL set, device or other conditions.
4. Ask why it matters. Connect the finding to page purpose, discovery, clarity, access or a measured business question.
5. Classify the verb. Is the right response to fix, inspect, refresh, consolidate, measure or defer?
6. Check ownership. Identify who can decide and who can verify the output.
7. Define the comparison. State what will be checked immediately and what needs a later search or business measurement window.
8. Read the limitation. Ensure the recommendation does not imply a ranking, traffic, lead, sales or revenue outcome.

If several recommendations depend on one unresolved decision, do not approve them as independent tasks. For example, deciding which of two pages owns an intent may come before rewriting both pages. A clear dependency prevents the site from accumulating competing answers.

What the data does not prove

An SEO recommendation does not prove that its proposed change will improve rankings, traffic, clicks, leads, sales or revenue. A Search Console row does not prove causation. A public audit finding does not reveal every user experience or private customer reason.

It does not prove that a new page is needed because a query exists, that more words are better, or that a higher audit score is the objective. It does not prove that a page should be deleted because one short window has no observed queries. Search data is incomplete and business context matters.

Recommendations also do not replace acceptance criteria. If the action is to improve a page, define what improved means in terms of clarity, scope, or a later measurement. If the action is to investigate, say what evidence would change the decision. Without that, the work can continue indefinitely while the original question disappears.

Checklist

• [ ] Target URL, template, query group or business path is named.
• [ ] Direct observation is separated from interpretation.
• [ ] Date, property, scope and conditions are stated.
• [ ] The reason the finding matters is explained without overclaiming.
• [ ] The action verb is precise: fix, inspect, refresh, consolidate, measure or defer.
• [ ] An owner and access requirement are clear.
• [ ] Immediate and later verification are defined.
• [ ] Dependencies and page ownership are considered.
• [ ] Limitations are attached to the recommendation.
• [ ] No ranking, traffic, lead, sales or revenue guarantee appears.

Optional next step

Read the SEO audit education hub for the wider evidence framework. Use the SEO audit report template for report-field context, or review SEO Audit Services when you need a defined scope and deliverable.

Sources and limits

This article uses dated consumer-facing evidence and clearly labelled teaching examples. Examples are directional, not forecasts or guarantees. No private client information, credentials or internal implementation details are included.

Related

• Search Visibility and Audit Literacy hub
• Average Position vs Ranking: What Search Console Really Shows
• Google Search Console for Small Business Owners

By GSA Growth · Updated 7 November 2026

Dated guidance. No ranking, traffic or revenue guarantee.