How to Read Your AI Visibility Scan—and Decide What to Do Next

What a free scan should help you decide

An audit is useful when it changes your next decision. That could mean repairing an inaccessible service page, investigating a weak Arabic customer journey or checking why an answer describes your company incorrectly. A coloured score without supporting evidence does little to help.

Mayar Vision's Free Scan is designed to combine a bounded website review with sampled AI-answer observations and available supporting signals. It normally presents five to seven useful priorities when the evidence supports them. Fewer supported findings are preferable to padding the report with speculative faults.

The example report uses fictional data to demonstrate the format. It is a teaching example, not a scan of a real customer or proof of live provider coverage. The interpretation principles below apply to a real report too: look at its actual coverage and evidence rather than assuming every possible check ran.

Check the scope before reading the headline

Confirm that the report refers to the right business and website. Similar trading names, multiple domains or an old company address can distort the starting point. A report about the wrong entity cannot become useful through a more attractive dashboard.

Next, look at the market and languages. A Saudi-focused report should state the location or market context used, the language of each question and the platform observed. Arabic wording is not, by itself, proof of a Saudi location setting. A city included in a question and a provider's configured location are also different pieces of context.

Check when the observations were collected. AI answers can change, and website content can be updated between runs. A report should preserve enough context to let a reviewer understand what was actually seen.

Finally, inspect the questions. Branded questions contain your company name; unbranded questions describe a need without naming you. “What does this company do?” tests something different from “Which firms provide this service in Jeddah?” Keep those groups separate when evaluating discovery. Otherwise easy name-based observations can make customer discovery look stronger than the sample supports.

Understand cited, named and absent

In the scan's summary, an answer is classified as cited when it cites your owned website. Named means the business was mentioned without an owned-site citation. Absent means neither was observed in a successfully measured answer. These summary categories are mutually exclusive; the source detail provides the fuller picture.

That last point matters. An answer may cite your website without naming the business in its prose. It may name your business while citing an external publisher. Open the actual answer and source links to understand what the classification represents.

Swipe across the table to see all columns.

An absent result is not evidence that every buyer will miss you. Equally, one citation is not proof of broad market leadership. Both are observations within a particular sample. Our authority guide explains why links, mentions and citations deserve separate treatment.

Read the denominator as carefully as the percentage

Consider a fictional teaching sample of ten scheduled tests on one platform. Eight return usable answers: three cite the website, two name the business without citing it, and three show neither. Two tests are unavailable.

The owned-site citation rate among measured answers is 3 out of 8, or 37.5%. Coverage is 8 out of 10 scheduled tests. Both facts belong together. Reporting only 37.5% hides the missing observations; reporting two unavailable tests as “absent” would claim knowledge the scan does not have.

This small example is not a Saudi benchmark, a customer result or the data in the demonstration report. It simply shows why the denominator matters. One additional citation in a measured sample of eight would move the rate by 12.5 percentage points. Large-looking changes can arise from a very small number of answers.

Keep platform and language results visible. If Arabic and English were tested with different questions or unequal coverage, explain that before comparing them. Repeated trials are useful for studying variation, but they should not quietly inflate the number of distinct buying questions tested.

Read coverage as a map of what remains unknown

The scan's broader framework covers 21 dimensions, including technical access, content, language, external signals, customer journeys and measurement. That does not mean a free automated run completes a full expert audit across all 21.

Some dimensions are sampled publicly. Others need owner access to Search Console or analytics, a specialist's judgement, or information that the scan did not collect. A “not assessed” label preserves that boundary. It should not be turned into a zero score or treated as a fault on your site.

For example, a public scan can inspect a visible service page. It cannot reliably tell whether your team received last month's enquiries or whether those enquiries became customers. A native editorial review is also different from a machine checking whether Arabic text exists.

Use these gaps to choose the next investigation. If the question is about real search impressions, owner reports matter. If it is about Arabic service terminology, a language and subject review matters. Buying more automated checks will not necessarily answer either question.

Turn each finding into a testable action

A strong finding has five parts: what was observed, where it was observed, why it may matter, what to do next and how to verify the result. Read all five before accepting the priority label.

Imagine a report identifies an indexing restriction on an important service page. First confirm the restriction is unintended. Then have the responsible person correct it and check the page again. The immediate acceptance check is that the instruction is correct; later search inclusion is a separate observation. Google's technical requirements explain why eligibility and indexing are not the same promise.

A performance finding needs similar care. A laboratory test can identify a slow loading element under its test conditions. It does not establish every visitor's experience. Compare like-for-like retests and review field data where available; lab and field measurements answer related but different questions.

For a backlink finding, inspect the actual referring sources. A low count in one provider's index is not sufficient evidence of poor credibility. For an Arabic finding, read the affected page and journey. An automated flag should lead to the relevant check, not substitute for it.

Choose the next step from the problem

Swipe across the table to see all columns.

A larger audit is valuable when it resolves uncertainty or changes the investment decision. It should not merely repeat the free findings in a longer document. Ask what additional access, sampling, specialist review or competitive investigation it includes.

For the same reason, a retainer is not the automatic next step after a free scan. A one-off repair may be enough. Ongoing work becomes useful when there is an agreed backlog of improvements, a need to maintain information or a measurement programme that supports decisions. See what monthly work should deliver.

Keep the report useful beyond the first conversation

Save the dated report and its scope. Write down the few actions you decide to pursue, who owns them and when they will be checked. Preserve the questions if you intend to compare future samples.

When reviewing a later scan, first compare coverage and conditions. A result can change because the website improved, the question changed, a provider returned a different answer or the earlier test was incomplete. Documenting those differences is part of analysis, not an excuse to avoid conclusions.

You can use the report yourself, share it with your website team or discuss it with Mayar Vision. The intended journey is to read the results first, with optional email delivery afterward. A WhatsApp conversation should begin with the question you want answered; it need not require handing over every detail of your business immediately.

Start with the Free Scan, explore the clearly labelled example, or compare the service scopes. The right next step should be understandable from the evidence in front of you.

Checked 26 September 2026. Report definitions and service descriptions follow Mayar Vision's current draft product design. The numerical chart is a fictional teaching example, not measured performance. Production provider coverage must be established by the actual report.