Know what the numbers mean.

What a scan asks, and who answers.

An initial scan asks one category-and-market question through OpenAI and Gemini, with web search enabled. It asks for local options without including your business name. Dashboard checks ask the questions you have chosen across your selected available providers.

API answers and consumer applications can differ in their retrieval, instructions and personalisation. Each observation identifies its collection method; a provider name alone does not make these results interchangeable.

• OpenAI
• Gemini

What the scan score is made of.

The scan score is the rounded average of up to four components on a 0–100 scale. Each measured component has equal weight. Components that could not be scored are left out, so read the coverage beside the score.

Website metadata

HTTPS, page title and meta description: 40, 30 and 30 points.

Local search match

Position in one Places text search: 100 for first place, falling by 12 per place, with a floor of zero.

Sampled AI answers

The percentage of complete grounded answers that name your business.

Metadata feedback

The model’s 0–100 assessment of the captured page metadata.

This scan score is a small diagnostic, not a search-engine score or a full website audit. Each report records the method used; compare reports only when their methods and measured components match.

What MyAly refuses to report.

No estimated prompt-volume bands. Nobody publishes how often a question is asked of an assistant. A “1K–10K searches” band beside an AI prompt is a guess with a number printed on it, and it would be the number people planned their content around. Suggested questions here are editorial examples whose demand has not been measured.

Collection methods stay visible. OpenAI, Claude, Gemini and Grok observations use APIs with web search enabled. Google AI Mode uses a search-results capture service. Neither method reproduces a signed-in user's personalised session. A provider stays unavailable until its collection method is qualified.

No sentiment inferred from a visibility score. Appearing often is not the same as being described well, and one run’s judgement about you says nothing about a competitor named in the same answer. Where sentiment was not measured for a business, it stays unmeasured rather than being derived from its share.

An observation has a scope.

Answers can change with the wording, location, model, available sources and collection time. A mention in one run does not guarantee a mention in another, or a future lead.

A share of answers describes the completed observations in which a business appeared, divided by the observations counted in that sample. Read the questions, coverage and denominator alongside the percentage. It does not describe every question people ask.

Share of voice is your mentions divided by all mentions of tracked businesses in completed answers. Each business counts once per answer; a mention is not necessarily a recommendation.

For example, if two completed answers both mention you, and only one also mentions a tracked competitor, visibility is 100% and share of voice is 2 of 3 mentions, or 67%. Businesses outside your tracked set are not part of that share.

A mention requires an unambiguous full business name. Repeated names count once; a name contained inside another tracked business name does not earn a second mention. Negative mentions still count. Aliases and sentiment are not inferred.

A failed or incomplete check is missing evidence, not proof that a business was absent. MyAly distinguishes available, partial and unavailable measurements. A scan score summarizes the checks that could be scored; it is not a search engine score.

Compare like with like.

Compare results only when the prompt set, collection method, scoring version and sample definition are compatible. Changing the businesses or questions being tracked changes the context. An improvement after a website edit does not, by itself, prove causation.

Follow the source, not just the summary.

Our market guides link local statements to their sources and show when those sources were reviewed. Those dates are our review dates, not the sources’ publication dates. Suggested questions are editorial examples; their search volume has not been measured.

Illustrative dashboard and report examples explain the interface. They are not measured results for the businesses shown or evidence of an endorsement. Use the evidence in your own completed run to assess your business.

Make useful facts easy to find.

Clear content, working links, crawlable pages and accurate structured data help explain a website. They do not guarantee inclusion in an AI answer. Google does not require special AI schema or use llms.txt to improve Search rankings.

Use the platforms’ own guidance to assess recommendations: Google Search’s AI guidance and Bing’s explanation of AI citation measurements.