INSIGHTS / AI VISIBILITY
What AI visibility can actually tell you.
You ask an AI system to recommend a business like yours. A competitor appears. Your brand does not. It is tempting to treat the answer as a verdict.
Before deciding what it means, look at what you actually tested. Which system did you use? What did you ask? Did you name your brand, specify a location or provide earlier context? Did the answer cite sources? What did those sources say?
An AI visibility test becomes useful when someone else can understand the conditions behind the result. A screenshot alone leaves too much unexplained.
Start with a question your customer might ask.
A direct question about your company tests something different from an open request for a recommendation. The first can help you inspect how your business is described. The second can help you observe which businesses appear for a particular need.
Keep those tests separate. If you insert your own brand into every prompt, your report should not suggest that customers would encounter it without being prompted.
Choose questions connected to real decisions: what you offer, where you work, who a service is suitable for, or which providers fit a particular brief. Record the intended audience and market so the exercise has a clear purpose.
Keep the evidence attached to the answer.
Our recommended observation record contains enough detail to distinguish what happened from what we think it might mean.
| Record | Why it matters |
|---|---|
| System and available model or mode label | Identifies the environment tested without guessing hidden settings. |
| Exact prompt and relevant conversation context | Shows whether the brand, location or preferred answer was supplied. |
| Date, time, language and market | Gives the observation a defined context for later comparison. |
| Whether browsing or search was enabled, where visible | Separates known test conditions from assumptions about retrieval. |
| Full answer and visible source links | Preserves what was said and which evidence was shown. |
| Presence and factual accuracy | Distinguishes being mentioned from being described correctly. |
| Test count and repeated runs | Makes the size of the observation set clear. |
If a setting is not visible, mark it as unknown. If a response has no citation, record that absence. Do not fill the gap with a source the system did not display.
A mention, a citation and an accurate description are different findings.
A brand can be mentioned with an incorrect service description. A source can be cited without supporting the sentence next to it. An answer can describe the business accurately without recommending it.
Each deserves its own check. Open the cited page and compare the relevant claim with what is actually there. Check descriptions against approved business information. Keep the saved answer beside the finding so a reviewer can follow the reasoning.
If you report a mention rate, state the denominator: how many answers, across which prompts and systems, in which test period. A percentage from a defined sample is an observation about that sample. It does not establish your share of every AI conversation or a universal ranking.
Turn the finding into a question you can act on.
Suppose an answer describes a service you no longer offer. Check the sources it cites and your public pages for that language. If the cited information is outdated, you have a specific correction to investigate. If you cannot find the origin, preserve that uncertainty instead of inventing a cause.
For Google's AI Overviews and AI Mode, Google's own guidance (Opens an external website) says established SEO practices remain relevant. Pages need to be indexed and eligible to appear with a search snippet to qualify as supporting links; meeting the requirements does not guarantee inclusion. This guidance concerns Google's features and should not be presented as a specification for every AI system.
Our practical view is to begin with information you can inspect and improve: clear service descriptions, accessible priority pages and evidence that supports your claims. Choose work against a documented gap, then record what changed. A later answer is another observation, not automatic proof that your edit caused it.
Keep visibility and business response in view.
An appearance in an answer does not, by itself, tell you whether anyone visited, enquired or bought. Where reliable referral and conversion data is available, examine it alongside the visibility observations and be explicit about what cannot be attributed.
The result should help decide the next piece of work. Sometimes that is correcting public information. Sometimes it is fixing a website foundation. Sometimes the evidence calls for more investigation.
Can Morph guarantee an AI recommendation?
No. We can examine defined tests and address agreed gaps in the information and website foundations within scope. Third-party systems determine their own answers.
Where do Pulse and Search Readiness fit?
Morph Pulse (Opens an external website) is the separate AI visibility product pathway. Search Readiness is a defined website-foundation offering, with eligibility and implementation scoped separately. Both connect to Digital & Performance.
Bring the answer and the question behind it.
If a result concerns you, share the prompt, the answer and any visible sources. That gives us something specific to examine.
Bring us the brief
Explore Morph Pulse (Opens an external website)
More perspectives