When businesses think about being discovered online, the traditional question has often been:
“Which keywords do we need to rank for?”
But AI-driven search introduces a different challenge.
Customers are increasingly asking AI systems to help them make decisions:
- Which software should I choose?
- Which supplier is right for my situation?
- Which service provider understands my specific problem?
- What options should I consider?
The question for businesses is therefore not only:
“Are we visible for a search term?”
It is:
“Does an AI system understand when our business is a relevant solution to a customer’s specific situation?”
A recent Google AI Mode experiment provides an interesting example of why this distinction matters.
The aim was not to prove exactly how AI Mode works internally. We cannot see the processes behind the answers it generates. However, repeated observations can help us form reasonable hypotheses about how businesses can prepare for AI-driven discovery.
A repeated Google AI Mode experiment
The test scenario was:
“I run a small UK marketing agency with eight employees. We need project management software that includes time tracking, allows clients to view project progress, integrates with Xero, and costs no more than £100 per month. Which three options should we shortlist, and why?”
The prompt remained exactly the same across multiple Google AI Mode tests.
However, the recommendations changed.
Across the different runs, AI Mode recommended a range of solutions including:
- Teamwork
- Xero Projects
- Avaza
- ClickUp
- Paymo
- Productive.io
- ProWorkflow
The interesting point was not simply that the answers varied.
The more interesting observation was that different solutions appeared to become more prominent depending on which aspect of the customer’s situation seemed to be emphasised.
For example:
- When Xero integration appeared to be the strongest signal, Xero Projects became prominent.
- When agency workflows and client visibility appeared more important, Teamwork became prominent.
- When affordability and small-team suitability appeared important, Avaza became prominent.
- When flexibility and customisation appeared important, ClickUp became prominent.
- When time tracking and billing appeared important, Paymo became prominent.
The customer scenario had not changed.
The prompt had not changed.
The interpretation of which requirements mattered most appeared to change.
We cannot see the AI’s internal reasoning
It is important not to overstate what this experiment proves.
We cannot say:
- “Google AI Mode gives this requirement a 40% weighting.”
- “This exact factor caused this recommendation.”
- “Using certain words guarantees visibility.”
Those would be assumptions beyond the evidence.
However, we can make a reasonable observation:
AI systems need to understand the relationship between a customer’s needs and a business’s offering before they can consider that business as a possible solution.
This leads to an important question:
How can a business increase the chances that an AI system understands where it fits?
AI systems cannot know what you do unless you explain it
A common mistake businesses make is describing themselves in broad terms.
For example:
“We provide affordable project management software.”
That statement may be true, but it leaves many unanswered questions.
Affordable compared with what?
Affordable for whom?
What type of business?
What problem does it solve?
Now compare:
“We provide project management software for small marketing agencies with teams of 5–20 people. The platform includes time tracking, client project visibility, Xero integration, and monthly plans designed for businesses with limited budgets.”
The second description provides much more context.
It explains:
- who the customer is;
- the situation they are in;
- the problems they need to solve;
- the features that matter;
- the constraints affecting their decision.
Those details create more opportunities for an AI system to recognise a potential match.
The goal is not to write for AI
This does not mean businesses should start creating content designed to manipulate AI systems.
That approach risks repeating mistakes from early SEO, where businesses created content primarily to target search algorithms rather than help customers.
The better approach is simpler:
Explain your business clearly.
A human customer benefits from understanding:
- who you help;
- what problems you solve;
- what makes your approach different;
- what situations you are particularly suited for.
AI systems also need this information if they are going to understand your relevance.
The same principles that make a website clearer for customers can also make it easier for AI systems to interpret.
Start with customer scenarios, not prompts
This experiment also reinforces an important part of AI visibility testing:
The starting point should not be finding random prompts.
The starting point should be understanding your customers.
A business should ask:
Who is our ideal customer?
For example:
- small manufacturing companies;
- independent restaurants;
- marketing agencies;
- nervous dental patients;
- growing professional service firms.
What situations are they facing?
For example:
- “I need an accountant who understands manufacturing.”
- “I need a dentist who can help someone with severe anxiety.”
- “I need software that works with my existing systems.”
What factors influence their decision?
For example:
- price;
- location;
- experience;
- integrations;
- specialist knowledge;
- speed;
- customer support.
Those scenarios can then become the basis for realistic AI visibility tests.
Visibility begins with being understandable
A business cannot expect AI systems to identify it as a solution to problems that it has never clearly connected itself to.
A company that says:
“We provide business services”
has given very little information.
A company that says:
“We help UK manufacturing companies using Xero improve production reporting, stock control, and management information”
has created a much clearer picture of where it fits.
This does not guarantee that AI systems will recommend that company.
There are many other factors involved, including competitors, available information, authority, reputation, and the quality of the underlying sources.
But without clear information about what you do, who you help, and why you are relevant, an AI system has fewer opportunities to understand where you belong.
The future of AI visibility may be less about ranking and more about relevance
Traditional search often focused on winning a position.
AI-driven discovery may require businesses to think differently.
The question may become:
“Have we clearly demonstrated the situations where our business is the right answer?”
The businesses that succeed in AI-driven discovery are unlikely to be those that simply repeat broad claims about their products.
They are more likely to be those that clearly explain their expertise, their customers, and the specific problems they solve.
AI visibility starts with customer understanding.
The prompt is simply the test.