The Companies AI Will Recommend

A company can be excellent at its work and still give an AI system very little useful information about why someone should hire it.

Its strongest projects may live in private presentations. The people with the most experience may barely appear on the website. Customers know what the company does well, but their reviews say little beyond “great service.”

That leaves a gap between the business’s capabilities and the evidence someone can find before speaking to it. When a buyer asks AI for a shortlist, that gap becomes consequential.

This series began with AI Is Becoming the Intelligence Layer Above the Internet, which explored how some of the research and comparison people once did themselves is moving into an intermediary that assembles an answer for them. The second essay, AI Doesn’t See Your Brand the Way You Do, and Why That Matters for GEO, examined the difference between the organization you know from the inside and the picture available through its public information.

Why GEO Is Bigger Than SEO for AI then asked what happens after information is found: what can a system reasonably conclude about the company behind it?

For this final installment, I want to bring that question closer to an actual buying decision.

Generative Engine Optimization (GEO) means improving the information available about a business so AI systems can find it and represent its capabilities more accurately.

Start with the buyer’s problem

Imagine an operations director arranging an office move for 140 employees. The move has to happen over a weekend. Equipment needs to arrive at the correct desks, and everyone needs to be working on Monday.

“Who are the best moving companies?” barely describes the job.

The buyer needs a company that can plan around building access, coordinate equipment, assign responsibilities, and handle problems without throwing the whole schedule off. A useful recommendation should account for those requirements.

Now consider a moving company that does this work regularly. Its website says it offers commercial moving, but most of the content discusses residential moves. Its photographs show houses. Its testimonials praise friendly crews. There is no explanation of how it plans an office relocation.

The company may be a strong candidate. The available information makes that difficult to assess.

I would start there. Describe the jobs the business wants to win and the conditions that make those jobs difficult. Then look at whether the website and other public sources give a prospective buyer enough information to judge its suitability.

That exercise often exposes a positioning problem before it exposes a content problem. If leadership cannot agree on which work the company is particularly good at, the website will struggle to explain it.

Show the work behind the claim

“Experienced commercial movers” is a claim. A detailed account of a completed office relocation gives someone a basis for evaluating it.

A useful case study would explain the size of the move, the time available, and what the company was responsible for. It would describe the planning decisions and report what happened. If the client confirmed the outcome, include that confirmation with enough context to understand it.

You do not need a dramatic success story. You need an account that helps another buyer recognize relevant experience.

This is where many case studies fall short. They announce that a project was successful while leaving out almost everything that would help someone assess whether the company could handle their own job.

The same problem appears in expert content. An article about “five benefits of hiring professional movers” tells an operations director little they could not work out themselves. An explanation of what must be settled before a weekend office move could be far more useful.

Ask the person who plans those jobs. What does the client usually overlook? What changes the estimate? Which decisions become expensive when they are left until Friday afternoon?

Those answers are worth publishing.

Google’s content guidance encourages original information, evidence of experience, clear sourcing, and material that helps its intended audience accomplish a goal. Those are sensible standards for evaluating the substance of a page, regardless of where someone encounters it.

Evidence outside the company’s website can add another perspective. A client describing how a difficult project was handled supplies information the company’s own account may not. A relevant industry article or professional profile can help establish the context around a person’s expertise.

Before pursuing another mention, ask what it would help a buyer establish. The answer should be more specific than “it gets our name out there.”

Make the company’s history easy to follow

Businesses change. Their online information often catches up unevenly.

A company expands into a new service, changes its name, or acquires another business. The website gets updated, while an old directory entry or executive biography continues to describe an earlier version.

Some differences are perfectly reasonable. A ten-year-old article should describe the business as it existed ten years ago. The current website should help someone understand what changed.

Part II introduced Signal Architecture as a way of thinking about how these pieces relate to one another. In practice, that includes explaining the connections between names, people, services, and locations so the company’s history is easier to follow.

Correct profiles you control. Explain a name change where it matters. Make current services and operating locations clear. Give experienced staff biographies that accurately describe their responsibilities and background.

Technical information can help too. Google documents how organization structured data can clarify administrative details and distinguish a business from similarly named organizations. It should communicate accurate facts that agree with the visible information on the site.

The website remains an important place to make those facts understandable. It should help someone who knows nothing about the business get their bearings.

Give the work an owner

This work crosses departments.

Sales knows which questions prospects ask before committing. Operations knows what delivering the service involves. Leadership decides which markets and capabilities deserve investment. Marketing turns some of that knowledge into material people can find and use.

Someone needs to bring those contributions together.

For the moving company, the objective might be to establish its suitability for office relocations with tight schedules. That would guide the projects selected for case studies, the questions covered on service pages, and the expertise described in team biographies.

A practical first assignment could be small: interview the project manager, document one relevant job, revise the commercial moving page, and correct the profiles that describe the business inaccurately.

Then assess what still needs work.

That is how I would approach the problem at Darby. Start with the buying decision and examine the information available to support it. The next task should follow from what is missing or unclear.

There is little value in commissioning a large publishing program before understanding where the evidence falls short.

Check whether it is bringing in business

Better information does not give a company control over an AI response.

Google states that its AI search features use established SEO practices, require no special AI markup, and offer no guarantee of inclusion. Accessibility and useful content remain the foundations within that environment.

An AI answer can also sound certain and be wrong. We need to judge the answers we observe rather than assume the systems are carrying out a reliable assessment of every provider.

Choose a small set of questions that reflect the work you want to win. Record how your company appears, which competitors are included, and which sources are cited. Repeat those observations over time.

Look for patterns before deciding what to change. A single omission gives you very little information about its cause.

As Part III discussed, a mention, a citation, and a recommendation are different outcomes. Your article might supply information for an answer without your business being suggested as a provider. Keep those distinctions visible when reporting progress.

Then bring the evaluation back to sales.

Are inquiries becoming more relevant to the work you want? Do prospects understand your capabilities before the first conversation? Are they asking about a project or an explanation they encountered during their research?

Ask what influenced them to get in touch. You will not capture the entire journey, but the answers can tell you whether the information you are publishing is helping people make a decision.

AI visibility is an indicator. Qualified opportunities and sales are what make the effort worthwhile.

The larger shift explored in this series is that buyers can now encounter an interpretation of your business before they encounter the business itself. Some of their research may already have been summarized, and some options may already have been narrowed.

My expectation is that companies with specific, accessible evidence of their capabilities will be better placed to contribute to those decisions. That is the practical case for doing this work, even as the systems change.

When the operations director asks who can get 140 employees moved and working again by Monday, your digital presence should give them a reason to consider you, and enough detail to decide whether that reason holds up.

Explore how Darby evaluates your company’s AI visibility. →