Who Does AI Say Your Competitors Are?

You have a list of competitors in your head right now. The same names that have been on the pitch deck for years.

Here is the problem. The AI answering your buyers has its own list. And I will bet it does not match yours.

Hi! I am Bob Michaels, and this is Evolving the Web. Every two weeks I talk through the essays I just published, in plain language, as one connected argument. These two weeks were one continuous argument, four essays, one idea. So let me make it.

Strip away all the noise, and success in business is simple. More customers than your competition. That is the whole game, and it has not changed.

What changed is where the choosing happens.

Buyers used to search, click around, and compare on their own. More and more, they now ask an AI. Who is the best option for this. Who should we hire for that. And the AI does not hand back ten blue links. It hands back a decision. A short list, a pick, and the reasons.

A recommendation is a delivered decision. Many buyers never look past it.

So here is the metric I care about above everything else in this field, and the first essay makes this case as bluntly as I can make it. When an AI compares you, head to head, against the competitors it chooses to name, who wins? And why?

Everything else people measure, mentions, rankings, coverage, those are instruments on the dashboard. The head to head is the destination. It is the one measurement shaped like the question your business actually asks. Are we winning the choosing, or losing it?

The second essay is about why the standard playbook cannot answer that question.

You know the playbook. Mine the keyword tools, harvest the questions people ask, publish posts on every topic in the category, watch the traffic. I call it topic chasing. And I want to be fair. It is real work, and coverage does help. Coverage is fuel.

But it has a ceiling, and the ceiling is this. Topic chasing measures what you put in. Posts published, keywords covered, traffic earned. It never measures what comes out the other end, which is the recommendation. You can rank beautifully and still lose the choosing. In one study of software companies, nearly half the brands sitting on page one of Google got zero mentions from ChatGPT on matching questions.

Ranking and being recommended are two different games, played on the same field.

So keep the content program. Change the finish line. The finish line is a locked series of head to head questions, run on a schedule, recording who wins, who gets named, and why.

The third essay is the one I would hand a stranger first, because it starts with the question almost nobody can answer. Do you actually know who your competitors are, according to AI?

Notice the phrasing. According to AI. Because when a model answers a buyer, it names the comparison set right inside the answer. These are the options. Here is the pick. That set is your real competitive field, and it is discovered, not assumed. Nobody asks your permission before naming your rivals.

And the discovered field can be genuinely strange. In one snapshot study, eighty one percent of the brands ChatGPT recommended did not rank in Google's top ten for the same keyword. The AI shortlist is not your search shortlist wearing a costume. It is a different list.

Finding yours is not hard. Ask the five major platforms the three kinds of questions a buyer would ask. Best options in the category. Who to hire for a specific job. This company versus that one. Write down every name that comes back, and how often. The names that keep recurring are your field. The names that show up once are noise.

While you are in there, check one more thing. Make each platform describe your company, and read it closely. AI sometimes blends two similarly named companies into one, and you inherit the other company's industry, their reviews, sometimes their failures. If that is happening, no other work matters until it is fixed.

Which brings us to the fourth essay. Say you run the questions and you are losing. What moves the recommendation needle?

There is a loop, and it has four beats. Ask, decode, build, verify.

Ask, you already know. The locked questions, on a cadence.

Decode means reading the answers that beat you like a brief. Not to sulk, to learn. What themes do the winners get praised for? What evidence gets quoted? Which sources does the model lean on? That is the machine telling you exactly what it finds convincing. It is the brief for what you build next.

Build means putting that evidence on your own domain first. Plain pages that state what you do, for whom, with what proof, in language a machine can lift and quote. In one large study, corporate websites took roughly three quarters of the citations in AI answers. Your own site is still the main stage. Then make sure your site, your social presence, and your machine layer all tell the same sharp story, because the comparison engine reads all of it as one witness.

And verify means running the same locked questions again, and taking the answer honestly. I will give you both of my endpoints. One program I ran moved from winning about six percent of its head to heads to winning about thirty six percent, in eleven weeks. Another program I ran stayed flat. Both results are real, and the flat one is exactly why the loop retests instead of declaring victory.

Anyone who guarantees you the first result is selling. The honest offer is the loop.

The two week arc in one breath.

The business question is the head to head, who wins the choosing. The old playbook measures inputs and cannot answer it. The competitor list is discovered by asking, never assumed. And when you are losing, the loop is ask, decode, build on your own domain, verify, and repeat.

All four essays are on the blog at bobmichaels dot ai, and the episode page links each one.

Running that first head to head is the first engagement I do with every client. Real buyer questions, five platforms, and you see exactly who the machines recommend in your category and why. Consultant or fractional, a slice of my week for as long as you need it.

Thanks for listening. And go ask an AI the one question you are afraid of. Us, or them.