My content is good. Why doesn't ChatGPT recommend me?

You are winning the human game and losing the machine game. Here is the difference, and how to close it.

AI engines do not recommend the best writing. They recommend the content that most clearly matches the question someone typed. That is why your work can be good, loved, and consistent, and ChatGPT can still name someone else.

You run a service business built on your expertise. A practice, a studio, a consultancy, a creative business where the work is personal and the brand is unmistakably yours. You know your content is strong, because people tell you. They reply, they share, they book. Attention has never been your problem.

Then you ask an AI engine a question your business should own, and it recommends a competitor. Often one whose content you would not be proud of.

Here is the uncomfortable part. The two are not the same skill. Winning attention rewards emotion, story, and reach. Winning a recommendation rewards a clear claim a machine can lift out of your page and repeat. Most creative, voice-led businesses have mastered the first and never knew the second existed.

This is not a call to flatten your voice into beige how-to content. It is a guide to the one place that decides whether the machine can read you at all, and how to fix it without losing a thing you care about.

Why does good content still get skipped by ChatGPT?

Good content gets skipped when the answer is buried. An AI engine reads your page looking for a plain claim it can quote next to the question a buyer asked. If your answer sits three paragraphs into a beautiful story, behind a clever heading, the engine finds nothing to carry and moves on to a page that states the point in the first line.

Think about how you write when you are at your best. You open with a scene. You build to the insight. You let the reader feel it before you name it. That is the craft, and it is exactly why a human stays. It is also why a machine leaves, because the machine does not read to the end. It scans for the cleanest match to the words in the question.

So the gap is not quality. It is shape. Your titles are often metaphors. Your headings carry brand phrases that sound wonderful and that nobody types into a search box. Your best answers are real, but they live inside the narrative where neither a rushed reader nor an engine leads with them.

In our experience scanning small, content-rich service businesses, this is the single most common pattern. The writing is strong. The entry points are missing. The page is built to be read by a person who already trusts you, not found by a stranger who is still looking.

A missing front door is one of several reasons an engine can skip you. For the full diagnostic, across positioning, reviews, and structure, read why ChatGPT doesn't recommend your business. This post stays on the one gap that trips up content-rich businesses most: the shape of the words themselves.

Is this a new rule, or an old one wearing new clothes?

This is an old rule. In 2017, Marcus Sheridan named it: They Ask, You Answer. If your buyers are asking questions, answer them plainly and openly on your own site, and trust will follow. Businesses grew on that idea long before AI engines existed.

Nothing about the principle has changed. The questions simply moved. They used to go into a search bar. Now they also go into ChatGPT, Perplexity, Claude, and Gemini. The machine did not write a new rulebook. It made the old one measurable, because now you can see, question by question, whether your page actually answered.

This matters for you specifically, because creative people are right to be wary of gaming a system. So let me be clear. You are not gaming anything. Naming the question your buyer asks, and answering it in plain words near the top of the page, is the most honest move in marketing. It is hospitality. You are opening the door and telling the person they are in the right place, before you take them on the journey you do so well.

If the link between this and classic SEO is what nags you, we cover it in how AI recommendations differ from SEO. The short version is that the basics are shared, and the new work sits in the answer.

What do our early scans show about getting recommended?

Our early data points to one lever, and it is not volume. It is match. In our first roughly one hundred business scans, when a page's content closely matched the words of the question, an AI engine recommended the business about 70 percent of the time. When the content barely matched, that fell to under 20 percent. How much the business published did not predict the result. How well it matched did.

The same early set showed a second pattern you may recognise in yourself. Businesses were recommended on almost every search that used their name, and on only about half of the searches a stranger types, the plain category questions. Roughly one in four were found by their name alone and named on none of their category searches. Beautiful, known to the people who already know you, and invisible to the ones still shopping.

This is early data from a small sample, and we hold it lightly. It points in the same direction as the strongest outside research. The 2024 Princeton study that named Generative Engine Optimisation tested what actually changes whether an AI engine cites a page. The wins were not louder adjectives. They were citing sources, adding real figures, and writing in a clear, steady voice, each lifting citation rates by up to 40 percent. Plain, specific, liftable. The same thing our scans keep showing.

One more number worth sitting with. Across our scans, ChatGPT was the strictest of the major engines, recommending the business on the smallest share of the questions we tested. The room you most want to win is the hardest room to win. That is not a reason to despair. It is a reason to give it the clean answer it needs.

We are careful not to overclaim from an early sample. We did not find that publishing more often, on its own, helped or hurt citation in any clean way. What we saw again and again was different. On the category questions a stranger types, the engines often named a competitor whose page stated the plain answer, even when that competitor's content was thinner and plainer than yours. The machine was not rewarding better. It was rewarding clearer. That is the whole gap in one line, and it is a gap you can close.

Why is the machine now part of your audience?

The machine is part of your audience because your buyers increasingly arrive with one. Gartner reports that most consumers have already used AI agents for everyday buying tasks like travel and shopping, and predicts that by 2028 AI agents will handle around a fifth of the interactions on storefronts built for humans. The near future is not one assistant for everyone. It is each buyer sending an agent ahead to read, shortlist, and decide before a person ever lands on your page.

Most people are not there yet, and that is the point. The share is small today and climbing. Every piece you publish now is read by people and, more and more, by the agents those people trust to filter the noise.

So treat the engine like a new member of your team, not an enemy at the gate. It is the literal-minded researcher you hired, fast and tireless, who will recommend you to everyone it serves. It has one quirk. It only repeats what it can quote cleanly. Hand it a sentence it can lift, and it will carry your name into rooms you will never stand in. You already shift your tone for a client call, a newsletter, and a quick message. This is one more listener, with one clear preference.

The Front Door Fix: name the door, answer at the threshold, signpost the rooms

The fix is not a rewrite. It is the front door of a single page. Everything a reader loves stays inside. You change only what the engine meets first. We call it the Front Door Fix, and it has three parts.

Name the door. Point the title at the real question your buyer types, not the metaphor you love. The metaphor is not lost. It moves to the subtitle, where it still does its work on the human. A branding coach's post called "Your Brand's North Star" becomes "How do I describe what my business does in one sentence?" with "Your Brand's North Star" sitting underneath as the line that gives it soul.

Answer at the threshold. Lead the first two or three lines with the plain claim, then drop into the story you tell so well. The engine reads the threshold first and finds something to carry. The reader gets the answer, then stays for the journey. Open with "A clear brand message names who you help and the result you create, before it says anything clever," then tell the story of the client who cut her about page from three paragraphs to a single line.

Signpost the rooms. Turn two or three of your headings into the questions people actually ask. A heading that reads as a search query is a heading an engine can match and a reader can scan. "From forgettable to unmistakable" becomes "How do I make my brand memorable?" The section underneath does not change.

Notice what survives. The voice. The story. The brand. The whole experience a human came for. You added one liftable line and pointed your existing craft at the words your buyer uses. That is not selling out. It is translation, and translating feeling into words a particular reader can receive is the thing you already do better than most.

If the plain title ever feels too bare for you, remember the subtitle is still yours. That is where the art now lives, one line lower, doing the same work it always did. Nobody asks you to choose between being found and being yourself. The front door is a small surface. Everything behind it stays exactly as you built it.

How do you rewrite one post this week without losing your voice?

You do not touch the whole blog. You run one experiment on one page, and you let the result decide the rest.

Step 1. This week, pick one post you are proud of that should be winning a question and is not. Just one.

Step 2. Before you change a word, write down the exact question a buyer would type to find it. Plain words, the way they would say it to a friend.

Step 3. Rewrite the title as that question. Move your original title to the subtitle so the voice stays.

Step 4. Add two or three plain sentences at the very top that answer the question outright, then let the existing story run underneath, untouched.

Step 5. Reshape two headings into questions, publish, then re-scan the page across the engines and watch whether the recommendation moves.

When you see one post shift from skipped to named, you will trust the pattern, and you can decide which page gets the same small treatment next. One door at a time.

The bottom line

Your content is not the problem. Its front door is. You have spent years earning the human game, and the machine game asks for one honest change, not a new personality.

Run a free AI visibility scan and see, question by question, where the engines name you and where they name someone else. It takes about five minutes, no credit card. If you want to know what the scan looks at before you run it, see our common scan questions.