Your Made-Up Customer Is Why Nobody Buys Your Digital Product

I once spent a whole afternoon building a picture of my ideal customer. Name, age, job, income, the coffee she drinks, her Sunday-night worries. It felt like real work. It was worth nothing, because I had made her up. The people who will actually pay for your digital product are out there tonight, typing their problem in their own words. There is a way to go and read them before you build a single thing. Learning that took me far longer than it should have.

The desk exercise that feels like progress

Here is the before-state, and I suspect you know it. You have an idea for a guide, a course, a small service. You go looking for advice on who to sell it to. Every article says the same thing: define your ideal customer. Give her a name. List her demographics, her interests, her pain points. Then pick a niche and narrow it until it feels specific.

So you open a document and start typing. Thirty-five to forty-five. Works full time. Wants extra income. Feels overwhelmed by technology. It looks like a plan. You feel organised. You close the laptop believing you have done the hard part.

Then you go and build the product, and it lands on nobody.

The problem with a person you invented

The made-up buyer has one fatal flaw: there is no evidence in it. Every line came out of your own head, which means one of two things happened. Either you described yourself, or you described a person who does not exist.

That matters more than it sounds, because everything downstream inherits the guess. The title of your ebook comes from the guess. The bullet points on your sales page come from the guess. The videos you make to attract people come from the guess. If the guess is wrong, you do not find out for three months, and when you do, you cannot tell which part was wrong.

The other advice on offer is worse. Run a survey. Hold a focus group. Check your analytics. All three assume you already have an audience. If you had an audience you would not be reading this. This is the gap in almost everything written about building extra income with AI: the advice starts at step five and quietly skips the part where you have nobody.

The honest position is this. You are not short of imagination. You are short of evidence.

You do not invent an audience, you find one already talking

Here is the shift, and it is the whole method.

Somewhere online, right now, people are describing your buyer's problem better than you ever could, for free, unprompted, in public. They are not being polite about it. They are not filling in a survey. They are frustrated at 11pm and typing it out. Your job is not to imagine those people. Your job is to go and read them, collect their exact words, and let those words decide the product, the title and the copy.

This is the People step. In the method I teach it runs Problem, People, Product, Pull, and People sits second on purpose. People come before product, always. Build after you have read, not before.

Where to actually go

These are the places I use, in rough order of how much honest language they give back per hour spent.

Reddit. The single richest source, because people post there to be understood, not to be seen. Search your problem area on Reddit search and sort by top from the past year. Read the posts, then read the replies underneath, which is where the real specifics live. Look for the threads with lots of comments and no accepted answer. That is an unmet need with a queue of people standing in it.

Reviews of the things they already bought. Find the three best-selling books or courses in your area and read the two and three star reviews. Five stars tell you nothing. Two and three stars tell you exactly what the buyer wanted, paid for, and did not get. That gap is your product.

Comments under the top videos. Search your topic on YouTube and TikTok, open the videos with the biggest view counts, and read the top comments plus the replies. Pay special attention to comments that start with "but what about" or "this doesn't work if". Those are the objections your future sales page has to answer.

Facebook and Skool groups. Join two or three groups where your buyer already gathers, and use the group search rather than the feed. Search a keyword like "stuck" or "help" and read what comes back.

What people type into Google. Start typing your topic into the search bar and read the autocomplete suggestions, then scroll to the People Also Ask box on the results page. Those are literal questions, in the buyer's grammar, at real volume.

Two hours across three of those sources is enough to start. You are not doing research for a dissertation. You are collecting raw language.

What to collect (and what to ignore)

Open one document. Do not summarise, do not tidy, do not paraphrase. Copy and paste the sentences whole, exactly as written, typos included. Paraphrasing is where your assumptions sneak back in.

Collect four things:

1. The complaint. What is not working, in their words.
2. The failed attempt. What they already tried and why it did not stick.
3. The wish. Any sentence starting with "I just want" or "if only there was".
4. The identity line. How they describe themselves. "I'm not techy." "I'm a mum with two hours a night."

Ignore anything that is someone selling. Ignore the advice replies. You want the people with the problem, not the people answering it.

Aim for something like a hundred pasted quotes. It sounds like a lot. It takes an evening.

Let AI find the pattern, not the person

This is where AI earns its place, and it is not the vague "analyse your data" you see everywhere. You are not asking it to invent a customer. You have banned invention. You are asking it to read two hundred real complaints faster than you can and hand back the repeating shapes, in the original language.

Paste your document into Claude or whichever assistant you use, and run this:

Below are real, unedited quotes I collected from people talking
about [topic] on Reddit, in reviews and in video comments.

Do not invent, summarise generously, or add anything that is not
in the text. Work only from these words.

1. Group the quotes into recurring problems. Name each group using
   a phrase taken directly from the quotes, not marketing language.
2. For each group, tell me how many quotes fall into it and paste
   the three most vivid verbatim examples.
3. List the exact words and phrases people repeat most often when
   describing the problem, and separately when describing what they
   want instead.
4. List every solution they say they already tried, and the reason
   they gave for it not working.
5. Tell me which group has the most emotion in the language, and
   quote the line that shows it.

[paste quotes here]

What comes back is not a persona. It is a map of a real conversation, ranked. The biggest cluster with the most emotional language, where the failed attempts pile up, is your product. The phrases in point three become your title and your headline, because they are already the words the buyer thinks in. You are not translating your idea into their language. You are just using their language.

Run it once, then read the source quotes for the top cluster yourself. The AI finds the pattern quickly. You still need to feel it.

The test that tells you these are real people

There is one check I run before I build anything, and it is embarrassingly simple.

Can you name three specific places, right now, where at least twenty of these people are gathered, and can you paste a real quote from each?

If yes, you have found an audience. They exist, you know where they are, and you know how they speak, which means you already know where to show up and what to say when you get there. If you cannot do that, what you have is still a fantasy with demographics attached, and no amount of narrowing will fix it.

The second check: does the problem show up in more than one place? A pain that appears on Reddit, in the reviews and in the comments is a real pain. A pain that appears in one thread is one person having a bad week.

This works for you whether or not you have a following, a list, or any traffic at all, because none of it depends on you. It depends on other people having already spoken. That is why I like it. It is the one piece of market research an absolute beginner can do as well as a professional, on a Tuesday night, for free.

You do not invent an audience. You find one already talking.

If you want the full picture of how the People step fits with finding the problem, building the product with AI and getting the right people to come to you, I have put the whole thing in a free five-step guide. It is the same sequence I use myself, written plainly, no fluff. You can read it here: The Practical Income Method.