How to Give an AI Writer Your Own Product Opinions (So the Guide Sounds Like You Tested It)
A generated buyer guide knows everything the listings say and nothing the listings don't. It can tell you the lumens and the battery hours; it can't tell you the mount cracks in the cold, or that the cheaper model is the one you'd buy. That second kind of knowledge is what makes a guide worth reading, and it's the one thing the tool has no way to get except from you.
This post covers how to add personal experience to AI content the honest way: a short notes file per topic, what goes in it, where it feeds in, what changes in the guide when it does, and the line between your opinion and an invented test. It follows the E-E-A-T post, which covers what Google means by experience; this is how to get yours onto the page.
The short answer
Before a batch, write ten lines of notes on the topic: the product you'd buy, the one that's overrated, the spec that matters and the one that doesn't, the dealbreaker, the thing people get wrong, and what you've used yourself. Put them in the tool's brief or notes field.
The guide's verdict, product sections and "how we picked" now carry a point of view. Say where the knowledge comes from, and never let the tool claim a test that didn't happen.
What the tool can't know
Everything on a listing is public and every generator has it: specs, price band, review counts, the common complaints. A guide written from only that reads like every other guide written from only that, because it is.
Google's reviews guidance asks for evidence of experience and an opinion the reader can act on; readers ask the same thing less formally. Both are asking for what's in your head and nowhere else.
The notes file: ten lines per topic
| Line | Example (bike lights) | Where it shows up |
|---|---|---|
| What I'd buy with my own money, and why | The 800-lumen Cygolite; the mount survives winter, the others crack | Verdict box, top pick section |
| What's overrated | The 2,000-lumen ones: too bright for roads, battery dies in 90 minutes | A product section's caveat, or a "what to skip" line |
| The spec that matters | Beam shape, not lumens; a cut-off beam doesn't blind drivers | "How we picked", the comparison table columns |
| The spec that's marketing | Modes; nobody uses more than two | Intro, "how we picked" |
| The dealbreaker | No USB-C in 2026 is a no | Selection filter, caveats |
| What people get wrong | Buying one light; you need a rear one more | Intro, FAQ |
| What I've used, for how long | Cygolite Metro for two winters; a Knog for one summer | Experience sentence in those product sections |
| Who this isn't for | Mountain bikers: different lights, different guide | Intro |
| Price reality | Under $30 is fine for a commuter; $100+ is for trail riding | Verdict, budget pick |
| Source of the knowledge | Daily commuter, 8 years; haven't used the brand-new models | "How we picked", author note |
Ten minutes, once per topic, reused across every guide in it. If you can't fill five of the lines, that's a signal about the topic, not the method; an opinion is one of the five tests for choosing a topic in the first place.
Where it feeds in
Into the brief or notes field, not the keyword and not a prompt that says "sound experienced". The notes are the material; the tool's job is to place them.
Good tools weave them into three spots: the verdict (your pick, your reason), the product sections (the caveat for the overrated one, the experience sentence for the ones you've used), and the "how we picked" section (the spec that matters, the dealbreaker, the source of the knowledge).
The brief post covers the other inputs; the notes are the one that makes the guide yours.
What changes in the guide
- The verdict has a reason. "Our pick is the Metro because the mount survives winter" instead of "Our pick is the Metro, with 800 lumens and a 4-hour battery."
- One product gets told off. A guide that says the 2,000-lumen light is wrong for roads is a guide with judgment; a guide that praises all ten is a listing.
- "How we picked" explains a choice. Beam shape over lumens is a decision the reader can disagree with, which is the point.
- The table's columns change. Beam type and mount material replace the modes column nobody reads.
- Experience sentences appear where they're true, and only there. Two products out of ten get "I've run this for two winters"; the other eight get the review-pattern caveat and no pretence.
The line not to cross
Your opinion, stated as yours, is the asset. A claimed test that didn't happen is the liability, and the tool will produce one if the notes invite it ("make it sound like we tested everything").
Keep to three rules: the experience sentences cover only products you've used; the rest of the guide says what it's based on ("from the specs and what buyers report"); and no sentence appears that you couldn't defend to a reader who emailed to ask.
The Amazon rules post covers why the invented test is also a program violation, not just a credibility problem.
Doing it with AffSERP
AffSERP's X Best Review tool takes a notes field per site and per batch, and it's built to place the ten lines above: your pick and reason in the verdict, the overrated product's caveat in its section, the spec that matters in "how we picked", and an experience sentence only on the products your notes say you've used.
Everything else is written from the specs and the review pattern, and says so.
Write the notes once per topic and every guide in the batch carries them; the fifteen-minute editing pass adds the rest.
FAQ
How do I make AI content sound like first-hand experience?
Give the tool first-hand material. Ten lines of your own notes on the topic (which product you'd buy, which is overrated, the thing that breaks, the question people get wrong) turn a generic guide into one with a point of view, because the tool now has something to say that the listings don't. Without the notes it can only rephrase the specs; with them it can take a position.
Is it OK to add 'in my experience' to AI-written reviews?
If it's your experience, yes, and it's the best thing you can add. If it's invented, no: it's a false claim under the FTC's endorsement rules and Amazon's accuracy rule, and it reads as fake to anyone who's used the product. The test is whether you could defend the sentence if a reader emailed to ask. 'I've run the 800-lumen version for two winters' passes; 'we tested twelve models' when nobody did fails.
What if I haven't used any of the products?
You still have opinions worth more than the listing: which spec actually matters for the reader, which feature is marketing, what the one-star reviews agree on, what you'd buy with your own money and why. Write those down and say where they come from ('from the review pattern', 'from using the previous model'). A guide written from honest judgment beats one written from invented testing.
Where do I put my notes so the AI uses them?
In the brief or notes field the tool offers per topic or per site, not in the keyword. Keep it short and specific: product names, the opinion, the reason. The tool weaves them into the verdict, the product sections and the 'how we picked' section; a sentence that says 'use my notes' does nothing on its own.
Keep reading
