Why this signal matters for Google
A real product review is a test you ran yourself and that a reader could run again. It is not a rewrite of the maker's spec sheet. It is not a roundup of what other people said. You had the products in your hands, you used them under the same conditions, you wrote down what happened, and you say which one you would pick. A bookkeeper who enters the same twenty invoices into three invoicing tools and times each run has done a test. A site that copies the feature table for those three tools has made a catalogue. So learning how to write a product review comes down to three moves: write the protocol (what you will measure, and how) before you start, compare several products instead of one, and end with a clear call. Your own product photo proves you held the thing. The test proves what you did with it.
Google tells its raters to ask the same question a buyer would. These are people who score pages by hand, following instructions Google makes available to anyone: the Quality Rater Guidelines, usually shortened to QRG. In section 3.4, experience is explained with a very simple example: which would you trust, a review from someone who has used the product, or one from someone who has not? Section 7.3 adds that this kind of experience needs no degree: an ordinary person who really tried something can produce a high quality page. Section 5.1 describes the opposite case, the restaurant review written by someone who never ate there. That is what your protocol is for. It puts the proof that you tried right in front of the rater.
Google also spells out what it expects from a review. Its guidance on how to write high quality reviews is aimed at bloggers, merchants and editorial teams. It asks for quantitative measurements, evidence of your own experience (visuals, audio, links) and an explanation of what sets a product apart from its competitors. "Great battery life" does not clear that bar. "11 hours 20 minutes of video playback at 50% brightness, against 8 hours 45 for model B" does. The same page notes that reviews often use affiliate links, meaning links that pay you a commission when the reader buys. Nothing wrong with that, on one condition: say so. A free sample or a hidden commission is a conflict of interest, and a reader who finds out alone stops believing any of your numbers. When I audit small business websites, the "comparison" page is nearly always a table lifted from the makers' spec sheets: no photo taken in-house, no test date, no figure measured on site.
On health and money, your review has a line it should not cross. The QRG label these subjects YMYL: the ones where a mistake can cost the reader dearly. Section 3.4.1 separates two things there. Describing your trial is experience: you used three blood pressure monitors for a month and you say which is easiest to read. Saying which reading should worry someone is expertise, and that call is not yours to make unless you are a clinician. AI tools sort content in a similar way. Ask ChatGPT, Claude or Perplexity "which payroll software for a five-person company": the sentence they can reuse holds a number, a condition and a verdict, not a string of adjectives. "Tested in March 2026 on twelve payslips, 4 minutes per payslip, one contribution error" can be quoted as is. "Great software, highly recommended" gives a machine nothing to repeat.
The NeuroEEAT scoring grid (0-10)
Reread your best review and ask what a reader could repeat at home using only what is on the page.
| Score | Description | Concrete indicators |
|---|---|---|
| 0-2 | No trial | Products are described from spec sheets or from reviews read elsewhere. No in-house photo, no date, no "I tried it". |
| 3-4 | Trial claimed | The page says "we tested" and shows nothing. No conditions, no measurement, a score out of 10 from nowhere. |
| 5-6 | Trial visible | Your photos show the product in use, the length of the trial is stated, and two or three precise remarks could only come from someone who used it. But only one product is tested, or each product is tested a different way. |
| 7-8 | Test with a protocol | Criteria are written before results, several products go through the same steps, measurements sit in a table, and the verdict says which one to pick and for whom. |
| 9-10 | Repeatable test | The full protocol is published with dates, equipment, raw data and the limits of the test. The page says who paid for the products. It is updated when a product changes, and other sites cite it. |
How to audit it on your site
Open the page where you recommend a product or service and look for these pieces of proof, one at a time.
- 01
Find the sentence that says how you tested
Near the top, a reader should learn in ten seconds what you tried, for how long and under what conditions. "Three invoicing tools, twenty invoices each, January 2026" is enough. Without that sentence, the rest of the page reads as an opinion.
- 02
Count the products that went through the same steps
A product tested alone has nothing to be compared with: "fast" next to what? Check that every product got the same trials, in the same order, on the same equipment. Two products compared fairly beat ten skimmed.
- 03
Spot the numbers you measured yourself
Highlight every figure on the page. Did it come from your stopwatch, your scale, your own log, or the maker's spec sheet? Google's guidance on reviews asks for quantitative measurements. A copied number does not prove you had the product.
- 04
Check that the verdict actually picks
Read only the last paragraph. Does it say which one to choose, for what use, and which to avoid? "It depends on your needs" is not a verdict. A good one names a winner per situation: the cheapest to start with, the most complete for a team.
- 05
Look at what the page says about money
Who paid for the products? Do the links earn you a commission? Look for that disclosure near the top, not buried in the terms. Check the pros and cons too: a review that lists no drawback reads like an ad.
How to activate or improve it
Start with one solid review, of the product your customers ask about most. The method comes after.
Write a first honest review
- Pick two or three products from your own trade that you already know, and actually get hold of them.
- Write three criteria before you start, for example setup time, real cost over a year and quality of support.
- Run every product through the same steps and note what you see right away, with the date.
- Take your own photos or screenshots during the trial, not afterwards.
- End with one sentence that picks: which one, for whom, and why.
Write the protocol down and show it
- Add a "How we tested" box at the top of each review: products, duration, conditions, equipment.
- Put measurements in a table, one row per criterion, one column per product, with units.
- List pros and cons for each product, taken from your trial notes.
- State who paid for the products and whether links earn you a commission.
- Date the review and name the exact version or model, so readers know whether it still applies.
Make your testing a reference
- Create one "How we test" page, linked from every review, that describes the full protocol.
- Publish the raw data (logs, measurement file, unedited photos) so someone can check your work.
- Rerun the test on a fixed schedule or at each new version, and keep the history of results.
- Declare the review to search engines with Review markup in JSON-LD (code that crawlers read and visitors never see), with pros and cons in positiveNotes and negativeNotes.
- On a health or money topic, have a named professional in the field check the protocol.
5 common mistakes to avoid
Sector specifics
Devices and wellness products
Blood pressure monitor, thermometer, supplement: your trial tells readers what is easy to use, readable and sturdy. It does not tell them what treats a condition. QRG section 3.4.1 separates lived experience from medical advice, which needs a professional. State clearly where your review stops, and have a named clinician check the page.
Software, accounts and investments
Reviewing a budgeting app or accounting software means measuring entry time, fees actually charged and the quality of exports. Do not let the trial drift into investment advice. Show the fees recorded on your own account, with the date, instead of the ones in the brochure.
Review what you install or sell
A roofer who compares three membranes on their own job sites, a wine merchant who tastes every delivery: you already test, you just never write it down. Note the place, the date and what you found six months later. No large comparison site can copy that.
Review the tools of your trade
A law firm that tries three e-signature tools on real files, made anonymous, has more to say than a general comparison site. Describe the typical file, the number of trials and what got stuck. Your clients are asking the question you were asking.
Signals to activate in parallel
The review says what you measured. These signals show it was really you doing the measuring:
FAQ
How to write a product review, with an example?
Open with what you tested and how, then give the results, then pick. An example of the pattern: "I entered the same twenty invoices into three invoicing tools in January 2026. Tool A took 14 minutes, tool B 22, tool C 31. For a solo business, A wins; for a team that needs approvals, B." Three sentences, and the reader has the method, the numbers and the verdict.
How do you write a product review blog post?
Use the product first, write second. Build the post in five blocks: how you tested, the results table, pros and cons from your notes, who each product suits, and your verdict. Add your own photos taken during the trial. Say at the top who paid for the product and whether the links are affiliate links.
How can you say a product has good quality?
By using it long enough to see what gives out, not by reading the label. A spec sheet or a rating app is a starting point, not a test. Look at what happens after ten, thirty, a hundred uses, next to an equivalent product. Write down what held and what failed, with the date.
Is there a product review blog template?
Yes, and it fits in six lines: what was tested, how and when; the criteria; results per criterion; pros and cons; a verdict per type of buyer; the disclosure. A template only gives you the order. What fills it has to come from your own trial, or you end up with the same page as everyone else who used it.
Can you review a product you got for free?
Yes, if you say so and the protocol stays the same. A sample sent by the brand goes through the same steps, the same criteria and the same list of drawbacks as a product you bought. State at the top who supplied it and whether a link earns you a commission. What destroys trust is finding out afterwards, not the free product.
Case study
Before · a hands-on review site doing well in Google
After · Google's March 2024 update
Source: HouseFresh (Gisele Navarro), 2024, https://housefresh.com/how-google-decimated-housefresh/ · Figures reported by the site itself. This case shows a limit, not a recipe: a real testing protocol is needed to earn a reader's trust, but on its own it does not shield a site from a drop in visibility. Do not test to please an algorithm. Test because your readers need it.