The review you didn’t test shows
Structure and markup refreshed for current answer engines; the original analysis is preserved.
Google’s July product reviews update — the fourth in fifteen months — keeps rewarding the same thing: evidence that you actually used the product. Original photos, real measurements, what surprised you, what the spec sheet left out. The engine is learning to measure the difference between a review you lived and one you assembled. The one you didn’t test shows.
the short answer
The July product reviews update — the fourth in fifteen months — rewards evidence that you used the product, not a claim that you did: original photos, real measurements, the trade-offs you noticed, what the manufacturer would not tell you. The engine cannot read your mind, but it reads the evidence, and real use leaves traces copying a spec sheet does not. Affiliate and comparison sites built on manufacturer descriptions are the most exposed. If you do test, make it legible — reward follows detectable evidence, not assertions.
key takeaways
- Google’s July product reviews update — the fourth in about fifteen months — keeps rewarding the same thing: evidence you actually used the product.
- Original photos, real measurements, the trade-offs you noticed, what the spec sheet left out: the marks a person actually handled the thing.
- The engine cannot read your mind, but it can read the evidence, and real use leaves traces that copying a spec sheet does not.
- Affiliate and comparison sites built on manufacturer descriptions are the most exposed — that model is exactly what the update discounts.
- If you do test, make it legible: show the photos, the numbers, the conditions, how you reached a verdict. Reward follows detectable evidence, not claims.
assembled vs lived
The left page can be written without leaving your chair; the right one cannot. That is the whole distinction the update is built around — not how polished the prose is, but whether the marks of real use are present. Polish is cheap now; evidence is not.
Why this is a measurement story, not a morality tale
It is easy to read these updates as Google scolding lazy reviewers, but that framing misses what is actually happening, which is a measurement problem being slowly solved. Google cannot directly observe whether you used a product; it can only observe your page. So over four rounds it has been refining its read of the things that correlate with real use — original media, specific measurements, lived trade-offs, detail that no spec sheet contains — and weighting them up while weighting down the markers of paraphrase. The honesty of the review matters only because honesty leaves a measurable trace. Seen this way, the update is not a moral judgement on your effort but a steadily improving instrument for detecting a particular kind of evidence, and your job is to make sure the evidence you legitimately have is actually on the page where the instrument can register it.
That reframing matters because it tells you what does and does not work in response. Protesting that your reviews are sincere does nothing, because sincerity is not the measured quantity; the measured quantity is the evidence of use, and a sincere review with none of it scores the same as an insincere one. Equally, dressing up a paraphrase with confident language does nothing, because confidence is not evidence either. The only move that registers is adding the real traces — the photo, the number, the specific failure, the genuine comparison — because those are what the instrument is tuned to find. Measurement, not motive, is the lever, and it is a more useful lever because you can actually act on it.
The argument, in three parts
The signal is evidence of use rather than a claim of it; an assembled review gives itself away by what it lacks; and the expense of real use is not a problem with the strategy but the strategy itself. Open each part.
01 The signal is evidence of use, not a claim of it
Read the July update alongside the three before it and the message is consistent to the point of monotony: Google wants proof that a real person used the product. Its guidance asks for your own photographs and audio, quantitative measurements, the decision factors that matter in the category and how the product actually performs against them, and design choices and their effects beyond what the manufacturer says. Every item on that list is a trace that genuine use leaves and that copying cannot. The update is not testing your sincerity or your credentials in the abstract; it is looking for the physical residue of experience on the page. That is a measurement problem reframed as an honesty one — the engine rewards what it can detect, and what it has learned to detect is the difference between having handled a thing and having read about it — a difference that turns out to leave surprisingly clear marks on a page.
02 What gives an assembled review away
A review built from spec sheets has tells, and they are the inverse of evidence. It restates specifications anyone can find, because that is its source. It stays general exactly where lived experience would get specific — the odd failure mode, the thing that is better or worse than the number suggests, the trade-off you only feel after a week. It reaches a verdict that could apply to anything in the category, because no particular product left its mark on the writing. Polish does not hide this; a perfectly written paraphrase is still a paraphrase, and the absence of first-hand detail is itself a signal. The engine does not have to prove you did not test the product. It only has to notice that the marks of testing are missing, and with four updates of practice it is getting good at noticing.
03 The expense of real use is the point
The uncomfortable part, especially for affiliate and comparison sites, is that there is no clever way around this, only a costly way through: actually get the products and use them, or work with people who have. That is more expensive than rewriting a product page, and the expense is not a bug in the strategy — it is the strategy. The whole reason the update can reward first-hand experience is that first-hand experience is hard to fake, which means the cost you resent is also the moat your thin competitor cannot cross. Spending it buys you something durable: a page full of evidence no spec-sheet rewrite can match, on a topic where the engine has spent four rounds learning to tell the two apart. The sites that treat the cost as an investment will pull away from the ones still looking for the shortcut, and the gap will widen with every update that sharpens the instrument.
What to do with this
If you genuinely test products, your task is legibility: make the evidence you already have impossible for either a reader or the engine to miss. Put your own photographs on the page, the measurements you recorded, the conditions you tested under, the trade-offs you actually hit, and a plain account of how you reached your verdict; Google even suggests a standing page describing your testing method, linked from each review. Where you recommend a best pick, say why in terms of first-hand evidence rather than restated specifications. The point is to close the gap between the experience you have and the experience the page can prove, because the engine can only reward what it can detect.
If you do not test — if your reviews are paraphrased product pages with buy links attached — the honest read is that the shortcut is closing, and four updates in fifteen months say it will keep closing. The response is not a better rewrite but a different model: obtain the products and use them, or partner with people who do, and let that evidence carry the page. It costs more, and the cost is exactly what protects you once you have paid it, because it is the thing a thinner competitor cannot copy. Being able to show, not just assert, that you used the thing is the durable position — and being the genuinely experienced, well-evidenced source is the groundwork the AC Group has built for 27 years.
What four updates in fifteen months is really telling you
It is worth pausing on the cadence, because the cadence is the message. This is the fourth product reviews update since April 2021 — four deliberate passes at the same problem in about fifteen months — and that rhythm says something a single update never could. Google does not return to a problem four times unless it considers it central and unsolved, and it does not keep refining the same instrument unless it intends to keep using it. The sites treating each round as a discrete event to weather, tweaking a few pages and waiting for the next one, are misreading the situation. The honest read is that a standard is being set and tightened on a schedule, and the only stable response is to meet the standard rather than to manage each occurrence of it. A direction this consistent is not noise to ride out; it is the shape of where review content is going.
That changes the calculation for anyone publishing reviews. If this were one update, gaming it briefly might make sense; against a sustained direction, every shortcut is borrowed time, because the next pass is already coming and is built to find what the last one missed. The compounding logic runs the other way for genuine experience: a page rich in real evidence does not just survive this update, it survives the next three, because the thing it has — proof of use — is exactly what each successive round is tuned to reward more sensitively. Reading the cadence correctly turns a stream of worrying announcements into a single, clear instruction: build the kind of review that gets stronger, not weaker, every single time the instrument improves — and stop spending effort on the kind that gets more exposed each pass.
What evidence actually looks like, by category
Evidence is concrete, and it helps to see what it means in practice, because the word can sound abstract until you tie it to a product. Google’s own guidance gives the shape of it: identify the decision factors that actually matter in a category and show how the product performs against them. For a car, that is fuel economy, safety, and handling, measured and described from real driving rather than lifted from a brochure. For a washing machine, it is the strengths and weaknesses of each cycle and feature, named in the language of someone who has run laundry through it. For a laptop, it is the battery under your actual workload, the keyboard after a week of typing, the thermals when you push it. In every case the move is the same: go past what the manufacturer claims to what you observed, and put a number, an image, or a specific consequence next to it.
Notice that none of these is reachable from a spec sheet, which is the entire point. The manufacturer’s page can tell you a battery is rated for a certain capacity; only use tells you it gets you through a real day with the screen bright and a dozen tabs open, and only use tells you the hinge loosens or the fan whines after a month, the sort of detail a buyer remembers and no spec sheet carries. That gap — between the stated specification and the lived consequence — is where evidence lives, and it is precisely the gap the update rewards you for closing. When you describe the consequence rather than restate the spec, you are not just writing better; you are leaving the exact trace the instrument is built to find, and doing for the reader the one thing the manufacturer’s copy never will, which is the truth about living with the product rather than the case for buying it.
The product reviews update, plainly: quick answers
What is the July product reviews update actually rewarding?
Evidence that you used the product, not a claim that you did. This is the fourth product reviews update in about fifteen months, and the through-line never changes: Google keeps asking for original photos, real measurements, the trade-offs you noticed, the things the manufacturer would not tell you — the marks that a person actually handled the thing. What it discounts, with each round more firmly, is the review assembled from spec sheets and other people’s words. The update does not ask whether you say you are an expert; it looks for the traces of genuine use, and rewards the pages that have them.
How can an algorithm tell a real review from a repackaged one?
Not by reading your mind, but by reading the evidence, and real use leaves evidence that copying does not. Your own photographs of the product in use, measurements you took, specific quirks and failure modes, comparisons grounded in having both things on your desk, the surprises that only show up after a week — none of that can be lifted from a spec sheet, because it did not exist there. A repackaged review, however polished, tends to restate known specifications, stay vague where lived detail would go, and reach a generic verdict. The engine does not need certainty about your experience; it needs the correlates of it, and those correlates are exactly what an assembled review lacks.
Who is most exposed to this?
Affiliate and comparison sites that built their model on manufacturer descriptions, because that model is precisely what the update discounts. If your reviews exist mainly to place buy links and the body is a paraphrase of the product page, you are the case these rounds were designed to find, and four updates in fifteen months signals this is a sustained direction, not a one-off. The way through is not a trick but a cost: actually obtain and use the products, or partner with people who have, and let the evidence of that use carry the page. It is more expensive than rewriting specifications, which is the point — the expense is the moat, because it is what a thin competitor cannot fake.
We genuinely test products. How do we make that legible?
Show the testing, do not just assert it. If you ran a product through real use, put the proof on the page: your own images, the numbers you recorded, the conditions you tested under, the trade-offs you hit, a clear account of how you reached a verdict. Google even suggests a standing page explaining how you test, linked from each review. The aim is to make your genuine experience visible to a system that can only reward what it can detect — so the gap between what you did and what the page shows closes. You earned the experience; the task now is to leave its fingerprints where both a reader and the engine can find them.
A note on sources and timing
This is written at the end of July 2022, as the latest product reviews update rolls out. We have described it as Google framed it: the fourth such update in about fifteen months, asking for demonstrated first-hand experience — original media, measurements, lived detail, design effects beyond the manufacturer’s account — and discounting reviews assembled from spec sheets, with affiliate and comparison sites most exposed. We have not predicted future updates. The durable point holds regardless of the next change: the engine rewards the evidence of use it can detect, the review you did not test shows, and being a genuinely experienced, well-evidenced source is the groundwork the AC Group has built for 27 years.