Seller Ratings and Product Ratings in Google Ads 2026: How to Earn Stars in Your Ads
Four ads on the results page. Three look identical — headline, description, sitelinks. The fourth carries a line underneath: “4.8 ★ Seller rating (1,204 reviews)”. Bids are comparable, copy is roughly equal in quality. The click goes to the fourth, and that is not a matter of taste: stars change behaviour in the SERP more reliably than another headline variant.
Seller ratings and product ratings are one of the few ad improvements that cost nothing in media budget. You do not switch them on and you cannot buy them — Google attaches them once you clear the thresholds. The catch is that those thresholds run on a rolling window, so stars disappear as quietly as they arrived. Here is what the thresholds are, how seller and product ratings differ, how to collect reviews without breaking policy, and why your stars may have vanished last week.
Two ratings that get confused constantly
| Seller rating | Product rating | |
|---|---|---|
| What it describes | The store: delivery, service, returns, support | A specific product: quality, accuracy of description |
| Where it shows | Below Search ads, on Shopping ads, on your store page | On product listings in Shopping and on shopping ads |
| Data source | Aggregated store reviews over a rolling 12 months | A product review feed keyed to GTIN |
| What you configure | Nothing — you just need reviews | Programme enrolment plus a review feed |
| Typical effect | Higher CTR at the same bid | Higher CTR and conversion rate on the listing |
Both can appear at once — a shopping ad can show product stars and the store rating together. They do not compete and they are computed from different data.
Seller ratings: three thresholds you hold at once
- At least 100 unique reviews in the last 12 months, counted per country. It is a rolling window: a review that turns twelve months old drops out.
- An average of at least 3.5 stars. Below that, nothing is displayed, no matter how many reviews you have.
- Language and country must match the market you advertise in. Spanish reviews will not light up stars for a German campaign.
The practical consequence: 100 reviews is not a one-off task, it is a rate. Collect 300 reviews in one quarter and then stop, and a year later the stars vanish as those reviews age out. The metric that matters is not “how many reviews do we have” but “how many new reviews per month”, and it needs to stay comfortably above 100 a year in every market you advertise in.
Track velocity, not the total. Roughly 10–12 new reviews a month per market holds the threshold with headroom. Anything under 8–9 a month is a star loss waiting to happen within the year.
Where Google gets store reviews
- Google Customer Reviews — Google’s own free programme. After an order, the customer receives an emailed survey. You enable it with a snippet on the order confirmation page. The cheapest and most overlooked source.
- Google-approved third-party review platforms — review services that officially syndicate data to Google. Sensible if you already collect reviews elsewhere and do not want a parallel process.
- Reviews Google collects across its shopping surfaces — aggregated automatically, with no direct lever on your side.
What does not count: reviews you gather and publish on your own website. A five-star widget on your product page contributes nothing to stars in the SERP — the data has to arrive through a recognised channel.
Product ratings work differently
Product ratings live at item level, not store level, and they need a separate review feed. The structure is simple: product identifier (ideally GTIN), review text, rating, reviewer, date. The feed is uploaded to Merchant Center as its own data source.
Thresholds worth holding in mind (directional, not a guarantee):
- the programme expects a minimum catalogue-wide volume — on the order of 50 reviews — before a feed is accepted;
- an individual product generally needs around three reviews before stars appear; one enthusiastic review will not do it;
- reviews aggregate by GTIN, so reviews for the same product sold by other retailers can feed the same score.
That last point explains the recurring “I have three reviews but the listing shows 412”. Correct: a product rating is a rating of the product in the market, not just at your store. It cuts both ways — you can inherit a strong score for a good product, and an unflattering one for a product that simply is not good.
Ratings depend on feed hygiene
Product ratings attach by identifier. If a GTIN is missing or wrong, reviews will not bind to the listing even with a perfectly formed review feed. Feed quality is therefore a precondition — see feed diagnostics and product disapprovals and the wider feed rules and supplemental feeds workflow.
What it is actually worth
Industry estimates for the CTR lift from seller ratings cluster around 10%. Treat that as directional: the real number depends heavily on your vertical and on whether competitors have stars too. Two situations produce the largest effect:
- You are the only listing with stars. Contrast matters more than the absolute score.
- High-consideration or high-risk purchases. The bigger the basket and the greater the “will it even arrive” anxiety, the more weight a store rating carries.
And one where the effect approaches zero: when everyone shows 4.7–4.9, stars stop being a differentiator and become hygiene. Not having them is a penalty; having them is table stakes.
The second, less obvious effect is economic. More clicks at the same bid usually means better placement at the same CPC. That is why ratings belong in the same category as ad assets — they improve auction outcomes without raising bids, a mechanic covered in ad extensions and assets in Google Ads.
Why your stars disappeared, in order of likelihood
- Reviews aged out of the window. The most common cause by far. You collected volume a year ago and stopped, and one day the trailing-12-month count slipped under 100.
- Average dropped below 3.5. Usually after peak-season fulfilment problems.
- A new market. You launched in a new country with no reviews there; the threshold is per country.
- Domain or store change. Ratings are tied to a domain and do not always survive a migration.
- Your review platform broke. A syndication failure looks, from Google’s side, like the review stream simply stopped.
- Too few reviews in the campaign’s language. The total is fine; the language-specific count is not.
- The review feed failed. For product ratings: a broken upload or mismatched identifiers.
What to check first
- Open Google’s seller rating page for your domain and read the current score and review count Google actually counts.
- Compare that number against 100 for each country you advertise in, separately.
- Plot new reviews by month. A declining trend means the problem is your collection process, not Google.
- For product ratings, check the last successful review feed upload and the share of rows with errors.
Collecting reviews without breaking policy
It is easy to damage yourself here. The rules are blunt: no buying reviews, no writing your own, no rewarding positive ratings, and no gating — that is, no asking only satisfied customers to leave feedback. Review gating is detectable from an implausibly high share of five-star ratings and is punished by removing the rating outright.
What works legitimately:
- Ask everyone, and ask at the right moment. A few days after delivery, once the product has been used but the impression is fresh.
- Ask twice. One request after delivery, a second a week later to non-responders. The second touch usually contributes a meaningful share of volume.
- Remove friction. One-click rating inside the email converts far better than a link to a form behind a login.
- Answer the negatives. Resolving a complaint often prompts the customer to revise the score, and always reduces the share of one- and two-star ratings.
- Ask in the customer’s language. An English email to a Polish buyer gets a fraction of the response, and Google counts thresholds by language.
A useful planning figure: even a well-run process converts only a low single-digit percentage of orders into reviews. The arithmetic follows — to hold 100+ reviews a year you need an order volume several times larger. With low order volume, stars take longer to earn, and that is simply how it works.
The product review feed: what goes in it and how it breaks
The product review feed is a separate XML file or sheet uploaded to Merchant Center as its own data source. The minimum field set per review:
| Field | Contents | Common mistake |
|---|---|---|
| Product identifier | GTIN, MPN or your SKU | SKU instead of GTIN — reviews never bind to the listing |
| Rating | A number on the declared scale | Mixing 1–5 and 1–10 scales in one feed |
| Review text | A substantive comment | Empty strings and one-word reviews |
| Reviewer | Name or pseudonym | Hundreds of reviews from one author |
| Date | When the review was written | Export date instead of review date |
| Review URL | Page where the review is published | Homepage instead of the product page |
Three rules that save weeks of troubleshooting:
- Upload daily. Reviews age, and infrequent uploads mean the system is working from yesterday’s picture while new positive reviews never arrive.
- Do not filter by rating. Exporting only fours and fives is the same violation as asking only happy customers. The distribution has to reflect reality.
- Watch the error rate. Above a few percent of rows, the problem is systemic — usually encoding, date format or missing identifiers.
Multi-market accounts: where stars usually go missing
Accounts trading in several countries lose ratings more often than single-market ones, and nearly always the same way. The home market produces 90% of orders and clears the threshold easily; the other four accumulate 20–40 reviews a year each. The global counter looks excellent while stars appear in exactly one country.
What to do:
- Track the trailing-12-month review count per country, never as a store-wide total.
- Localise the request emails: language drives both response rate and whether the review counts for that market.
- Budget six to nine months before stars appear in a new market, and do not plan the first season around them.
- Do not scale spend into a market where you have no reviews, no promotions and no price advantage — you are entering the auction with nothing to differentiate you.
Domain changes, rebrands and migrations
Seller ratings are tied to a domain. Move to a new domain and the review history does not transfer automatically, so a store with five years of reputation suddenly appears in the SERP without stars. If a migration is unavoidable:
- start collecting reviews on the new domain before the switch, not after;
- keep permanent 301 redirects from old URLs — this matters for ratings and for landing pages generally;
- update the domain inside your third-party review platform, or the data stream stops silently;
- plan for a one- to two-month CTR dip and offset it with promotions and assets.
What to do while the threshold is out of reach
A young store or a narrow niche will struggle to hit 100 reviews a year in every market. Until stars arrive, build SERP trust by other means:
- Promotions. A discount annotation is a visible differentiator with no review threshold attached — see Merchant Center promotions.
- Assets. Callouts, structured snippets and links to warranty or delivery pages answer the same objections a rating would.
- Price. If you are genuinely below market, prove it and use it — measurement is covered in price competitiveness and best sellers.
- Specific copy. Without social proof, concrete detail carries the ad: delivery windows, return terms, stock status. What to change and what to watch is in Ad Strength and RSA asset reporting.
A 90-day rollout
| Period | Actions | Outcome |
|---|---|---|
| Week 1 | Enable Google Customer Reviews, add the snippet to the order confirmation page, verify opt-in handling | Review flow started |
| Weeks 2–3 | Set up the two-email sequence and localise it per market | Higher review conversion rate |
| Week 4 | Assemble the product review feed, audit GTINs across the catalogue | Foundation for product ratings |
| Month 2 | Apply to the product reviews programme, automate a daily feed upload | Stars begin appearing on listings |
| Month 3 | Build monitoring: trailing-12-month counts per country, average score, feed errors | Star loss detected in days, not quarters |
Monitoring is last on the list and first in importance. Most merchants discover a lost rating from a CTR drop a month after the fact. A simple rule: when the trailing-12-month count falls below about 130, you have roughly a quarter before the rating goes — which is enough time to fix it.
How ratings fit with everything else
Stars influence whether the ad gets clicked. Price decides whether the click is economically winnable. The landing page decides what happens next — and in automated campaign types that page is increasingly chosen by Google rather than by you, as covered in final URL expansion in Performance Max and AI Max. The three only work together: a high CTR pointed at a weak landing page just buys expensive traffic.
Across many accounts and markets, threshold monitoring is far easier to run centrally from a manager account — see Google Ads agency account rental. More material sits in the PPC Rebels blog.
FAQ
How many reviews do seller ratings require?
At least 100 unique reviews in the past 12 months per country, with an average of 3.5 stars or higher. The window rolls, so older reviews drop out of the count.
Do I need to enable seller ratings in Google Ads?
No. It is an automated asset: once you clear the thresholds it appears on its own. There is no “show stars” toggle.
How do seller ratings differ from product ratings?
Seller ratings score the store — delivery, service, returns. Product ratings score a specific item. The first needs store reviews; the second needs a product review feed.
Do reviews on my own website count?
Not unless they are syndicated through Google Customer Reviews or an approved third-party platform. An on-site widget alone produces no stars in the SERP.
Why does a listing show more reviews than I have collected?
Product reviews aggregate by product identifier, so reviews for the same item from other retailers are included. That is expected behaviour.
Can I delete a negative review?
Not on request. You can dispute a review that breaks policy — spam, abuse, a review about someone else’s order — and you can respond, which frequently leads the customer to revise the score themselves.
What happens if I offer a discount for a review?
Rewarding a positive review specifically is a violation. Asking for an honest review with no strings is acceptable, but any scheme that skews the rating distribution risks costing you the rating entirely.
Are ratings counted separately per country?
Yes. The 100-review threshold applies per country of serving, and review language has to match the market.
How quickly do stars appear once I qualify?
Usually within a few weeks of the data accumulating and passing validation. It is not instant.
Do ratings affect Quality Score?
Not directly, but they raise CTR, and expected click-through rate is one component of ad quality. The effect is indirect and durable.
Are product ratings useful if I do not run Shopping?
Much less so — they surface primarily on shopping formats. A Search-only advertiser should focus on the seller rating.
What should I do after a wave of negative reviews from a logistics failure?
Work the orders, respond publicly, and keep collecting. Pausing collection accelerates the decline, because the window is then left containing nothing but complaints.
