Discussion starter: a local SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- Local SEO
Review count, recency or rating: which has correlated most with map pack gains for you, and how did you rule out other changes?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
If you run the correlation, review count usually comes out on top. That says more about what count stands in for (business age, size, brand demand, links and citations) than about what Google rewards. Rating rarely correlates strongly, partly because most businesses in a competitive map pack sit within a few tenths of a star of each other, and a variable with almost no spread cannot explain much variation even if it matters. Recency is the one practitioners increasingly argue is underweighted, but no controlled public test isolates it cleanly.
So there is not enough evidence to rank one of the three above the others as a universal rule. The more useful question is the second half of yours: how to stop a review campaign being credited with gains that actually came from a category change, a competitor suspension or a rank-tracking artefact. Most “reviews moved us up” stories do not survive that check, and the ones that do usually show that reviews moved as a bundle, not which part of the bundle did the work.
What Google documents and what practitioners claim
Google’s Business Profile help page explains local results through relevance, distance and “how well-known a business is” (the factor practitioners call prominence). On reviews it is direct: “More reviews and positive ratings can help your business’s local ranking.” That supports count and rating as inputs. The page says nothing about recency, velocity or review text.
Practitioner evidence fills the gap, with the usual limits. Whitespark has argued that review recency is the most underrated local ranking factor, and Sterling Sky has published case studies on both review recency and review count. These are valuable because they come from people who track local rankings closely, but they are individual case studies: one or a few businesses, one market, limited controls. Whitespark’s Local Search Ranking Factors survey is expert opinion rather than measurement. It is a good read on consensus, not proof of causation.
Why rating looks weak in correlation data:
Correlation measures how well differences in one variable line up with differences in another. If every business in the pack sits between roughly 4.6 and 4.9 stars, rating has too little spread to line up with anything. Count spans orders of magnitude (a new clinic with 12 reviews against an established one with 900), so it shows up strongly even when part of that strength belongs to the things count proxies for.
Why the three are hard to separate by observation
A review request programme changes count, recency and usually rating at the same time, because customers asked right after a good job skew positive. It also changes a fourth variable people forget: review text, which often names the service and the suburb. When the pack improves after a campaign, you have four candidate causes that all moved together.
Count
Plausible mechanism: prominence, judged relative to the competitors at each search location rather than as an absolute number. Likely diminishing returns once you are level with the pack. Confounded with business age, brand searches and links.
Recency
Plausible mechanism: evidence the business is active now. The practitioner claim is that rankings soften when reviews stop. Confounded with active marketing, seasonality and other profile activity happening at the same time.
Rating
Google names positive ratings as helpful. The clearer effect is on behaviour: users filter by rating in Maps and choose between listings on stars. Confounded with service quality, which drives calls and direction requests anyway.
Review text
New reviews add service, product and place words to the profile. Often the real change after a campaign, and the one almost nobody tracks separately.
What usually moves the map pack and gets credited to reviews
Before attributing anything to reviews, eliminate the changes that tend to move local rankings faster and more visibly.
Alternative cause Typical signature How to rule it out Primary or secondary category change Step change on category-matching keywords across most of the grid within days Profile edit history and a change log; compare category keywords against non-category keywords Business name edits (keyword additions breach Google’s guidelines) Sharp jump on name-matching terms only Change log; check whether competitors edited names too Competitor changes: suspension, move, new listing, spam removal You improve only at grid points where a competitor vanished Weekly snapshot of the top three at each grid point Tracking method Gains that appear in one tool or grid size but not another Fixed grid, fixed spacing, same tool, same time of day Opening hours at time of search Rankings differ by time of day (a field observation several local specialists report, not documented by Google) Track at a consistent time; log hours changes Website or landing page changes Organic web rankings for the same terms improve too Check the linked page’s organic rankings over the same window Links, citations, local press Gradual lift across many keywords Link monitoring alongside review tracking Local algorithm shifts Competitors also move, often in the same week Compare against competitors and locations that changed nothing Reviews filtered or removed Count dips, sometimes after a burst Weekly count snapshot, not just “reviews received” A measurement setup that can attribute a gain
- Fix the rank measurement first. Use a geogrid sized to the real service area, the same keyword set (category terms, service terms, “near me” variants) and the same time each week. Report average rank across grid points and the share of points in the top three. A single rank check from the office is not data.
- Keep a change log. Record every edit to categories, name, hours, website URL, services, photos and posts, plus site changes and new links. Most false attributions come from an undocumented edit.
- Track the review variables separately. Weekly: total count, reviews in the last 30 and 90 days, days since last review, rating, and how many new reviews mention the target service or suburb.
- Track competitors the same way. For the top three at your weak grid points, record their count, rating and last review date. Express your count as a ratio to the competitor median. Closing a gap is a better predictor than an absolute number.
- Decide the response window before you look. If you pick the window after seeing the chart, you will find a correlation.
- Use a control. For a multi-location business, stagger the rollout across matched locations. For a single location, controls are weak: use keywords the campaign could not plausibly affect and competitors who changed nothing, and treat the result as suggestive.
Hypothetical example:
A ten-clinic physiotherapy group matches clinics into two groups of five with similar starting review counts, ratings and competition. Group A starts SMS review requests in March; Group B starts in June. Between March and May, Group A’s share of top-three grid points rises while Group B’s stays flat, in the same algorithm environment. In June and July, Group B catches up. That pattern supports “the review programme caused the gain”, but it cannot say whether count, recency, rating or review text did the work.
To separate recency from count, the group pauses requests at two Group A clinics six months later. If their rankings soften while their total count stays high, recency matters beyond count. If nothing happens over a reasonable window, the recency claim weakens for this market. Rating cannot be ethically manipulated, so it has to be judged on natural variation and on conversion data.
Where rating earns its keep
Even if rating barely moves rank in your data, it moves revenue per impression. Users filter and choose on stars, and Business Profile performance data (calls, direction requests, website clicks) shows that effect more directly than a rank tracker. Treat rating as a floor to protect rather than a lever to pull.
Review content may also matter more than it used to. Ask Maps (Gemini in Google Maps) launched in the US and India in March 2026 and expanded to English globally in August 2026, with some features US-only. It is plausible that review text feeds conversational answers there, but Google has not documented how businesses are selected, so treat that as a reason to encourage specific, detailed reviews, not as a ranking claim.
Watch for this:
Google’s review policies prohibit incentivised reviews and selectively asking only happy customers (review gating). Fake or incentivised reviews are also a consumer-law issue in many markets, including Australia and the US. Unnatural bursts risk filtering and removals, which can wipe out the very gain you are trying to measure. A steady, ask-everyone process is both the compliant option and the one that produces cleaner data.
What I would prioritise, and what would change that
If this were my local account, I would run a steady, policy-compliant review flow that asks every customer, and never let it lapse. That keeps recency covered, builds count relative to the specific competitors at your weakest grid points, and protects the rating floor that drives conversion. I would log review text separately, because it is the variable most likely to be doing hidden work.
What would change the plan: a staggered test where count rises without any rank movement (count is not the constraint, so look at category, distance and relevance instead), or a pause test where rankings hold for months (recency matters less in your market than the case studies suggest). Until you have one of those, report review gains as “the review programme, as a bundle”, not as count, recency or rating on their own.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.