Google has confirmed that reviews factor into local ranking as part of “prominence,” one of the three named local ranking factors alongside relevance and distance, but Google has not disclosed a formula, a weighting breakdown, or even a fixed list of review sub-signals it evaluates. Anyone stating that recency counts for a specific percentage, or that velocity outweighs rating diversity by some defined margin, is inventing a number Google has never published. The honest answer separates two categories cleanly: what Google confirms (reviews affect prominence, and review count and score are explicitly named), and what practitioners reasonably infer from observed behavior and general ranking-system logic (that recency, velocity, diversity, and response rate plausibly matter in some form). Treating the second category as settled fact is the most common mistake in this space.
Direct answer
Google’s own “How Local Search Results work” help page states that local results are ranked based on relevance, distance, and prominence, and that prominence “takes into account information that Google has about a business from across the web, like links, articles, and directories,” along with explicitly naming review count and review score as contributing factors. That is the full extent of Google’s public disclosure on reviews as a ranking input. Google has not published, and has not confirmed through Search Liaison or engineer statements, any specific weighting for review recency versus review velocity (the rate at which new reviews accumulate) versus rating diversity (a healthy spread of detailed reviews versus a suspicious cluster of identical five-star ratings) versus how often and how well a business owner responds to reviews. Everything beyond “review count and score contribute to prominence” is inference, some of it well-grounded in observed patterns and Google’s general anti-spam logic, but inference nonetheless, and this piece flags it as such throughout.
What Google has actually confirmed
The Google Business Profile help documentation and the “How Local Search Results work” page are the two primary sources practitioners should ground claims in, and both are narrower than most SEO content implies. The local search help page confirms:
- Prominence is one of three ranking factors (relevance, distance, prominence).
- Prominence reflects information Google has about a business “from across the web,” a deliberately broad phrase that covers links, articles, directories, and other off-site signals.
- The number of reviews and the review score are explicitly called out as inputs to prominence.
Separately, Google’s Business Profile help documentation on managing and responding to reviews confirms that responding to reviews is a recommended practice that can influence customer trust and engagement, and that Google’s own guidance encourages timely, thoughtful responses. What that documentation does not say is that response rate itself is a ranking input measured and weighted by the local algorithm. Google frames responding to reviews primarily as a customer-experience and trust practice, not as a disclosed ranking lever. It is entirely plausible that responses correlate with better outcomes because responsive businesses tend to be more actively managed, more customer-attentive, and more likely to accumulate the kind of review profile that helps prominence generally, but that is a correlation argument, not a confirmed direct algorithmic weight on “response rate” as its own measured variable.
What’s reasonably inferred but unconfirmed
Several ideas circulate in the SEO community that have plausible logical grounding but no direct Google confirmation, and each should be presented to clients or stakeholders with that caveat attached.
Review recency. It’s reasonable to infer that Google’s systems would favor businesses with an ongoing stream of recent reviews over a business whose most recent review is several years old, because recency is a common signal type across many of Google’s ranking systems generally (freshness is a well-established concept in web ranking broadly), and because a business with no recent activity presents a plausible signal of reduced current relevance or even closure. But no Google source states “reviews older than X carry reduced weight” or specifies any decay curve. The inference is directionally reasonable; the specifics are not knowable.
Review velocity. The idea that a sudden, unnatural spike in review volume can trigger scrutiny is grounded in Google’s general anti-spam posture rather than a disclosed prominence formula. Google’s review policies explicitly prohibit fake and incentivized reviews and describe enforcement action against review manipulation, which indirectly supports the idea that velocity patterns are monitored, at minimum for policy violations, if not as a direct positive ranking multiplier. Treating “steady, natural velocity is safer than review spikes” as sound practical advice is reasonable; treating it as a disclosed ranking weight is not.
Rating diversity. The intuition that a spread of detailed, varied reviews (mentioning different aspects of a business, written in different lengths and voices) looks more organic than a wall of terse five-star reviews is consistent with how most trust-and-quality systems are generally understood to work, and aligns with what Google’s review policies say about authenticity. But this is again an inference from general principles and stated anti-spam goals, not a disclosed sub-signal with a named weight.
Response rate as a direct ranking input. This is the least supported of the four in terms of direct disclosure. It is well supported as a customer-experience recommendation from Google’s own help content, and reasonably inferred to correlate with better prominence outcomes indirectly, but there is no statement from Google confirming that response rate itself is measured and weighted as a standalone ranking factor.
A worked example of confirmed versus inferred
Suppose two hypothetical dental practices, Practice X and Practice Y, sit in the same local pack. Practice X has 85 reviews averaging 4.6 stars, accumulated steadily over three years, with a mix of short and detailed reviews and a visible pattern of the owner responding to nearly every one. Practice Y has 90 reviews averaging 4.7 stars, but 60 of them arrived within a single two-week span last spring, nearly all five stars, nearly all short and similarly worded, and the owner has never responded to any of them. If Practice X consistently outranks Practice Y in the local pack, it’s tempting to build a tidy story: “recency and steady velocity beat a review spike, and response rate mattered.” That story might even be directionally correct. But nothing about this comparison lets you conclude Google weighted velocity at some specific percentage or that response rate contributed a measurable ranking increment, since Google has only confirmed that review count and score feed prominence at all. The honest read is narrower: Practice Y’s pattern is consistent with the kind of review-manipulation profile Google’s policies target for enforcement, and Practice X’s steadier profile is safer and more defensible, but the exact mechanism and weighting connecting that pattern to the ranking gap between them isn’t something either practice’s SEO could actually verify.
What to do about it
Given the confirmed-versus-inferred split, the practical posture is to treat reviews as a genuinely important prominence input (that part is confirmed) while refusing to over-engineer a strategy around unconfirmed sub-weights. Concretely: encourage a steady, ongoing flow of genuine reviews rather than one-time pushes, since steady accumulation is defensible under both the confirmed “review count” signal and the reasonable “avoid suspicious velocity spikes” inference. Respond to reviews consistently, both positive and negative, because Google’s own guidance recommends it for trust and engagement even though it isn’t confirmed as a direct ranking multiplier. Avoid review-gating or incentivized review schemes, since these violate Google’s explicit review policies and risk enforcement action regardless of any ranking-weight question. And when reporting results to clients or stakeholders, resist the temptation to present a specific weighting model for recency, velocity, diversity, or response rate as if Google disclosed it, since doing so misrepresents the actual state of public documentation and sets expectations that can’t be verified or debugged when they don’t hold. The defensible claim is narrower than the popular narrative, and that narrower claim is the one worth building strategy on.
