What AI Insights and Review Sentiment Add

Two features, deliberately labeled apart

SeoMap now has three separate layers of analysis stacked under a scan’s metrics, and it matters that they’re not the same thing wearing different names. First is the automated analysis that was already there — a handful of concrete, rule-based observations (“SoLV jumped from 12% to 34%,” “a competitor overtook you at six points”) generated by a deterministic engine, no model involved, labeled plainly as automated analysis rather than AI. On top of that sits a genuinely new layer: a short AI-generated paragraph, produced by an actual language model reading the same numbers, and labeled “AI analysis” specifically so nobody mistakes it for the rule-based summary next to it. A third, separate feature — AI review sentiment — does something the first two don’t: it reads real customer text instead of scan numbers.

What the AI paragraph actually does

The rule-based analysis and the AI paragraph run on the exact same inputs — the same scan, the same competitor list, the same deltas against the previous scan — so the AI layer isn’t finding new data, it’s writing about the same facts in a way that reads like an explanation instead of a bullet list. It’s generated automatically once a scan finishes, with a manual “regenerate” button for anyone who wants a second pass on the same numbers. For an agency handing a report to a client who doesn’t think in ARP and SoLV, that’s the practical value: the deterministic bullets stay auditable and exact, while the AI paragraph turns them into something a non-specialist reads in ten seconds and actually understands.

Review sentiment: a different kind of input entirely

Review sentiment analysis works on text SeoMap has never touched before — up to 50 of a business’s most recent Google reviews, pulled on request and classified for sentiment split (positive/neutral/negative), recurring themes, and a handful of representative quotes. It’s opt-in per project, off by default, and capped at once a week per project — not a background job quietly running on every account, something a user turns on deliberately when they want a read on what customers are actually saying, not just where the business ranks. Pairing that alongside Share of Local Voice fills a real gap: SoLV says how visible a business is, review sentiment says something about whether the people who found it liked what they got — two different questions a heatmap alone was never going to answer.

What none of this does

Worth stating plainly, because “AI” invites more assumptions than the feature actually claims: none of this writes back to Google. There’s no auto-reply to reviews, no automatic edit to a Business Profile, no AI agent taking action on a listing on a business’s behalf — review sentiment reads and summarizes, it never posts. The separate, non-AI profile watch feature works the same way on the monitoring side: it flags that something on a profile changed, never who changed it or why, and it never changes anything itself. Both features are read-only by design, same as the API and the same as everything else that touches a business’s public listing.

The takeaway

The useful distinction isn’t “AI vs. no AI” — it’s what each layer is actually built to answer. The rule-based analysis stays exact and auditable for anyone who wants the raw facts. The AI paragraph turns those same facts into something readable for a client who’d rather not decode a metrics table. Review sentiment answers a question neither of those touches — what customers are actually saying — read from real review text, opt-in, and strictly read-only. Together they cover ranking, explanation and reputation from the same scan, without any of the three claiming to be the other.