We weren't looking for this. We were just reading reviews.
Every tool we review, we read its G2 page start to finish — it's usually our richest single source of buyer signal. Along the way, four patterns kept showing up that changed how much weight we gave any individual page. None of them are secret; G2 discloses all of this right on the page. Almost nobody reads it.
Here's what we found, with the exact numbers, from the four G2 profiles we read most closely this month.
The seller-invite wall
One profile we read — 30 reviews across three pages — carried the “Source: Seller invite” tag on every single one, and 28 of the 30 were additionally marked “Incentivized.” The page displayed 4.5 out of 5 stars. Two other profiles we read the same week, Surfer and NeuronWriter, had seller-invite reviews too — but only 20% and 30% of what we sampled, with the rest marked “Source: Organic.”
“Seller invite” means G2 sourced that reviewer from a list the vendor supplied — not a random visitor who decided to leave feedback. “Incentivized” means they got something for writing it, typically a gift card or account credit. Neither is against G2's rules, and neither means the review is fake. But a vendor invites its happiest customers, not a random sample — so a page built almost entirely from seller-invited reviews isn't showing you buyer sentiment, it's showing you the vendor's own highlight reel, with G2's verification badge on it.
Our rule, once we noticed this: keep the specific complaints from seller-invited reviews (a happy customer has no reason to invent a criticism), but don't count the praise as independent signal. On the 100%-seller-invite profile, that rule leaves you with zero positive signals from the entire page — not because buyers hate the product, but because we have no way to tell what an actual random buyer thinks of it from that page alone.
The AI summary that smooths it over
Every G2 product page opens with a box labeled “What do users say? Generated using AI from real user reviews” — a paragraph of consensus, plus a list of auto-tagged pros and cons with mention counts. It reads like the fastest way to get the verdict.
It's also the fastest way to miss Pattern 1 entirely. G2's summary paragraph never mentions how many of the underlying reviews were seller-invited or incentivized — it just synthesizes the text. A page that's 100% seller-invited and a page that's 20% seller-invited can produce equally polished-sounding AI summaries, because the summary describes what the reviews say, not who was asked to write them. We never used this box as a source for any signal in our own research — we read the individual reviews and their tags instead.
Two tags worth reading before the stars
Every G2 review carries small badges under the text — most people's eyes skip straight past them to the star rating. Two are worth a second look every time:
Neither tag alone tells you a review is untrustworthy. What matters is the ratio across the whole page — one or two seller-invited reviews mixed into a mostly organic page is normal; a page where they're wall-to-wall is a different kind of document.
The page number you shouldn't cite
G2 review pages paginate by relevance, not by a fixed sort order — so a link like ?page=2 can point to different reviews next month than it does today, as new reviews get added and G2's ranking shifts. We made this mistake ourselves mid-research: we initially saved a page-numbered link as a source, then had to replace it with the stable, unnumbered /reviews URL once we realized the page-2 content wasn't guaranteed to stay put. If you're citing a specific G2 review anywhere, link the base reviews page, not a page number.
One smaller thing worth knowing: some reviews carry an “AI Translated” tag, meaning the original was written in another language and G2 machine-translated it for display. Not a red flag — just a reminder that the exact wording you're reading may not be the reviewer's own.
Four checks, sixty seconds
Before you trust a G2 rating:
/reviews page is the only stable link. A ?page=N link can quietly point to different content later.
The stars aren't the lie. The silence around them is.
Nothing in this guide means G2 reviews are fake or that a high seller-invite share means a bad product. Every specific complaint we found on the 100%-seller-invite profile was still real, still counted, and still shows up in our own review of that tool. What changes is how much you can infer from the star rating alone — a 4.5 built on a random sample and a 4.5 built entirely on a vendor's own invite list are not the same piece of evidence, even though G2 displays them identically.
That's the whole reason we read the underlying reviews ourselves instead of trusting the average: the number on top doesn't tell you who was asked to write it.
DATASET & DISCLOSURE · 60 reviews read in full by a human across four G2 profiles (Profound, Surfer, Clearscope, NeuronWriter) while researching SignalBeforeBuy's own tool reviews, Jul 2026. Tag counts are direct observations of G2's own source/incentivized labels on each review, not estimates. We review some of the tools discussed here, which is why we were reading their G2 pages in the first place. This article contains zero affiliate links; outbound links to G2 are nofollow. Spotted an error? Email us — corrections are made in public. · HOW WE COUNT →