{"site":{"name":"Koji","description":"AI-native customer research platform that helps teams conduct, analyze, and synthesize customer interviews at scale.","url":"https://www.koji.so","contentTypes":["blog","documentation"],"lastUpdated":"2026-08-10T21:28:49.466Z"},"content":[{"type":"blog","id":"b16d6feb-61e0-48a9-8891-a3d7f811f9e6","slug":"software-review-sites-b2b-buyers-2026","title":"G2 vs Capterra vs TrustRadius (2026): What Software Review Sites Can and Cannot Tell You","url":"https://www.koji.so/blog/software-review-sites-b2b-buyers-2026","summary":"G2 acquired Capterra, Software Advice and GetApp from Gartner in early 2026, consolidating four major B2B review properties under one owner and removing the independence of cross-checking one review site against another. The FTC Consumer Review Rule (16 CFR Part 465, effective 21 October 2024) bans incentives conditioned on review sentiment, insider reviews without disclosure, and review suppression, with penalties up to 53,088 dollars per violation and first warning letters issued 22 December 2025. Review data remains useful for category discovery and shortlisting but is structurally incapable of revealing why deals are lost, because only buyers who purchased can review.","content":"Software review sites are an excellent way to build a shortlist and a poor way to make a decision. They tell you what buyers who already purchased a product think about it after the fact. They cannot tell you what the buyers who chose someone else were thinking, which is usually the more valuable half of the market.\n\n**Short answer:** use G2, Capterra, TrustRadius and Gartner Peer Insights to discover categories, narrow a field of thirty vendors to five, and check whether a feature actually exists in the shipping product. Do not use them to understand why deals are won and lost, to price a contract, or to predict whether a tool will survive contact with your team. Those questions need conversations with real buyers, including the ones who said no.\n\nTwo things changed the landscape in the last eighteen months: a consolidation that quietly removed the independence of the second opinion, and a federal rule that made a common review-generation tactic illegal.\n\n## The second opinion collapsed in February 2026\n\nOn 29 January 2026, G2 announced an agreement to acquire **Capterra, Software Advice and GetApp from Gartner**. Gartner confirmed the sale from its own side: in the Q4 2025 results released on 3 February 2026, chairman and chief executive Gene Hall listed among the quarter's events that the company had \"entered into a definitive agreement to divest the Digital Markets business.\"\n\nThe scale of the combined entity, from G2's announcement:\n\n- Roughly **6 million verified customer reviews**\n- More than **200 million annual software buyers** reached\n- More than **10,000 software vendors** served\n- More than **2,000 software and services categories**\n\nG2 co-founder and chief executive Godard Abel called it \"a transformational moment for G2 and, more importantly, the global B2B software industry.\" Trade press put the price at around 110 million dollars and the close in early February; the announcement itself did not disclose terms.\n\nHere is why a buyer should care, and it has nothing to do with the price.\n\n**The standard due-diligence move in B2B software evaluation was to cross-check a vendor's G2 profile against its Capterra profile.** Two independent datasets, two moderation teams, two ranking algorithms, two populations of reviewers. Agreement between them was meaningful evidence. Disagreement was a flag worth chasing.\n\nAfter the deal, four of the largest review properties in B2B software sit under one owner. Whatever their datasets do next, they now share a corporate incentive structure, and over time they will tend to share moderation policy, verification standards and scoring methodology. **Independence is a property of ownership, not of the number of domain names in your browser tabs.** A single change to a moderation rule now moves every property at once, in the same direction, and no amount of cross-referencing will surface it.\n\nThis is not an accusation of bad faith. It is a structural observation about correlated error, and it changes what a \"second source\" means in a vendor evaluation. If your procurement checklist says \"verify on two review sites,\" that checklist is now measuring something different from what it measured in 2025.\n\n## What the FTC Consumer Review Rule actually made illegal\n\nThe Federal Trade Commission's **Rule on the Use of Consumer Reviews and Testimonials, 16 CFR Part 465** (published at 89 FR 68077 on 22 August 2024), took effect on 21 October 2024. It is worth knowing precisely, because most summaries get the key provision backwards.\n\n| Section | What it prohibits |\n| --- | --- |\n| 465.2 | Writing, creating, selling or disseminating reviews that misrepresent that the reviewer exists, used the product, or had the experience described |\n| 465.4 | Providing compensation or incentives conditioned, expressly or by implication, on the review expressing a particular sentiment, positive or negative |\n| 465.5 | Insider reviews and testimonials by officers, managers, employees or agents, and their immediate relatives, without a clear and conspicuous disclosure of the relationship |\n| 465.6 | Misrepresenting that a review site the business controls, owns or operates provides independent opinions |\n| 465.7 | Review suppression, including unfounded legal threats to remove a review, and displaying a review set as complete while withholding reviews for being negative |\n| 465.8 | Selling or buying fake indicators of social media influence |\n\nOn **22 December 2025** the FTC sent warning letters to ten companies over potential violations. Christopher Mufarrige, Director of the Bureau of Consumer Protection, framed it plainly: \"As consumers increasingly depend on online reviews, the FTC is committed to ensuring companies comply with this Rule.\" The agency's own business blog put the exposure at up to **53,088 dollars per violation**, a figure that is adjusted annually for inflation. The letters were not findings of violation, but they marked the shift from education to enforcement.\n\nNote what 465.4 does and does not say. **It does not ban paying for reviews. It bans paying for a sentiment.** A 25 dollar gift card offered to any customer who writes an honest review is lawful. A 25 dollar gift card offered to customers who write a five-star review is not.\n\nThe review platforms are written to that line. G2's published community guidelines cap any incentive at **100 dollars** across all forms, including cash, gift cards, swag, tokens, credits and subscriptions, and state that \"eligibility to receive an incentive is never based on the opinions, positive or negative, expressed in the review.\" Incentivized reviews are tagged as such, alongside four other labels: Validated Reviewer, Current User, Source, and Rating Update. G2 also says it will remove reviews if it finds evidence of a seller soliciting only positive ones, and that it does not remove reviews at a seller's request.\n\n## The three limits that survive perfect compliance\n\nThe rule polices deception. It does not police sampling, and sampling is where review data actually goes wrong.\n\n### 1. Compliant incentives do not fix a selected population\n\nA vendor emails a review request to 400 delighted customers and nobody else. Every review that results is honest. Every incentive is compliant with 465.4. Every reviewer genuinely used the product. And the resulting rating is still a measurement of the vendor's ability to identify happy customers.\n\nThis is a selection effect, not a lie, and no disclosure label catches it. It is the same mechanism covered in our guides to [sampling bias](/docs/sampling-bias-research) and [survey response bias](/docs/survey-response-bias): the answer is fine, the sample decides the result. G2's guideline against soliciting only positive reviews addresses the crudest version, but a vendor that simply chooses which segment to ask has broken no rule at all.\n\n### 2. A five-star review can come from someone who never deployed the product\n\nG2's guidelines explicitly allow **trial and evaluation reviews**, on the reasonable grounds that early impressions are useful. The moderation team may return a review if the experience looks too limited, but the category is permitted by design.\n\nFor a buyer this is a serious readability problem. Onboarding is the part of a B2B product that vendors optimise hardest, and it is the part a trial reviewer is best placed to judge. **The failure modes you actually care about, such as what happens in month nine when the admin who set it up leaves, are invisible to the reviewer population that is easiest to recruit.** The label tells you the review is from a current user or not; it does not tell you how long the product survived in their stack.\n\n### 3. Only customers can review you, and losses are where the information is\n\nThis is the structural limit, and it is not fixable by any platform.\n\nA review site can only ever collect testimony from people who bought the product. The buyers who ran your evaluation, sat through your demo, read your pricing page and then signed with a competitor are absent from every review dataset that exists. So is the buyer who almost renewed and quietly did not.\n\nThat absent group holds the answers to the questions with the most commercial value: which objection actually killed the deal, which competitor claim landed, what the champion could not defend internally, and what the buying committee believed about your product that was not true. **Review sites are a census of your wins and a blind spot the exact shape of your losses.** It is [survivorship bias](/docs/survivorship-bias-customer-research) with a five-star average on top. The instrument for that is win-loss interviewing, covered in our [win-loss analysis guide](/docs/win-loss-analysis-guide) and the practical [win-loss interview questions](/blog/win-loss-interview-questions-2026), and there is no substitute for asking.\n\n## What each property is good for\n\n| Property | Owner after Feb 2026 | Best used for | Weakest at |\n| --- | --- | --- | --- |\n| G2 | G2 | Category discovery, mid-market and enterprise SaaS breadth, grid-style shortlisting | Long-horizon reliability, losses, negotiated pricing |\n| Capterra | G2 | SMB and vertical software, long-tail categories | Enterprise depth, independence from G2 as a second source |\n| Software Advice / GetApp | G2 | Guided shortlisting, buyer-advisor referral flows | Neutrality signals, since referral economics drive the model |\n| TrustRadius | Independent | Longer, more structured B2B reviews with more context per review | Volume in smaller categories |\n| Gartner Peer Insights | Gartner | Enterprise-grade verification, large-account perspective | Coverage of smaller vendors and newer categories |\n\nOne more thing worth naming, because it is disclosed and standard rather than hidden: review platforms are also advertising businesses. G2 sells buyer intent data, market intelligence and ad placement on category and competitor pages. That does not make the reviews false. It does mean the **page** you are reading is a media property with its own ranking and placement economics, not a neutral index of a market. Read the reviews; do not read the layout.\n\n## The research review sites cannot do, and how to do it\n\nOnce the shortlist exists, every remaining question is about people rather than products. The productive move is to stop reading testimony and start collecting evidence, which is the same argument we make about [analyst reports](/blog/analyst-reports-vs-customer-research-2026) and about buying [expert network calls](/blog/expert-networks-vs-customer-interviews-2026) as a proxy for your market.\n\nKoji is an AI-native research platform built for exactly this loop. You write a brief; Koji drafts an interview guide; an AI moderator conducts the interviews itself, by voice or by text, in the participant's own language, and probes when someone says something the guide did not anticipate.\n\n- **AI-moderated voice and text interviews** running in parallel, at any hour, with no scheduling and no calendar tetris with a busy buyer.\n- **Six structured question types** in a single study: `open_ended`, `scale`, `single_choice`, `multiple_choice`, `ranking` and `yes_no`. That means one study returns both a satisfaction distribution and the reasons behind every score. The [structured questions guide](/docs/structured-questions-guide) explains how the two aggregate together.\n- **Automatic thematic analysis** using descriptive and in-vivo codes grounded in the participant's own words, then clustered into one codebook across the study, as described in our [thematic analysis guide](/docs/thematic-analysis-guide).\n- **One-click reports** where every theme traces back to the transcript line that produced it, so a skeptical stakeholder can audit a claim in seconds.\n- **No moderator bias and no moderator fatigue.** Interview 30 is conducted exactly like interview 1, which matters when you are comparing accounts.\n\nThe practical pairing looks like this. Use review sites to build the shortlist. Use [review mining](/docs/review-mining-customer-reviews) to extract themes from public review text at scale, and [secondary research](/docs/secondary-research-guide) to frame the category. Then run interviews with your own won and lost buyers to answer the questions the public data structurally cannot, and use the [surveys versus interviews](/docs/survey-vs-interview) decision tree if you are unsure which instrument the question needs.\n\n## What it costs to close the gap\n\nKoji publishes its pricing rather than gating it behind a call:\n\n- **Insights: 29 euro per month**, including 29 credits.\n- **Interviews: 79 euro per month**, including 79 credits.\n- **Annual plans at 290 and 790 euro**, which is two months free.\n- **Credits: a text interview costs 1, a voice interview 3, a report refresh 5.** Extra credits are 1 euro each, with packs at 50 for 50 euro, 100 for 95 euro and 250 for 225 euro.\n- **A quality gate** means only conversations scoring 3 or above consume a credit. A junk interview is free.\n\nTwenty win-loss interviews with recent buyers is 60 credits. That is a smaller number than most teams spend on a single review-generation campaign, and it answers the question the campaign cannot.\n\n## Frequently Asked Questions\n\n### Are G2 and Capterra still independent of each other?\n\nNo. G2 announced an agreement to acquire Capterra, Software Advice and GetApp from Gartner on 29 January 2026, and Gartner confirmed the divestiture in its Q4 2025 results on 3 February 2026. The properties may continue to run separate datasets and interfaces, but they share an owner, and over time they are likely to converge on shared moderation and scoring policy. Cross-checking a vendor across G2 and Capterra is no longer a genuine second source. TrustRadius and Gartner Peer Insights remain separately owned.\n\n### Is it legal for vendors to give gift cards for reviews?\n\nYes, with an important limit. Section 465.4 of the FTC Consumer Review Rule prohibits compensation or incentives conditioned, expressly or by implication, on the review expressing a particular sentiment. Paying for an honest review is lawful; paying for a positive one is not. G2's published guidelines cap incentives at 100 dollars across cash, gift cards, swag, credits and subscriptions, state that eligibility never depends on the opinions expressed, and tag incentivized reviews. The FTC has warned that violations can carry civil penalties of up to 53,088 dollars each.\n\n### Has the FTC actually enforced the Consumer Review Rule?\n\nIt has begun. On 22 December 2025 the FTC sent warning letters to ten companies over potential violations of the rule, based on consumer complaints and information the companies provided. The letters were not formal findings of violation, but they required recipients to confirm in writing what they had done to comply and signalled the move from guidance to enforcement. The rule itself has been in force since 21 October 2024.\n\n### Can review sites tell me why we lose deals?\n\nNo, and this is their most important limitation. Only customers who bought a product can review it, so every review dataset is a census of wins with a blind spot in the exact shape of your losses. The objection that killed a deal, the competitor claim that landed, and the belief your champion could not defend internally are all held by people who never became reviewers. Win-loss interviews with recent buyers, including the ones who chose a competitor, are the only instrument that reaches them.\n\n### How many reviews does a rating need before I should trust it?\n\nVolume matters less than composition. A 4.7 average built from 300 reviews collected in one incentivized campaign aimed at a vendor's happiest accounts is weaker evidence than a 4.3 built from 40 reviews arriving steadily over two years from mixed segments. Look at the distribution of dates, the mix of company sizes, whether reviews are tagged as incentivized or as trial and evaluation experiences, and whether critical reviews exist at all. A category with no one-star reviews is telling you something about collection, not quality.\n\n### What should I do after the shortlist is built?\n\nSwitch instruments. Review data has already done its job once you have five candidates. The remaining questions are about your own context: what your buying committee will object to, what your team will actually adopt, what your lost deals believed about you, and which vendor claims survive contact with a real customer. Those are interview questions. Run them with your own won and lost buyers rather than with a public reviewer population you did not select.\n\n## Ask the buyers who are missing from the reviews\n\nEvery review dataset you can read was written by someone who bought. The decision you are making right now depends just as much on the people who did not.\n\n**Start free with 10 credits.** Write a brief, launch interviews with your recent won and lost deals, and read coded themes today. No card, no scoping call, no procurement cycle.","category":"Comparisons","lastModified":"2026-08-09T03:27:43.724626+00:00","metaTitle":"G2 vs Capterra vs TrustRadius (2026): What Software Review Sites Can and Cannot Tell You","metaDescription":"G2 bought Capterra, Software Advice and GetApp from Gartner in February 2026, ending the independent second opinion. What review data proves, what the FTC Consumer Review Rule bans, and the question review sites cannot answer.","keywords":["software review sites","g2 vs capterra","trustradius reviews","gartner peer insights","b2b software reviews","ftc consumer review rule","are g2 reviews reliable"],"aiSummary":"G2 acquired Capterra, Software Advice and GetApp from Gartner in early 2026, consolidating four major B2B review properties under one owner and removing the independence of cross-checking one review site against another. The FTC Consumer Review Rule (16 CFR Part 465, effective 21 October 2024) bans incentives conditioned on review sentiment, insider reviews without disclosure, and review suppression, with penalties up to 53,088 dollars per violation and first warning letters issued 22 December 2025. Review data remains useful for category discovery and shortlisting but is structurally incapable of revealing why deals are lost, because only buyers who purchased can review.","aiKeywords":["software review sites","g2","capterra","trustradius","ftc consumer review rule","win-loss research","b2b software buying"],"aiContentType":"guide","faqItems":[{"answer":"No. G2 announced an agreement to acquire Capterra, Software Advice and GetApp from Gartner on 29 January 2026, and Gartner confirmed the divestiture in its Q4 2025 results on 3 February 2026. Cross-checking a vendor across G2 and Capterra is no longer a genuine second source. TrustRadius and Gartner Peer Insights remain separately owned.","question":"Are G2 and Capterra still independent of each other?"},{"answer":"Yes, with a limit. Section 465.4 of the FTC Consumer Review Rule prohibits incentives conditioned on the review expressing a particular sentiment. Paying for an honest review is lawful; paying for a positive one is not. G2 caps incentives at 100 dollars, states eligibility never depends on the opinions expressed, and tags incentivized reviews.","question":"Is it legal for vendors to give gift cards for reviews?"},{"answer":"It has begun. On 22 December 2025 the FTC sent warning letters to ten companies over potential violations, requiring written confirmation of compliance steps. The letters were not formal findings of violation. The rule has been in force since 21 October 2024, with civil penalties up to 53,088 dollars per violation.","question":"Has the FTC actually enforced the Consumer Review Rule?"},{"answer":"No. Only customers who bought a product can review it, so every review dataset is a census of wins with a blind spot the exact shape of your losses. Win-loss interviews with recent buyers, including those who chose a competitor, are the only instrument that reaches them.","question":"Can review sites tell me why we lose deals?"},{"answer":"Composition matters more than volume. Check the distribution of dates, the mix of company sizes, whether reviews are tagged incentivized or trial and evaluation, and whether critical reviews exist at all. A category with no one-star reviews is telling you something about collection, not quality.","question":"How many reviews does a rating need before I should trust it?"},{"answer":"Switch instruments. The remaining questions are about your own context: what your buying committee will object to, what your team will adopt, and what your lost deals believed about you. Those are interview questions, run with your own won and lost buyers.","question":"What should I do after the shortlist is built?"}],"relatedTopics":["software review sites","b2b software buying","win-loss research","vendor evaluation","ftc consumer review rule"]}],"pagination":{"total":1,"returned":1,"offset":0}}