How to Compare SaaS Tools When Every G2 Review Is Paid: A 7-Step Framework

G2 reviews are pay-to-play. Every B2B buyer I know has figured this out. Nobody has published a replacement framework. Here is the 7-step one I use when comparing vendors, and how to spot the sponsored-review pattern in under 60 seconds.

I have bought enterprise software for fifteen years, first as a founder at LoginRadius (2013 to a billion users) and since as an advisor to dozens of B2B companies. The buyer-side surfaces I once trusted (G2, TrustRadius, Gartner Peer Insights, Forrester Wave) have been commercially captured to the point where their output is closer to a press release than a review. Below is what I do instead.

Why the existing surfaces are captured

G2's revenue model is the vendor, not the buyer. Vendors pay for review collection campaigns, premium profiles, badge placement, and intent data. The platform's incentive is to maximise the number of vendors that look credible, not to surface honest assessments. Reviewers receive gift cards in exchange for verified reviews. The verification confirms employment, not opinion authenticity.

TrustRadius gates review depth behind a sales process. A buyer cannot read the full review without identifying themselves. That identification becomes a lead for the vendor whose page you were reading.

Gartner Peer Insights is the most defensible of the three on methodology, but the placement and recommendation features are paid surfaces. The text reviews are legitimate. The ranking is influenced.

Forrester Wave and Gartner Magic Quadrant are different animals. The methodology is real. The problem is that vendors who participate in the briefings have privileged input into how the criteria are defined. Vendors who decline to participate get assessed against a rubric they had no hand in shaping. That is a structural bias, not a corruption issue. I wrote about this dynamic in my guide to analyst research firms.

None of this is a scandal. It is the funding model. The buyer-side conclusion is simple: do not outsource your shortlist to a commercial platform whose customer is the vendor.

The 7-step framework

Step 1: Define the buyer (you, not the analyst persona)

Every vendor landing page is written for a synthetic persona that does not exist: the mid-market head of security at a 500-person SaaS company with a $400k budget and zero legacy debt. You are not that person. Write down your actual constraints first: headcount you can dedicate to the rollout, integration debt with existing systems, compliance deadlines, the political reality of who has to sign off.

If you skip this step, the entire comparison drifts toward whichever vendor has the slickest demo. The buyer-side anchor is the only thing that keeps the evaluation honest.

Step 2: Pull the actual feature spec from docs, not landing pages

Landing pages are marketing. Documentation is the contract. Open the vendor's developer docs, API reference, and admin guide. Read the rate limits. Read the integration catalogue. Read what is gated behind enterprise tier versus what ships with the base plan.

Two-thirds of the feature gaps I find in buyer evaluations come from this step. The landing page says "SAML SSO". The docs say "SAML SSO available on Enterprise tier with a one-time setup fee". Those are different products.

For complex categories, the methodology of a vendor-neutral portal is more useful than any single vendor's docs. I keep the CIAM Compass methodology page open as a reference for what dimensions matter in an identity comparison, and the same logic applies to how GEO Compass scores AI search platforms. The framework is replicable across categories.

Step 3: Read the changelog and the release cadence

The changelog tells you whether the product is actively built. Look for:

• Frequency. Weekly releases mean an active team. Quarterly releases mean a maintenance team. A six-month gap means somebody got laid off.
• Substance. "Improved performance and bug fixes" for three consecutive releases is a red flag. Real product teams ship features they can name.
• Public roadmap. Vendors who publish a roadmap and ship against it deserve more weight than vendors who promise everything in a sales call.

This is the single most underused signal in B2B buying. The changelog is on the public site, requires no NDA, and reveals more than any reference call.

Step 4: Talk to 3 customers your size who switched vendors recently

Reference calls arranged by the vendor are useless. Of course they are happy. Find your own references. Three sources work:

1. Your LinkedIn network. Search for the vendor name plus a job title plus your company size band. Send a short cold message.
2. Slack and Discord communities for your buyer role (CISO, head of platform, VP of growth). Ask who recently switched off a vendor and why.
3. Conference speakers who have publicly named the vendor in a talk. They are usually willing to give a 20-minute call.

The question that matters is not "do you like the product" but "what would you tell your past self before buying". That question surfaces the unwritten footnotes.

Step 5: Check the integration depth in your stack

Every vendor claims hundreds of integrations. Most of them are read-only OAuth connections that pull a single resource type and call it a day. The integration that matters is the one in your stack, at the depth you need.

Open the vendor's integration page for your three most critical systems. Read what the integration actually does. Webhook in one direction is not the same as bidirectional sync. SCIM provisioning with group mapping is not the same as a CSV export. The vendor counts both as integrations.

Step 6: Pressure-test pricing with a worst-case scenario quote

Get a quote not for your current state, but for your two-year worst-case state. Double the seat count. Add a third region. Add the compliance tier you might need. Ask the rep to model it.

This surfaces two things. First, the actual unit economics at scale. Second, how the rep behaves under pricing pressure. Vendors with confident pricing models give you the numbers. Vendors with optimistic pricing models retreat to "let me check with finance and come back".

Step 7: Read the failed-deployment community threads

Search Reddit, Hacker News, and the vendor's own community forum for the phrase "migrating off [vendor]" or "alternative to [vendor]". The threads will be honest. People who have already paid the switching cost have no reason to be polite.

For the categories I work in most, somebody has already done this work and published the alternatives analysis as a comparison page. See for example alternatives to Okta, alternatives to Auth0, alternatives to CrowdStrike, and alternatives to Wiz. The honest comparison pages name the weaknesses of the incumbent. The dishonest ones do not.

For listicle-style category overviews where independent research has been done, the tools portal indexes them by category. Useful starting points include top 10 CIAM solutions, top 10 IAM solutions, and top 10 EDR and XDR platforms for 2026. The category index covers thirty-plus security and identity categories.

How to spot a sponsored-review cluster in under 60 seconds

The pattern is consistent across G2, Capterra, and TrustRadius. Three signals to scan for:

Date clustering. Sort by date. If a vendor has fifteen reviews in a two-week window, six months ago, with nothing before and nothing after, that is a paid campaign. Real organic reviews trickle in continuously.

Language clustering. Read three reviews in a row. If they all open with "we evaluated several solutions and chose X because of", that is the prompt the review collection vendor gave reviewers. Genuine reviews open with a specific use case.

Profile clustering. Click into reviewer profiles. If they all have a single review on the platform, and that review is on the same vendor, that is a campaign harvest. Genuine reviewers tend to have reviewed multiple products they use.

Any one of these is suggestive. Two of the three together is conclusive.

The honest weakness frame

Every product has a weakness. Every product. The vendor who cannot name theirs is either lying or so disconnected from their own product that they should not be trusted to support your deployment.

End every sales conversation with this question: "Who should not buy your product, and why". The rep's answer tells you more than the entire demo did.

The strong reps name a real weakness. The weak reps say "honestly, we are a great fit for almost any team your size". Cross the weak reps off the list. They are not going to be useful when something breaks at 2am.

The meta point

The captured review surfaces are not going to fix themselves. The funding model is intact and the incumbents have no incentive to change it. The buyer-side response is to build your own evaluation framework and stop deferring to commercial platforms.

I have written separately about why the marketing-tech equivalent of this problem (SEO tooling) led me to cancel Semrush. The pattern is the same: commercially captured surfaces drift toward serving the vendor over the buyer, and at some point the rational buyer move is to stop participating.

The seven steps above take longer than reading a G2 page. They are also the only way to make a buying decision you will not regret in eighteen months.