An analyst report is a structured, comparable, defensible summary of what a small number of very well-informed people believe about a market. It is not a measurement of your buyers, and it was never built to be one. Confusion between those two things is the most expensive mistake in enterprise vendor evaluation.
Short answer: buy analyst research when you need category structure, a defensible shortlist for a procurement committee, or a fast education in a market you do not know. Do your own customer research when you need to know what your buyers will object to, why your deals are actually lost, or whether a vendor claim survives contact with someone who paid for it. The two are complements, and the failure mode is substituting one for the other.
The 2025 financials of both major firms are public, and they say something useful about which of those two jobs is getting harder to sell.
What the numbers say
Gartner (NYSE: IT), full year 2025, reported 3 February 2026:
- Revenue of 6.5 billion dollars, up 4 percent as reported and 3 percent FX neutral.
- Contract value of 5.2 billion dollars, up 1 percent year over year FX neutral.
- Global Technology Sales contract value of 3.9 billion dollars, flat FX neutral; Global Business Sales contract value of 1.2 billion dollars, up 3 percent.
- Net income of 0.7 billion dollars, down 42 percent; adjusted EPS of 13.17 dollars, down 7 percent.
- In the fourth quarter, the Insights segment produced 1,283 million dollars of revenue at a 77 percent contribution margin, growing 1 percent FX neutral. Conferences grew 14 percent; Consulting fell 13 percent.
- The company repurchased 7.0 million shares for 2.0 billion dollars, cutting the share count by 8 percent.
Forrester (Nasdaq: FORR), full year 2025, reported February 2026:
- Total revenue of 396.9 million dollars, down from 432.5 million the prior year. Research revenue fell to 295.6 million from 316.7 million; consulting to 88.2 million; events to 13.1 million.
- A goodwill impairment of 110.7 million dollars, contributing to a GAAP net loss of 119.4 million dollars, or 6.28 dollars per share.
- Contract value of 292.4 million dollars, down from 311.9 million.
- 1,797 clients, down from 1,942. Wallet retention of 87 percent, down from 89 percent, against client retention of 77 percent, up from 73 percent.
- Total headcount of 1,474, with a sales force of 553.
- Guidance for 2026 of 345 to 360 million dollars, a decline of 13 to 9 percent.
Two very different firms. One conclusion that applies to both: the syndicated research product is not what is growing. Gartner's growth is coming from conferences, not from research contract value, and its research segment throws off a 77 percent contribution margin while growing 1 percent. Forrester is shrinking, and has told the market it expects to shrink again.
That is not a prediction about their futures. It is a signal about what has become scarce. Synthesising publicly available information into a comparable market view is exactly the task that got dramatically cheaper between 2023 and 2026. What did not get cheaper is evidence from people who are not in anyone's dataset, which is to say your own buyers.
The number that should reframe how you read a Wave
1,797.
That is Forrester's entire client base at the end of 2025: not seats, but client organisations. Gartner does not disclose an equivalent count, but its 5.2 billion dollars of contract value is concentrated among large enterprises by the same logic.
This is not a criticism. A premium subscription model is supposed to have a small, high-value client base, and by that standard both firms are working exactly as designed. But it has a consequence buyers rarely think through.
The "market view" you are deferring to is assembled largely from conversations with organisations who can afford a six-figure research subscription. The inquiry calls that shape an analyst's sense of what buyers want, what is hard, and which problems are urgent come from that population. If you sell to mid-market operations teams, or to designers, or to schools, or to anyone outside the large-enterprise IT buying centre, the analyst's market and your market overlap less than the report's confidence implies.
Our guide to secondary research covers the general discipline here: every secondary source was collected for someone else's purpose, and the first question is always whose.
Where the evidence in a vendor evaluation comes from
Analyst firms publish their evaluation methodologies, and they are more transparent than critics usually allow. The inputs to a major vendor evaluation typically include vendor briefings, a vendor-completed questionnaire, product demonstrations, publicly available data, analyst experience from client inquiry, and customer references supplied by the vendor being evaluated.
That last input deserves a moment.
In our companion piece on software review sites, the core limitation is that reviewers self-select: only people who bought can review, so the population skews. In a reference-based evaluation the problem takes a stronger form. The reference customers are chosen by the party under evaluation. Every vendor keeps a list of accounts that will take the call, and no vendor volunteers the account that is three months from churning.
This is well understood inside the firms, which is why methodologies weight references alongside demos, questionnaires and analyst judgement rather than treating them as decisive. But as a buyer you should read a reference-informed finding for what it is: evidence about the best version of a product, described by customers who agreed to describe it.
A second structural point, and this one saves real money. Most major evaluations have published inclusion criteria: minimum revenue, minimum customer counts, minimum geographic coverage. A vendor missing from a quadrant or wave has frequently failed a size threshold rather than a quality one. Reading absence as a negative verdict is one of the most common and most expensive misreadings in procurement, because it systematically excludes exactly the newer entrants whose economics differ most from the incumbents.
The divestiture is itself a data point
In January 2026 Gartner agreed to sell Capterra, Software Advice and GetApp to G2, a deal Gartner chairman and chief executive Gene Hall confirmed in the same Q4 2025 release quoted above, describing a "definitive agreement to divest the Digital Markets business."
Set aside the strategy. Notice what it says about the products.
Aggregated buyer testimony and analyst judgement had been sitting inside one company, and they turned out to be separable businesses with different buyers, different economics and different owners. They are not two grades of the same evidence. One is a large, shallow, self-selected population of people who bought. The other is a small number of experts synthesising briefings, inquiry and public data. Neither is a substitute for talking to your own market, and a buying process that reads both and interviews nobody has collected two kinds of secondhand information and called it diligence.
When to buy the report, and when to run the study
| The question you actually have | Best instrument | Why |
|---|---|---|
| What are the categories, and who is in them? | Analyst report | Category definition and cross-vendor structure are exactly what the format is for |
| Which five vendors should make the shortlist? | Analyst report plus review sites | Structured comparability, then breadth of user coverage |
| Will my procurement committee accept this vendor? | Analyst report | External name-brand defensibility is a genuine product |
| Why did we lose the last four competitive deals? | Customer interviews | Only the lost buyers know, and they are in no dataset |
| What will our buying committee object to? | Customer interviews | Objections are specific to your positioning and your market |
| Does this vendor claim hold up in production? | Customer interviews | Reference lists are curated; your own network is not |
| How should we price against the incumbent? | Customer interviews | Willingness to pay is measured, not summarised |
| Is this category actually growing for buyers like ours? | Both | Analyst view for the market, interviews for your segment |
The pattern is consistent. Analyst research is strongest on the structure of a market and weakest on the specifics of your buyers. Gartner's own widely cited B2B buying research makes the point from the other side: buyers spend only a small fraction of their purchase journey with suppliers, roughly 17 percent by that firm's estimate, and the rest researching independently. If most of the decision happens where you are not present, the only way to know what happened there is to ask the people who were.
Doing the primary half properly
Once the report has done its job, the remaining questions are about people, and they are answerable in days rather than quarters.
Koji is an AI-native research platform for exactly this. You write a brief; Koji drafts the interview guide; an AI moderator runs the interviews itself, by voice or by text, in the participant's own language, and probes when a respondent says something the guide did not anticipate.
- AI-moderated voice and text interviews in parallel, at any hour, without scheduling, which is what makes senior B2B respondents reachable at all.
- Six structured question types in one study:
open_ended,scale,single_choice,multiple_choice,rankingandyes_no. A single study returns a ranked list of decision criteria and the reasoning behind each rank. See the structured questions guide. - Automatic thematic analysis with descriptive and in-vivo codes grounded in participants' own words, clustered into one codebook across the study, as described in our thematic analysis guide.
- One-click reports where every theme links back to the transcript line that produced it, which is the audit trail a skeptical executive will ask for.
- No moderator bias and no moderator fatigue, so interview 25 is conducted exactly like interview 1.
The studies that most often replace an analyst-only decision are win-loss analysis, positioning validation, competitive intelligence interviews with buyers who evaluated both you and a rival, and mapping the real B2B buyer journey rather than the idealised one.
What each path costs
Analyst subscriptions are quoted, not published, and vary by seat count, coverage and inquiry entitlements. Forrester's disclosed contract value of 292.4 million dollars across 1,797 clients implies an average annual relationship in the low six figures, which is a reasonable order-of-magnitude anchor for an enterprise subscription.
Koji publishes everything:
- Insights: 29 euro per month, including 29 credits.
- Interviews: 79 euro per month, including 79 credits.
- Annual plans at 290 and 790 euro, two months free.
- Credits: 1 for a text interview, 3 for a voice interview, 5 for a report refresh. Extra credits at 1 euro, with packs at 50 for 50 euro, 100 for 95 euro and 250 for 225 euro.
- A quality gate: only conversations scoring 3 or above consume a credit.
- 10 free credits on signup, no card required.
Thirty win-loss and evaluation interviews is 90 credits. It does not replace a market map. It answers the questions a market map was never built to answer.
Related reading: analyst reports are one form of purchased evidence; see syndicated data vs custom research for the general rule, and NielsenIQ vs Circana vs Numerator for the retail measurement equivalent.
Frequently Asked Questions
Are Magic Quadrants and Waves biased toward large vendors?
They are structurally weighted toward scale, which is not the same as bias. Major evaluations publish inclusion criteria that typically require minimum revenue, customer counts and geographic coverage, so newer or smaller vendors are frequently absent because they failed a size threshold rather than a quality assessment. The practical implication for buyers is that absence from a quadrant carries far less information than a low position within one, and treating the two the same systematically screens out the entrants with the most different economics.
What did Gartner and Forrester actually report for 2025?
Gartner reported 6.5 billion dollars of revenue, up 4 percent, with contract value of 5.2 billion dollars, up 1 percent FX neutral, and net income down 42 percent to 0.7 billion dollars. Its fourth-quarter Insights segment ran a 77 percent contribution margin while growing 1 percent FX neutral. Forrester reported total revenue of 396.9 million dollars, down from 432.5 million, a 110.7 million dollar goodwill impairment, a GAAP net loss of 119.4 million dollars, contract value of 292.4 million dollars, 1,797 clients down from 1,942, and 2026 guidance implying a further 9 to 13 percent revenue decline.
Do analysts talk to a vendor's actual customers?
Published methodologies generally include customer references as one input among vendor briefings, questionnaires, demonstrations, public data and analyst inquiry experience. The important qualifier is that reference customers are usually supplied by the vendor being evaluated, so they represent the accounts most willing to advocate rather than a random sample of the customer base. Analyst firms account for this by weighting references alongside other evidence, but as a buyer you should read reference-informed findings as describing the product at its best.
Should I cancel my analyst subscription and do primary research instead?
Usually not. They answer different questions. Analyst research gives you category structure, cross-vendor comparability and external defensibility for a procurement committee, and none of those are produced by interviewing your own customers. What primary research replaces is the habit of using an analyst report as a proxy for buyer evidence: why deals are lost, what objections arise, whether a claim holds in production, and what your specific segment will pay. Most well-run functions keep both and stop asking the report to do the second job.
Why does Forrester having 1,797 clients matter to me?
Because it tells you which population shaped the market view. Analyst perspective is built substantially from inquiry conversations with subscribing organisations, and a subscription at that price point selects for large enterprises. If your buyers are mid-market, or sit outside the enterprise IT buying centre, the market described in the report and the market you sell into overlap less than the document's tone suggests. That is a reason to read the report as structure rather than as measurement.
How fast can primary research realistically be?
For interview-based work with your own contacts, days rather than weeks. An AI moderator runs interviews in parallel, at whatever hour the respondent is free, and codes each transcript as it arrives, so analysis is largely complete when fielding closes. The genuine constraint is recruitment: reaching lost buyers takes outreach, and no platform removes that. What disappears is scheduling, transcription, coding and the multi-week analysis phase that made a small question uneconomical to ask.
Stop reading about your market and talk to it
An analyst report tells you what a market looks like from above. Your buyers are the only source for what your deals look like from inside.
Start free with 10 credits. Write a brief, interview your recent won and lost buyers, and read coded themes today. No card, no scoping call, no subscription negotiation.
