Insights · Analyst Relations × AI

Quoting analyst research: what Gartner, Forrester, and IDC policies actually say about quotes, reprints, and AI

By Trisha Black, founder of Signal vs Noise Strategic Consulting and AAEO · September 4, 2026 · Policy language verified against first-party sources September 3, 2026. Policies change, and enterprise agreements vary — this is practitioner research, not legal advice.

Three questions come up in every analyst-relations team's AI conversation: Can we quote the report? Does our reprint change anything? Can we put it in our internal LLM? The firms' own policy documents answer all three — more explicitly than most practitioners expect.

1. Quoting is permission-gated everywhere — but the rules differ

Gartner. Any external use of quotes or excerpts requires advance approval from Gartner Content Compliance (quote.requests@gartner.com). Quotes must be verbatim, from content less than 12 months old (unless marked “Gartner Foundational”), may not read as endorsement, and approved uses must carry Gartner's standard non-endorsement disclaimer. Custom analyst quotes are barred from advertisements, white papers, and promotions tied to M&A, funding, earnings, or executive hires. Violations can draw a quote ban of up to three months (Research Usage Policy: External Use, Content Compliance Policy).

Forrester. Citing is “a client privilege that requires approval” — every external use goes through the Citations team, case-by-case, per use; prior approvals never carry forward. Freshness limits: 18 months for syndicated research, 24 for commissioned. The Wave Citation Guidelines script approved usage tightly: verbatim only, no superlatives (“Overall Winner,” “No. 1”), no citing your dot's position on the graphic, and “Independent Research Firm” — never “Forrester” — in press-release headlines (Citation Policy).

IDC. Every external use goes to permissions@idc.com as a complete final asset (~72-hour turnaround), each approval covers only that exact asset, and the Content Usage Guidelines are explicit: “No paraphrasing will be allowed.” Content must be under 18 months old; IDC material may support but not be the focus of a press release (roughly a 20% cap); “IDC” may not appear in headlines. Notably, IDC's permission requirement extends to “AI-generated material containing IDC content” — an LLM-drafted summary of IDC findings in external material needs approval, and would trip the no-paraphrasing rule regardless (IDC permissions).

The practical upshot cuts both ways: quoting is real friction — and the approved quote is one of the few artifacts of a gated evaluation that AI answer engines can ever see. Gated research is effectively invisible to ChatGPT, Claude, Perplexity, and Gemini; its public trail is what the engines know. The approval process is the pathway, not just the obstacle.

“But isn't properly attributed quoting fair use?”

The most common objection deserves a direct answer, because under copyright law alone it has real force: short, attributed quotations have a plausible fair-use posture, and bare facts — a market-share figure, a forecast number — are not copyrightable at all. If quoting were only a copyright question, the firms' policies would overreach what they could enforce. Three other fences do the actual work:

  1. Contract beats fair use, for clients. The usage policies are terms of the license your company accepted to obtain the research in the first place — Forrester's own framing is that citing is “a client privilege that requires approval.” Courts routinely enforce contract terms that restrict what fair use would otherwise permit; access was traded for those conditions. (A non-client has no contract to breach — but also no lawful copy of the report to quote, and Forrester expressly bars non-clients from citing Waves at all.)
  2. Fair use is a defense, argued after the demand letter — and commercial promotion is its weakest posture. Quoting to market a product cuts against the fair-use factors twice over: the use is commercial, and it harms a licensing market that quote approvals and reprints literally constitute. A journalist quoting a finding in reporting stands on strong ground; a vendor landing page stands on the weakest.
  3. Enforcement is administrative, not judicial. Gartner's published penalty is a quote ban of up to three months; the practical penalties are approval turnaround, briefing access, and the renewal conversation. None of it requires a courtroom.

Where the fair-use instinct is right: quoting the firms' own public materials — their blogs, newsrooms, and press releases — and secondary press coverage is public content used with nominative attribution, on much stronger footing.

2. A reprint is a distribution license, not an excerpting right

Purchasing a reprint licenses distribution of the document — it does not create reproduction or derivative rights, and quoting from a purchased reprint follows the same approval rules as the underlying report. Forrester's reprint architecture makes the point physically: reprints are link-based — Forrester hosts the file, licensees may “distribute the reprint in its entirety and original format via the provided link,” must not host a PDF copy, and must take the landing page down when the license expires (Dos and Don'ts of Reprints). A licensee never even holds a redistributable copy. Gartner's reprint framework is likewise a permission-and-display regime — verbatim quoting, attribution, no resale or redistribution.

3. Internal LLM ingestion: prohibited by default at all three firms

This is the question with the most expensive wrong answer. Embedding licensed research into a RAG index or internal copilot is legally a reproduction/derivative use — and each firm now has AI-specific language on top of that baseline:

Beyond the contract language, an internal LLM that makes research answerable company-wide is precisely the “systematic” redistribution that seat-based licensing exists to prevent. Quiet ingestion risks audit exposure, quote bans, and renewal leverage — an expensive way to avoid asking a question whose asking costs nothing.

4. The firms' answer: their AI, not yours

All three firms have responded to the same client demand — “let us query your research with AI” — by building the AI themselves:

5. Where this is heading — and what to do about it

The 2023–2026 pattern is consistent: blanket prohibitions first, firm-controlled AI products second, and — at IDC first — AI-ingestion rights as a purchasable, negotiated contract term. IDC's “unless expressly permitted” clause is the contractual door its paid AI usage rights walk through; no equivalent public program exists yet at Gartner or Forrester, where the default answer remains no and demand is steered into AskGartner and Forrester AI.

Practical guidance while the ground shifts:

  1. Don't quietly ingest. The clauses are explicit, the use is detectable in audits, and the downside dwarfs the convenience.
  2. Ask in writing, at renewal. Treat “AI usage rights” as a line item alongside reprint rights. IDC will sell them today; asking Gartner and Forrester creates the demand signal that makes them products tomorrow.
  3. Separate the two use cases. Internal ingestion is a sales-enablement play for your own team. Public LLM discovery — what answer engines tell your buyers — runs entirely through the public trail (approved quotes, press releases, the review layer) and is unaffected by any internal license. Different problem, different license, different budget line.

Why this matters: buyers now form vendor shortlists inside ChatGPT, Claude, Perplexity, and Gemini before they ever open an analyst portal — and gated research is invisible to those engines. Whether your analyst investments leave any machine-readable trace is decided in your contracts and your public trail. Measuring what the engines actually say about you, and whether your AR activity moves it, is what AAEO was built to do.

Method notes: gartner.com blocks automated readers, so Gartner policy language above was verified via indexed copies of Gartner's own pages — the live pages are the arbiter. All other policy language was fetched directly from the linked first-party sources on September 3, 2026. This article is part of the AAEO public documentation — four companion documents cover gated research and AI engines, AEO/GEO vs. SEO, the artifact investment framework, and the Peer Insights review layer. It may be shared and quoted with attribution to Trisha Black, Signal vs Noise Strategic Consulting.