Gemini knows Crunchbase.
It doesn’t know what Crunchbase became.
Across 20 buyer-intent queries and 60 AI responses, Crunchbase is cited 38 times (63.3%) and ranks first on 25 of them — a strong result by any benchmark. But every one of those Gemini citations describes a $49-per-month funding directory. On the five questions covering predictive intelligence — the product Crunchbase has spent eighteen months and a 900% enterprise ACV increase building — Gemini cites the company zero times, and hands exit forecasting to PitchBook and CB Insights instead.
This audit uses the Xtrusio methodology — a 20-query, three-platform framework designed to surface how AI platforms surface (or omit) B2B vendors during discovery-phase research by decision-makers.
Every citation below was produced by running real private-market research questions through ChatGPT, Claude and Gemini and recording which vendors each platform named, and in what order.
Crunchbase’s repositioning is visible only to platforms that search the live web. To the ones answering from memory, it does not exist.
On the five predictive-intelligence questions, the split is absolute: Claude cites Crunchbase 5 of 5, ChatGPT 4 of 5, Gemini 0 of 5. The two platforms that retrieved live sources found the MCP launch, the exit-prediction precision figures and the 84% funding-model recall. The platform answering from training data found a funding database and nothing more — and awarded acquisition forecasting to PitchBook Exit Predictor and CB Insights Mosaic. This is not a content gap. The content exists, it is indexed, and where it is retrieved it wins outright. The gap is that eighteen months of repositioning has not yet aged into what these models know by default.
Platform Scorecard
Where Crunchbase is found — and where the answer depends on whether the model searched
Three platforms, the same 20 questions, a 35-point spread. Claude and ChatGPT both ran live web retrieval during the session. Gemini answered from what it already knew. That single difference accounts for almost the entire gap.
Competitor counts verified across ChatGPT and Gemini. Crunchbase’s own three-platform rate is 63.3%.
AI Visibility Leaderboard
Who owns the private-markets conversation — and who owns which platform
Crunchbase leads the field overall. But the lead is thin and it is split: PitchBook out-cites Crunchbase on Gemini, Dealroom out-cites it on ChatGPT. There is no single displacement competitor here — there are two, operating on different axes.
AI Positioning Audit
20 buyer-intent queries — click any row to see the exact question
Every question was written from the perspective of a real decision-maker researching private-company intelligence during discovery — before they know which vendors exist. No question names Crunchbase or any competitor. These are the buyers whose AI answers decide whether Crunchbase enters the shortlist at all.
| # | Query Topic | Product Line | Claude | ChatGPT | Gemini |
|---|---|---|---|---|---|
| 1 | Daily funding-round feed | Prospecting | ✓ | ✓ | ✓ |
Exact question asked across all three AI platforms: “Our sales team wants to reach out to companies right after they raise funding. What’s the best way to get a reliable daily feed of new funding rounds that match the kind of companies we sell to?” | |||||
| 2 | Verified contacts at funded startups | Prospecting | ✗ | ✗ | ✗ |
Exact question asked across all three AI platforms: “We need verified email addresses and direct dials for decision-makers at recently funded startups. Which tools actually deliver accurate contact data at that stage?” | |||||
| 3 | Funding data into Salesforce | Prospecting | ✓ | ✓ | ✓ |
Exact question asked across all three AI platforms: “How can we push company and funding data into Salesforce automatically so our reps don’t have to research accounts manually?” | |||||
| 4 | Predicting who raises next | Prospecting | ✓ | ✓ | ✗ |
Exact question asked across all three AI platforms: “Is there a way to know which companies are likely to raise money in the next few months, so we can reach them before their budget is already committed?” | |||||
| 5 | Seed / Series A sourcing on a budget | Deal Sourcing | ✓ | ✓ | ✓ |
Exact question asked across all three AI platforms: “As a small early-stage fund, what’s the most cost-effective platform for finding startups that just raised seed or Series A rounds in a specific sector?” | |||||
| 6 | Fund performance and LP data | Deal Sourcing | ✗ | ✗ | ✗ |
Exact question asked across all three AI platforms: “I need fund performance benchmarks and limited partner information to compare our returns against peer funds. Where does that data actually come from?” | |||||
| 7 | Sub-$100/month research tools | Deal Sourcing | ✓ | ✓ | ✓ |
Exact question asked across all three AI platforms: “We’re a two-person investment team with a small software budget. Are there private company research tools under $100 a month that are genuinely usable for deal sourcing?” | |||||
| 8 | European and UK coverage depth | Deal Sourcing | ✓ | ✗ | ✗ |
Exact question asked across all three AI platforms: “Which private market databases have the deepest coverage of European and UK startups?” | |||||
| 9 | Acquisition and IPO prediction | Predictions | ✓ | ✓ | ✗ |
Exact question asked across all three AI platforms: “Are there tools that predict which private companies are most likely to be acquired or go public in the next year?” | |||||
| 10 | Published prediction accuracy | Predictions | ✓ | ✓ | ✗ |
Exact question asked across all three AI platforms: “How reliable are AI-generated predictions about private companies, and does anyone publish accuracy rates for them?” | |||||
| 11 | Spotting a sector early | Predictions | ✓ | ✗ | ✗ |
Exact question asked across all three AI platforms: “How do investors spot a sector that’s heating up before it becomes obvious in the press?” | |||||
| 12 | Behavioural vs filing-based signals | Predictions | ✓ | ✓ | ✗ |
Exact question asked across all three AI platforms: “Is there any private company data source that uses real user behaviour or engagement signals rather than just news articles and regulatory filings?” | |||||
| 13 | Licensing data into a product | Data & API | ✓ | ✓ | ✓ |
Exact question asked across all three AI platforms: “We’re building a fintech product and need to embed private company and funding data into our app. Which providers license that data for commercial use?” | |||||
| 14 | Continuous CRM enrichment | Data & API | ✓ | ✓ | ✓ |
Exact question asked across all three AI platforms: “What’s the best way to keep company records in our CRM continuously enriched and up to date without manual data entry?” | |||||
| 15 | Live data inside an AI assistant | Data & API | ✓ | ✓ | ✓ |
Exact question asked across all three AI platforms: “Can I connect private market and funding data directly to an AI assistant so it answers questions with live company data instead of guessing?” | |||||
| 16 | Headcount and revenue estimates | Data & API | ✗ | ✗ | ✗ |
Exact question asked across all three AI platforms: “Which company data APIs give the most accurate headcount and revenue estimates for private companies?” | |||||
| 17 | Building a niche landscape map | Market Research | ✓ | ✗ | ✓ |
Exact question asked across all three AI platforms: “I need to build a competitive landscape map of every startup in a niche category. What’s the fastest way to do that?” | |||||
| 18 | Tracking competitor moves | Market Research | ✓ | ✓ | ✓ |
Exact question asked across all three AI platforms: “How do I track when a competitor raises money, makes an acquisition, or hires new executives?” | |||||
| 19 | What analysts cite in reports | Market Research | ✓ | ✓ | ✗ |
Exact question asked across all three AI platforms: “Which sources do analysts trust for private company funding history when they need to cite numbers in a report?” | |||||
| 20 | Deal terms and cap tables | Market Research | ✗ | ✗ | ✗ |
Exact question asked across all three AI platforms: “For due diligence on a private acquisition target, where can I find deal terms, valuation multiples and cap table details?” | |||||
| TOTAL | 16/20 (80%) | 13/20 (65%) | 9/20 (45%) | ||
▹ Q8 and Q17 are recorded as citations on Claude, but in both cases Crunchbase is named as the weaker comparison — Dealroom and Tracxn are recommended over it.
Gemini’s Two-Year-Old Crunchbase
Where 35 points of citation rate disappear — and which competitor collects them
Gemini is not unaware of Crunchbase. It ranks the company first on six of twenty questions and describes it, on one, as the single mandatory software subscription for a lean investment team. That is not weak brand recognition. It is precise brand recognition — of the company Crunchbase was before February 2025.
The four questions below are where that costs the most.
“Are there tools that predict which private companies are most likely to be acquired or go public in the next year?”
“Is there any private company data source that uses real user behaviour or engagement signals rather than just news articles and regulatory filings?”
“How reliable are AI-generated predictions about private companies, and does anyone publish accuracy rates for them?”
“Which sources do analysts trust for private company funding history when they need to cite numbers in a report?”
Crunchbase’s content is not missing. ChatGPT retrieved the MCP launch within a week of release, the 84% funding-model recall, and the full exit-prediction precision table — and moved Crunchbase from absent to first place on the same question Gemini gave to PitchBook. The material works when it is found. What has not happened yet is the slower process by which eighteen months of repositioning becomes something a model knows without looking it up.
AI Topic Authority Map
Query heatmap — product line × platform
| Topic | AI Leader | Crunchbase Status |
|---|---|---|
| Daily funding-round alerts | Crunchbase | UNANIMOUS #1 (3/3) |
| CRM sync & enrichment | Crunchbase | UNANIMOUS (3/3) |
| Budget-tier deal sourcing | Crunchbase | UNANIMOUS #1 (3/3) |
| Data licensing & AI assistant access | Crunchbase | UNANIMOUS (3/3) |
| Exit & funding prediction | Split — Crunchbase / PitchBook | 2 of 3 — zero on Gemini |
| Engagement-signal differentiation | Harmonic | 2 of 3, rank 4 on ChatGPT |
| European & UK coverage | Dealroom | Named as the weaker option |
| Analyst-grade citation authority | PitchBook | 1 of 3 (33%) |
| Verified contact data | Apollo / ZoomInfo | INVISIBLE (0/3) |
| Fund performance, LP & deal terms | PitchBook / Preqin | INVISIBLE (0/3) |
4 queries
4 queries
4 queries
4 queries
4 queries
▹ Predictive Intelligence is the only Crunchbase product line that swings from 100% to 0% depending on which platform a buyer opens.
Methodology
How this Xtrusio AEO/GEO Audit was conducted
This research is based on Xtrusio’s proprietary AI visibility analysis framework.
Recommendations
Closing a positioning gap, not a content gap
The usual remediation does not apply here. There is no missing content — the predictive material exists, is indexed, and wins outright wherever a model retrieves it. The work is to shorten the distance between publication and model memory, and to defend three specific claims that are currently being argued against.
- Two platforms independently discounted the 80M-user signal in near-identical terms — that it measures who is researching a company, not who is buying from one. Publish the counter-evidence: correlation between engagement signal movement and subsequent funding or acquisition events, with the methodology shown.
- Move the exit-prediction precision table (0.90 acquisition, 0.74 IPO, 0.83 closure, 0.96 remaining private) and the 84% funding recall figure onto a permanent, crawlable methodology page rather than press releases. Gemini answers Q10 with “vendors rarely publish accuracy rates” — that page is the direct rebuttal.
- Rewrite pricing and comparison pages so predictive intelligence is not described as a Business-tier gate. All three platforms anchor Crunchbase at $49 Pro pricing while the enterprise motion drives the growth.
- Model memory follows third-party corroboration, not owned content. Place the prediction accuracy story with financial and technology press, and pursue independent evaluation or academic citation of the forecasting methodology — this is what moves a claim from retrieved to remembered.
- Contest Q19 directly. PitchBook is described as the institutional gold standard because its verification process is public and Crunchbase’s community-contributed sourcing is treated as a weakness. Publish the verification pipeline: what is machine-validated, what is human-reviewed, what the correction SLA is.
- Address the European gap or stop competing on it. Q8 is the only structural zero that looks addressable, and Dealroom wins it on every platform — on Claude, Crunchbase appears in the answer only as the weaker comparison.
- PitchBook shipped Time to Exit in July 2026, updated daily — a direct answer to Crunchbase’s exit-prediction signal. The category is being contested now, and whichever vendor defines its vocabulary will hold the default citation.
- Publish a recurring, dated prediction ledger — forecasts made, outcomes confirmed, accuracy tracked over time. Nothing else in this category produces citable evidence that compounds.
- Quarterly Xtrusio re‑audits to track when predictive positioning crosses from retrieved-only into default model knowledge on Gemini.
Your predictions are winning. Gemini hasn’t heard.
Let’s close the distance between what you publish and what the models remember.
