Nearly Half of B2B SaaS Is Invisible to AI. Are You?
Being liked and being retrieved are two different problems. Nearly half of B2B SaaS wins the first and loses the second.
In a 2026 benchmark, 44 of 50 brands earned perfect sentiment — and nearly half still scored below 50 out of 100 on visibility. The market doesn’t dislike them. The machine can’t recognize them.
A new benchmark just tested 50 B2B SaaS companies to see how often AI assistants name them when a buyer asks who to trust.
Nearly half scored below 50 out of 100.
It wasn't because the market necessarily dislikes them. When the study scored sentiment — how positively the AI talked about each brand when it did mention them — 44 of the 50 companies earned a perfect 20 out of 20.
The buyers' favorite tools and the buyers' invisible tools got the same warm treatment. The only thing separating them was whether the machine mentioned them at all.
That's the whole story of AI search in one dataset. Being liked and being retrieved are two different problems.
You can have a great product, happy customers, and a reputation your competitors envy — and still be missing from the answer your next buyer is reading right now.
I want to show you why that gap exists, what the companies on the right side of it are doing differently, and one client of mine who sits so far on the visible side that Google's AI cites them in roughly four of every ten answers about their category.
- 44% of 50 B2B SaaS brands scored below 50/100 on AI visibility in DerivateX’s 2026 benchmark — across 1,400 buyer-intent prompts in ChatGPT, Perplexity, Claude, and Gemini.
- The gap is recognition, not reputation. 44 of 50 companies earned perfect sentiment; visibility still varied by 87 points top to bottom.
- Ranking no longer guarantees citation. Only 38% of AI Overview citations come from top-10 pages, down from 76% a year ago.
- The cited side is buildable. One anonymized client is the #1 cited brand in all 10 of its core topics, named in ~39% of Google AI answers — built on human voice and proprietary proof, not volume.
- The fix is judgment, not volume: non-commodity pages, journey-deep coverage, off-domain corroboration, and measuring your top queries.
What does "invisible to AI" mean?
Definition
AI visibility is how often — and how prominently — AI assistants name your brand when buyers ask for recommendations in your category.
Measured in the DerivateX study as an “AI Presence Score” (0–100): mention rate and position carry 60 of the 100 points.
Being “invisible to AI” means that when a buyer asks ChatGPT or Perplexity or Google's AI to recommend software in your category, your name doesn't come up — even if you'd win the deal on the merits.
The study that measured this, DerivateX's 2026 benchmark, ran 1,400 buyer-intent prompts across the four major assistants and scored each of 50 companies from 0 to 100 on an "AI Presence Score."
| Category | More visible | Less visible | Gap |
|---|---|---|---|
| SEO analytics | Ahrefs — 83 | Semrush — 68 | 15 |
| Field service | ServiceTitan — 68 | Jobber — 41 | 27 |
| Payments | Stripe — 65 | Razorpay — 39 | 26 |
| Automation | Zapier — 63 | Make — 40 | 23 |
Mention rate and position carried 60 of those 100 points, because that's what a buyer sees: are you in the answer, and where. The average score came in at 56.9.
The spread was brutal — 87 points between the top company and the bottom, both selling into established, crowded categories.
Note the exact finding, because it's about to get mangled everywhere: 44% scored below 50 out of 100.
But hold on — not "44% don't show up at all." That precision matters, and I'll come back to why it's the whole game.
Get the number right
You’ll see this study reported as “44% of B2B SaaS don’t show up in AI.” That’s not the whole truth about what it found. 44% scored below 50 out of 100 on visibility — they under-perform, they’re not absent. On a topic about citation integrity, the precise and honest data point is what's worth repeating.
Why isn't my SaaS showing up in AI answers?
If you aren’t showing up, it’s likely because AI cites the most recognizable content. When a buyer asks an assistant a question, the model doesn't reread the internet and rank the smartest page.
It assembles an answer from the sources it already recognizes as belonging to that topic — the ones that show up consistently, distinctly, and in enough places that the machine treats them as an entity worth naming.
Recognition is the currency. Quality is table stakes you've already paid.
This is why the sentiment finding is so damning. The invisible 44% aren't losing on quality or reputation. They're losing because they read like everything else in their category, so the machine has no reason to reach for them specifically.
When your content sounds like every competitor's, you don't get penalized. You get unretrieved — which is worse, because it's silent.
AI doesn’t cite the best content anymore.
From here on out, it's citing the most recognizable content it can find fastest.
— Brad Bartlett
And you can't rank your way out of it anymore. Ahrefs analyzed 863,000 search results and found only 38% of AI Overview citations come from pages ranking in the top 10 — down from 76% a year earlier.
Nearly a third of citations now come from pages that don't crack the top 100 at all. So the reassuring "but we rank fine on Google" is no longer the defense it used to be. Ranking and getting cited have come apart.
What the visible companies do differently
Let me show you the other end of the spectrum, because it's the part that's actually buildable.
A client of mine sells software in a crowded category — the kind of space where a handful of legacy names have owned the search results for a decade. A year ago their content was doing what most good content does: sitting mid-page, behind the incumbents, technically fine and functionally invisible.
We didn't add volume. We added judgment.
Pillar-and-cluster content built around the exact terms a buyer types before they choose a vendor, and every piece written to sound like a sharp human peer instead of a brand account. Each page opened with a real answer, earned its authority instead of asserting it, and mapped to a question a buyer actually asks.
Then the client's research partner measured where Google's AI Mode actually pulls its answers from. The receipts:
They're the #1 cited brand versus every tracked competitor in all 10 of their core topics — often two to three times the nearest rival.
In 7 of those 10 topics, they're the #1 most-cited source on the entire web — ahead of Reddit, ahead of YouTube, ahead of national firms with real budgets.
Across roughly 10,700 sampled answers, Google's AI cited them in about 39% of them. Four in ten.
of Google AI answers in one client’s category cite them at least once
AI Mode citation share (Pillarbase). The #1 cited brand vs. every tracked competitor in all 10 core topics — and the #1 source on the entire web in 7 of them, past Reddit and YouTube.
But the number that tells you it's working isn't a number at all.
A property-management firm with more than 100,000 units under management emailed the client — not to buy anything, but to ask if they could be added to one of the articles. They'd decided that piece was the industry list worth being on.
Sit with that for a moment. A company that size, with its own marketing team, read a piece of content and concluded it was the authority. That's the market telling you who it recognizes.
Same category. Same warm sentiment. Two outcomes.
The only variable that moved was recognizability.
Reputation
Loved by customers — perfect sentiment
Content
Reads like the rest of the category
Google ranking
Fine — sits mid-page behind the incumbents
In the AI answer
Not named
“AI doesn’t cite the best content. It cites the most recognizable.”
Reputation
The same warm sentiment — no advantage here
Content
Distinct, human, built on proprietary proof
In the AI answer
#1 cited brand — named in ~4 of 10
The line between disappearing and getting chosen. What the cited side is built on:
The Citation Authority Flywheel: why recognition compounds
Here's the mechanism underneath that result, because I don't want you to think it was luck.
I call it the Citation Authority Flywheel.
Original data, off-site signals, and entity recognition feed each other — so each turn of the wheel makes the next citation more likely. Publish something distinctive that a model can't generate on its own, get recognized for it, earn a mention somewhere off your domain, and the machine starts treating you as the source for that topic.
That’s the flywheel part: Recognition begets citation, citation begets recognition. On, and on, and on.
But it’s voice that makes the loop legible. The same distinctiveness that makes a human reader think "a real person wrote this" is what makes a retrieval system treat you as an entity worth pulling from.
Flatten your voice to match the category and you don't just bore people — you erase the signal the machine uses to tell you apart from the forty other sources saying the same thing.
That's the difference between the invisible 44% and the client above. Same category dynamics. One built recognition on purpose. The other published into the void.
Are you in the invisible 44%?
Answer five honest questions. Your score reveals a verdict at the bottom — no email required.
1. Ask ChatGPT “best [your category] software.” Are you named in the answer?
2. Do your top pages open with a clear, quotable answer — not a category throat-clear?
3. Does your key content carry proprietary data, a named framework, or first-hand specifics a model couldn’t generate?
4. Are you corroborated off your own domain — reviews, a talk, video, earned mentions?
5. Have you tested your top buyer queries across ChatGPT, Perplexity, and Google in the last 90 days?
What to do if you're in the invisible 44%
If you suspect you're on the wrong side of this, here's where I'd start. Not hacks — there are no hacks left, and Google has spent its last two documentation updates saying so. Just the moves that actually move recognition.
Hold every page to the non-commodity bar
Before anything ships, ask the question Google now asks: could a model have produced this on its own? If yes, it can rank, but it will never be the citation. What clears the bar is a named position, proprietary evidence, and first-person specifics a machine can't invent.
Cover the buyer's whole journey on one deep page
Not a thin page per keyword. The AI splits a buyer's question into a spray of sub-questions and assembles the answer across them, so the page that resolves the whole cluster — each section leading with a clean, liftable answer — wins over a dozen shallow ones.
Be corroborable beyond your own domain
A brand that exists only on its own blog reads as unvalidated to every engine at once. Earned mentions, real reviews, a talk, a video walking through your framework — that's the corroboration layer the machines trust.
Make sure you really measure it
Test your top 15 to 20 buyer queries across ChatGPT, Perplexity, and Google's AI, and log where you're named, ignored, or beaten. Google Search Console now has a Generative AI performance report, so you can see it without guessing. You can't fix a visibility gap you've never looked at.
Four moves that build recognition
01
Hold every page to the non-commodity bar
Before it ships, ask what Google now asks: could a model have produced this? If yes, it can rank — but it will never be the citation. A named position and first-person specifics clear the bar.
02
Cover the whole journey on one deep page
Not a thin page per keyword. The AI splits a question into sub-questions and assembles across them, so the page that resolves the whole cluster — each section answer-first — wins.
03
Be corroborable beyond your domain
A brand that lives only on its own blog reads as unvalidated to every engine at once. Earned mentions, reviews, a talk, a video of your framework — that’s the corroboration layer machines trust.
04
Actually measure it
Test your top 15–20 buyer queries across ChatGPT, Perplexity, and Google’s AI. Search Console’s new Generative AI report shows it without guessing. You can’t fix a gap you’ve never looked at.
The honest caveat about all of this
One thing I won't do is sell you certainty I don't have.
A competitor is right behind that client on visibility. Authority in AI search is held, not won once — the flywheel spins for whoever keeps feeding it, and stopping is how you lose the lead you built.
This isn't a set-it-and-forget-it play. It's a compounding one, which means it rewards the brand that treats recognition as an ongoing practice, not a launch.
That's the honest version. And it's still a far better position than being one of the 44 companies the market likes and the machine can't find.
The honest caveat
A competitor is right behind that client on visibility. Authority in AI search is held, not won once — the flywheel spins for whoever keeps feeding it, and stopping is how you lose the lead you built. This is a compounding play, not a set-it-and-forget-it one.
Invisibility to AI isn't a reputation problem you fix with better messaging. It's a recognition problem you fix by building content distinctive enough that a machine reaches for you by name. The 44% aren't “bad companies” by any sense of the imagination. In fact, they may be far better than those showing up in AI.
They're simply less legible ones — and legibility is buildable, on purpose, starting with the next page you publish.
Volume is free now. Judgment isn't. The brands getting cited figured out which one the machines are actually paying for.
Sound human. Get cited.
Are you in the invisible 44% — or the cited few?
A new benchmark says nearly half of B2B SaaS is missing from the AI answers their buyers read. A GEO Visibility Audit tells you exactly where you stand: which queries name you, which skip you, and the fastest path to getting cited.
You get a scored content inventory and a prioritized fix list — specific enough to hand straight to a writer.
Book a GEO Visibility Audit →Fiverr Pro vetted · 4.9 stars · 1,600+ client reviews
Sound human. Get cited. · Made with 💙 in kcmo
Frequently Asked Questions
What does it mean for a B2B SaaS company to be “invisible to AI”?
It means AI assistants like ChatGPT, Gemini, and Perplexity rarely or never name the company when buyers ask for recommendations in its category. In DerivateX’s 2026 benchmark, 44% of 50 B2B SaaS brands scored below 50 out of 100 on AI visibility — despite near-perfect sentiment scores. The problem is retrieval, not reputation.
Why does my SaaS rank on Google but not show up in AI answers?
Ranking and citation have decoupled. Only about 38% of AI Overview citations now come from pages ranking in the top 10 (Ahrefs, 863K search results), down from 76% a year earlier. AI systems assemble answers from sources across a spread of related sub-queries and cite the most recognizable, distinctive source — not necessarily the top-ranked one.
How do I get my SaaS cited by AI tools like ChatGPT?
Publish content a model couldn’t have written for itself: a named point of view, proprietary data, and first-hand specifics — structured so each section leads with a clear, extractable answer. Then build corroboration beyond your own domain through earned mentions, reviews, and video. Recognition compounds; that’s the Citation Authority Flywheel.
Is being invisible to AI a reputation problem?
No. DerivateX found 44 of 50 companies earned perfect sentiment scores while visibility varied by 87 points between the top and bottom brand. Buyers don’t dislike the invisible companies — AI simply doesn’t retrieve them often enough to surface them.
Which AI platform is hardest to get cited by?
In the DerivateX study, ChatGPT and Gemini mentioned 100% of tested brands, while Claude mentioned only 88% — the most selective. Because a large majority of top-cited sources don’t overlap across engines, there’s no single “AI search” to optimize for. Each platform is its own channel with its own habits.
Written by
Brad Bartlett
Brad is a copywriter and content strategist who helps creators, brands, and organizations build content that's actually worth reading — and built to be found. He specializes in conversion-focused copy, brand voice, and SEO and AI search optimization, with a straightforward philosophy: great content has to be authentic before it can perform. He works comfortably across the AI content space, helping clients use the tools without losing the voice. Fiverr Pro vetted, 4.9 stars out of 5 across 1,600+ clients.