by Jay Singh ยท August 24, 2026 ยท AI visibility ยท 7 min read

AI Citation Study: What Gets Cited for B2B SaaS Buying Queries

We ran an AI citation study across 18 real B2B SaaS buying questions and logged every source that came back. G2 and Capterra, the review platforms most SaaS marketing teams have spent years building profiles on, surfaced in exactly two of them. What filled the gap instead is the part worth paying attention to.

The method

Six categories a B2B SaaS marketer genuinely searches. All of this was done on the free tier of Google Gemini, OpenAI’s ChatGPT, and Anthropic’s Claude. Four queries drafted per category, 18 run to completion with every source logged and categorised. The remaining six were near-duplicates of queries already covered, so they’re excluded rather than counted twice.

Each source was tagged by what it actually is:

TagSource typeExample
REVReview aggregatorG2, Capterra
VENVendor citing its own blogZapier writing about Zapier alternatives
AGYAgency ranking itselfAn agency’s own “best agencies” post
INDIndependent or niche publicationA small SaaS or marketing blog
PUBRecognised publicationTechRadar, Forbes Advisor
COMCommunity sourceReddit, DEV Community

The 18 queries, and what came back

Category 1: CRM (4 of 4 run)

QueryDominant source typesReview sites present?
Best CRM for B2B SaaS startupsVEN, INDNo
Best CRM for a 50-person sales teamVEN, IND, COMNo
HubSpot vs Salesforce for a growing SaaS companyVEN, INDNo
Best CRM alternatives to SalesforceVEN, INDNo

Category 2: Workflow automation (3 of 4 run)

QueryDominant source typesReview sites present?
Best workflow automation tools for tech teamsVEN, INDNo
Best Zapier alternativesVEN, IND, REVYes โ€” Capterra and G2
Make vs Zapier for a SaaS marketing teamVEN, INDNo

Category 3: AEO and AI visibility (2 of 4 run)

QueryDominant source typesReview sites present?
Best AEO tools for B2B SaaSVEN (heavily)No
How do I know if ChatGPT is citing my websiteVEN, INDNo

Category 4: Content marketing agencies (2 of 4 run)

QueryDominant source typesReview sites present?
Best B2B SaaS content marketing agenciesAGY (heavily)No
Best SEO agency for B2B SaaSAGY, INDNo

Category 5: Keyword research tools (4 of 4 run)

QueryDominant source typesReview sites present?
Best keyword research tools for SEOIND, PUBNo
Ahrefs vs Semrush for a small marketing teamVEN, IND, REVYes โ€” G2
Best free keyword research toolsIND, PUBNo
Best keyword research tools for B2B SaaS contentINDNo

Category 6: AI writing tools (3 of 4 run)

QueryDominant source typesReview sites present?
Best AI tools for writing SEO content (small teams)VEN, INDNo
Jasper vs Copy.ai for B2B contentIND, VENNo*
Best AI content tools that don’t sound roboticINDNo

* One source quoted G2 star ratings inside its own article without G2 appearing as a cited source. More on that below.

Finding 1: review sites appeared twice, and only on one query shape

Query typeQueries runReview sites appeared
Comparison and alternatives (“X vs Y”, “X alternatives”)52
Category listicles (“best X for Y”)130

Both review-site appearances were the same shape of question: a direct comparison, or an alternatives search, inside a mature software category where G2 and Capterra have years of accumulated review data.

  • “Best Zapier alternatives” returned Capterra’s Zapier alternatives page and a G2 competitors page.
  • “Ahrefs vs Semrush for a small marketing team” returned G2’s head-to-head comparison page, complete with star ratings and review counts.

In all 13 category-listicle queries, review platforms didn’t appear at all. Not for “best CRM for B2B SaaS startups.” Not for “best B2B SaaS content marketing agencies.” Not for “best SEO agency for B2B SaaS.” These are exactly the categories where G2 and Capterra dominate traditional Google results.

The practical read: review-platform authority appears to transfer to AI answers on comparison and alternatives queries, and not on category listicles. If getting cited by a specific engine is the goal, our guide on how to get cited in Perplexity covers the structural side of this in more depth.

AI citation study - Comparison Alternatives over Best X for Y.

Finding 2: vendors and agencies are citing themselves, constantly

The sources filling those 13 listicle queries follow a clear pattern.

Content marketing agency queries were the most extreme case of self-ranking. Nearly every result was an agency publishing its own “best agencies” list:

AgencyPublishes its own “best agencies” rankingAppears on it
AnimalzYesYes
Siege MediaYesYes
Omniscient DigitalYesYes
Grow and ConvertYesYes
RevenueZenYesYes, ranked first
First Page SageYesYes, ranked first

A buyer reading these answers is largely reading agency marketing arranged in ranked order. Several disclose the conflict openly. Some don’t.

AEO and AI-visibility tool queries ran the same pattern even harder. Nearly every citation for “best AEO tools for B2B SaaS” pointed to an AEO vendor’s own blog ranking the category it competes in: Conductor, AirOps, Omnibound, Indexly, and others, each appearing in the list they published.

CRM and workflow automation mixed vendor-owned blogs (Zapier writing about Zapier alternatives, Close writing about the best CRMs for SaaS, folk writing about the best CRMs for sales teams) with a long tail of smaller independent publications.

Finding 3: G2’s data gets laundered through third-party blogs

One nuance worth separating out. On the Jasper vs Copy.ai query, a cited source included the line “G2 scores: Jasper 4.7/5, Copy.ai 4.6/5″ inside its own article.

G2’s data made it into the answer. G2’s domain did not get cited. The review platform did the work of collecting those scores and a third-party blog captured the citation.

That’s a meaningfully different failure mode from simple absence, and it suggests review data can still influence AI answers indirectly while the platform itself gets no visibility from it.

Finding 4: the citation pool is mostly small sites

The most consistent characteristic across all 18 queries wasn’t a specific source. It was how thin the authority requirement appears to be.

Most cited sources weren’t major publications or established review platforms. They were small SaaS blogs, niche marketing sites, and tool vendors’ own content marketing, sites with modest domain authority publishing well-structured, current, specific comparison content.

Recognisable third-party publications appeared, but rarely: TechRadar, Forbes Advisor, a Gartner peer-review page, one Reddit thread, one DEV Community post. Real, but heavily outnumbered by the long tail.

That’s the encouraging finding for a smaller brand. The barrier to being a cited source here is not “outrank G2.” It’s closer to “publish something specific, current, and structured enough to be quoted.” That structural work is exactly what our AEO tools roundup evaluates tools against.

AI citation study - Citation Pool is mostly small sites

What this means for your AI visibility strategy

Comparison and alternatives pages are the highest-leverage format. They’re the query type where review platforms surface, which means they’re also where a strong owned comparison page competes directly against the aggregators rather than being buried under them. Our keyword research tools guide is one example of this format built deliberately.

Your own blog is a legitimate citation source, and your competitors already know it. The vendors dominating these results are cited because they published the category roundup. If you sell in a category and haven’t published a serious, honest roundup of it, someone else has.

Domain authority matters less here than structure and specificity. The citation pool is full of small sites. What they share is clear formatting, current information, and specific claims worth quoting.

Don’t abandon your review profiles, but don’t over-credit them. They earned their two appearances on genuine comparison intent, and their data reached a third answer indirectly. That’s a real slice of buyer research, just narrower than most teams assume.

Method notes and limits

  • 18 of 24 planned queries were completed with full source logging. The other six were near-duplicates of covered queries.
  • Category coverage is uneven. CRM and keyword research were fully covered at 4 queries each. AEO and agencies were covered at 2 each.
  • Results reflect one collection window. AI search results shift; a re-run in three months would not produce an identical source list.
  • This measures which sources get surfaced and cited, not whether being cited drives traffic or pipeline. Those are separate questions this dataset doesn’t answer.
  • Categories were B2B SaaS marketing and sales tooling. Findings may not transfer to verticals where review platforms have deeper coverage.

Where to go from here

For the fuller picture on the categories tested here, our AEO tools roundup for B2B SaaS goes deeper on AI-visibility platforms, our workflow automation tools roundup and keyword research tools guide cover two of the categories above, and our B2B SaaS content marketing agencies roundup is our own entry in exactly the self-ranking category described in Finding 2, with that conflict disclosed in the piece itself.

Want this run for your category?

This is a repeatable method, not a one-time snapshot. If you want to know which sources AI engines surface for your buyers’ actual queries, and where your brand sits in that pool, we can run it.

Think B2B is boring?

That's coz you are not subscribed to ours newsletter. We share a 100 bad days (of B2B marketing) made a 100 good stories. A 100 good stories make YOU interesting at parties.

Jay Singh writes most of what you read here โœ

Jay is the co-founder of LymLyt and an AI partnership consultant helping B2B SaaS companies grow through content strategy, SEO, and AI integration. He writes about content marketing, what makes B2B content convert, and how SaaS teams can build content systems that drive real pipeline.