Case study: Gumlet turned ChatGPT mentions into 20% of inbound revenue. Read it →
TL;DR
Most B2B SaaS teams still treat “AI search” as a content rewrite problem on their own domain. A fresh supplier-selection study says the answer engines are often not reading your homepage when a buyer asks who to hire.
They are reading directories.
That is uncomfortable if your GEO program is mostly blog posts and a pricing page. It is also the opening for an operator stack that most agencies will not copy cleanly: map the lists that actually get cited, earn structured presence there, then prove it in CRM.
This piece is DerivateX’s read of the Sector Directories Are Driving AI Overview Citations study, what to trust, what to discount, and what to do this month.
Citations.press ran 3,850 commercial prompts between June 10 and August 22, 2026 across four surfaces:
The focus was supplier-selection style prompts (category and provider recommendation asks), not general informational queries. That matters. A “what is payroll software” answer and a “recommend a payroll platform for 200-person US SaaS” answer pull from different webs.
Headline findings from the published study:
| Finding | Number |
|---|---|
| Directory share of citations on supplier-selection prompts | 41% |
| Brand-owned page share on the same prompt set | 18% |
| Brand homepage appearances overall | 14% |
| Brand homepage appearances on category (not company-named) prompts | 6% |
| Engineering: directory citations vs brand-site citations | 4.2 to 1 |
| Directory citations from list pages (not profiles) | 68% |
| Structured one-entity entries vs free-text listings | 3.1× citation rate |
Citation rates for sector directories on category prompts, by surface:
The study also reported a weak relationship between a directory’s domain authority and citation rate (Spearman ρ = 0.19). Coverage depth inside a narrow sector predicted citation better than broad high-authority directories with shallow category pages.
Overlap between the cited source sets across the four platforms was only 31%. Same buyer intent, different “webs” per engine.
Citations.press is transparent that it publishes AI citation research and operates the sector directories used as the study’s test set.
That does not make the data useless. It does mean you should not treat “directories win” as a universal law proven on an independent sample of the entire web. It means:
DerivateX’s first-party vs third-party citations and third-party assets for AI search pieces already argued that brand sites alone under-explain recommendation answers. This study is another datapoint in that direction, with the caveat above baked in.
When a buyer names a company, brand pages get a fairer fight. When a buyer names a category, engines need a compact, comparable set of entities. List pages are built for that job.
That is why 68% of directory citations in the study came from list pages, not profile pages, and why structured attributes beat free-text blurbs.
It also lines up with DerivateX’s AI Overviews citation research: commercial software answers lean on third-party lists and formats that look like evidence, not brochure copy.
If your “AI content plan” is still “rewrite the homepage for ChatGPT,” you are optimizing the page that showed up in 6% of category-prompt responses in this dataset.
Write 25 to 50 prompts a real buyer would ask an assistant:
Run them logged out across ChatGPT, Perplexity, Claude, Gemini, AI Overviews, and AI Mode on a fixed cadence. Log named brands, linked sources, and whether you appear.
From the panel, pull every domain that powered a recommendation. Cluster them:
Prioritize presence and accuracy on the short list that already appears, not every DR90 site on a media kit.
The study’s strongest operational hint is structural: one entity per entry, consistent attributes, list pages that models can parse. For SaaS that usually means:
You still need pages that third parties and models can quote: comparison pages, use-case pages, integration proof, methodology, and case studies. The point is sequencing. Off-site list presence without citable first-party proof is empty. First-party proof without off-site distribution often stays invisible on category prompts.
A directory mention that never shows in demos is a press clipping. Connect AI-influenced sessions and assisted opportunities in CRM the same way you would for any other channel. Gumlet is the public reminder that revenue attribution beats vanity visibility.
If you want a baseline of where you already appear in AI answers before a retainer conversation, start with the free AI visibility audit. Engagement options stay on pricing.
On 3,850 supplier-selection prompts across four AI surfaces, sector directories accounted for 41% of citations versus 18% for brand-owned pages, with list pages driving most directory citations.
No. It means category recommendation answers often pull from structured third-party lists. Your site still needs to be citable; it often will not be the primary shortlist source on “recommend a provider” prompts.
Citations.press operates the directories used in the test set. Directional findings can still be useful, but you should validate on your own category and engines.
Perplexity (47% on category prompts), then Google AI Overviews (39%), Google AI Mode (36%), and ChatGPT (22%).
Run a supplier-prompt panel, extract the list domains that appear, and fix structured presence on those domains before launching a broad “get on every directory” campaign.
It reinforces first-party vs third-party citation dynamics and the off-site work described in third-party assets for AI search, with a fresh 2026 supplier-prompt datapoint.
Shivanshi Bhatia is the co-founder of DerivateX, a B2B SaaS SEO and Generative Engine Optimization agency that engineers AI citations in ChatGPT, Perplexity, Claude, and Gemini and connects them to demo bookings and revenue pipeline. She runs operations and delivery, which means every audit, content brief, and published page ships through a system she built. She owns the client relationship from kickoff through reporting, so clients spend their time on decisions instead of chasing updates. She has worked in SaaS since 2019 and reviews client work before it goes live.
SEO Growth & Operations at DerivateX, where I focus on SEO, content publishing, AI search optimization and organic growth. Through my blogs, I share practical insights on SEO, AI tools, content strategy, digital marketing and workflows that help businesses improve search visibility and grow sustainably.
