How a B2B SaaS brand earned ChatGPT citations in 90 days.

Context.

A mid-market B2B SaaS brand operating in a workflow-automation category. They were ranking well on traditional Google — page 1–2 for most relevant keywords — but had no presence in AI search answers. When buyers asked ChatGPT or Perplexity "best workflow automation for [use case]", the brand wasn't mentioned. A competitor with weaker traditional SEO was being cited consistently.

The brand's assumption: more SEO content would fix it. The actual problem: the existing content lacked the structural signals AI extraction systems use to identify authoritative sources. Volume wasn't the gap; structure was.

The diagnostic.

We started with the standard 30-minute AI visibility audit. Findings:

• No entity disambiguation. The brand had a Wikipedia-style mention count but no consistent entity description across the surfaces AI systems triangulate (about page, schema, sameAs, Knowledge Graph).
• Content optimised for keyword density, not answer extractability. Pages buried direct answers in paragraph 5 of long intros. AI extraction prefers clear answer-first paragraphs near page top.
• Thin schema implementation. Organization schema present; Article, FAQPage, HowTo, and Product schema either missing or incomplete. AI systems use these signals heavily for citation confidence.
• No FAQ infrastructure on category pages. Their two main category pages had no FAQ section, no FAQ schema, and no answer-first content blocks. The two formats AI systems most reliably cite.

The structural work.

The 90-day engagement focused on four workstreams running in parallel:

• Entity reinforcement. Rewrote the About page with a clear, structured entity definition. Updated Organization schema with comprehensive sameAs, knowsAbout, and areaServed properties. Added speakable markup where applicable.
• Content restructure on top 15 pages. Moved direct answers to page-top "Quick Answer" blocks. Added FAQ sections to every category and product page with full FAQPage schema. Rewrote intros to lead with answers rather than context.
• Schema graph upgrade. Added Article schema to blog content with named author bylines, datePublished, and dateModified. HowTo schema on tutorial content. Product schema on product pages. BreadcrumbList on every non-home page.
• Authority infrastructure. Built author pages with credentials, sameAs links to LinkedIn and conference talks, and explicit expertise statements. The E-E-A-T layer that AI systems use to verify whether a source is citation-worthy.

What changed.

The first documented ChatGPT citation appeared on day 47 of the engagement — for a mid-funnel category query the brand had previously been invisible on. By day 90, branded and category citations were appearing routinely across both ChatGPT and Perplexity. Google AI Overviews picked up the brand on 12+ distinct category queries within the same window.

The compounding effect on paid CAC was the surprise. As AI citations began substituting for paid acquisition on top-of-funnel category queries, blended CAC dropped roughly 22% over the quarter — without any change to paid spend or targeting. The organic substitution effect is real and underappreciated.

What this proves about the discipline.

AI search citation isn't a content volume problem. It's a structural problem. The brands that get cited have done specific structural work — entity signals, schema depth, answer-first content, E-E-A-T infrastructure — that most competitors haven't. The brands that don't? Usually they've written plenty of content; the structure just isn't there.

This is the work GliscoLab does. The structural pattern shown here is what we apply on every AI SEO engagement.