How an SMB LMS platform went from absent in AI answers to cited by Google and ChatGPT in two months

The challenge: buyers were shortlisting in AI, and the platform was on no list

Buyers were building LMS shortlists inside AI assistants before opening a vendor website. We tested 307 buyer questions across five AI engines. The four incumbents appeared in 37% to 76% of the answers. The client appeared in none.

Organic search told the same story, at around 100 sessions a month. Four structural gaps sat behind the numbers:

• Coverage gap. Of roughly 51,930 monthly searches in the category, 38,340 had no page on the site that could rank, even in principle.
• Demand arriving, then leaving. 25,054 search impressions produced 27 clicks, a click-through rate of about 0.1%.
• Stuck in striking distance. 24,840 impressions sat at positions 8 to 15 and converted at 0.10%.
• Site health. Technical issues across 15 categories capped what could rank at all.

Incumbents win head terms on a decade of authority. A smaller platform had to win somewhere else.

What we did

Three workstreams with one content engine underneath. SEO earns the position, GEO gets the brand recommended when buyers ask an AI assistant, and AEO gets it quoted inside the answer itself.

Step 1: Lock the baseline and diagnose

• A full technical audit of 500 pages, with every issue tied to a specific URL.
• A competitive AI baseline: 307 buyer questions tested against five competitors across five AI engines.
• 157 keywords researched, with 76 mapped to the nine core service pages.
• 73 buyer prompts researched and 67 approved across 14 page groups, tracked with 335 checks a month.

The July baseline was locked as the first deliverable, so every result below is measured against it.

Step 2: Build the architecture AI engines can retrieve from

• Eight topic clusters, not a list of blogs. Each blog cluster links to its matching service page, concentrating depth where it builds authority.
• A /compare/ page group for alternatives searches, the highest-intent territory for buyers weighing vendors.
• A /resources knowledge hub for short, question-shaped answers.
• Industry pages held back deliberately until vertical research is complete.

Step 3: Route every prompt to a page and a treatment

Not every buyer question needs a new article. Of the 73 prompts researched, 41 became FAQ blocks on existing pages, 12 became new pieces, 14 became knowledge hub pages, and 6 were held. Refusing to publish thin content was part of the method, and it kept GEO affordable.

1. 01

   Answer-first structure

   The direct answer within the first 60 words.

2. 02

   Question-shaped H2s

   Headings that match the way buyers actually phrase prompts.

3. 03

   Validated schema

   FAQPage, Article, Product, Organization and Breadcrumb.

4. 04

   Entity consistency

   Identical brand facts across the site and Organization schema.

5. 05

   Striking-distance programme

   Targeting the 24,840 impressions the site already held.

The first publishing wave shipped 20 blog posts and 10 page optimisations, from a production engine with a ceiling of 40 pieces a month split between blog and LinkedIn.

Results: from invisible to ranked and cited in one phase

Organic search also became the platform's best-performing channel, with users up 27.7% and an engagement rate of 60.8%, against 32.7% previously. Five new blog pages entered the site's top 10 landing pages from a standing start.

Asked which content formats the platform supports (SCORM, video, podcasts, assignments), Google AI Overviews, ChatGPT, Perplexity and Gemini all named the platform and cited its own pages as the source.

Phase 2 is now in production, with 40 topics finalised and expansion into priority industry verticals under way.