How a B2B Analytics SaaS Grew AI Referral Traffic 500% in 90 Days

A Series B analytics SaaS (ARR ~$18M, ~40 employees) came to us in December 2025 after noticing their competitors appearing in ChatGPT product recommendation answers — while they did not. Their traffic from AI sources was negligible: fewer than 200 visits per month across ChatGPT, Perplexity, and Gemini combined.

An initial AEO scan found three critical gaps:

• Robots.txt blocked GPTBot and ClaudeBot — a legacy of a 2023 policy decision nobody had revisited.
• No FAQ schema anywhere on the site — including on product and comparison pages that were natural citation candidates.
• No visible "last updated" timestamps on any blog post. Content was fresh but looked stale to crawlers.

The 90-day plan had three phases:

Phase 1 — Technical (Week 1)

• Removed blanket Disallow rules for GPTBot, OAI-SearchBot, ClaudeBot, and PerplexityBot
• Published an llms.txt file at the site root listing the 30 most important pages
• Added Organization and WebSite JSON-LD schema to the site template

Phase 2 — Content structure (Weeks 2-6)

• Rewrote 20 top-of-funnel pages to open each H2 section with a 2-sentence answer capsule
• Rephrased section headings as questions ("How does X work?" instead of "Product overview")
• Added FAQPage schema with 5-8 Q&A pairs on each rewritten page
• Added visible "Last updated" timestamps beside every title

Phase 3 — Authority (Weeks 7-12)

• Published a 4,000-word proprietary data report from the company's own usage metrics
• Pitched the report to 12 industry publications — got 4 placements with backlinks
• Added a customer case study library with real metrics and named enterprises

Total AI referral traffic went from 185 visits in November to 1,106 visits in February — a 5.98x increase. Conversions from AI traffic outperformed paid search channels, with a trial signup rate of 6.1% compared to 1.4% for paid.

Looking at the weekly data, three changes produced disproportionate lift:

1. Unblocking crawlers — Week 1 alone saw a 12x increase in AI crawler visits in server logs. This was literally a five-line change in robots.txt.
2. Answer capsules on the comparison pages — the "X vs Y" pages were the first to earn Perplexity citations, typically within 2-3 weeks of the rewrite.
3. The proprietary data report — when the report was picked up by a major industry publication in Week 10, ChatGPT citations for related queries jumped from 11 to 19 within 4 weeks.

• The easy wins are really easy. Three robots.txt lines and an llms.txt file accounted for the first 30% of the growth.
• Answer capsules move the needle fastest. New content took 4+ weeks to earn citations; retrofitted capsules on existing content earned citations in 2-3 weeks.
• Third-party mentions compound. The four industry-publication backlinks from the data report continued driving AI citations months after publication.
• Measurement is manual. There's no single dashboard for AI citations yet. Expect to set aside 30-60 minutes per week for manual audits across ChatGPT, Perplexity, and Gemini.