B2B SaaS: solid product, docs invisible to Google and AI

Sprint planning: SEO in the backlog — again

The CTO closed the Jira board. In “Marketing requests”, the SEO ticket had waited 4 sprints — always losing to bug fixes and feature releases.

Head of growth arrived with a screenshot: ChatGPT prompt “HR software for remote teams in Indonesia, what alternatives exist?” — five names appeared. Their product was not one.

“Not a fit problem. Developers researching don’t find us.”

CTO POV: machines must read our product

As a technical co-founder, the frustration was specific: we built good docs for users, but docs lived on Notion then moved to a subdomain blocked by robots.txt. Engineering shipped API reference — not connected to product entity on the marketing site. WordPress landing and React app were two worlds.

This was not a “need more blog” problem. It was infrastructure discoverability — like shipping an API without documentation. We shipped product without documentation for search engines.

What they had tried (engineering lens)

Public Notion docs — not indexed. SEO plugin on marketing — docs subdomain untouched. Six generic blog posts — no links to features or docs. Changelog on Twitter — unstructured.

Growth wanted ad budget. CTO wanted foundation first. Deadlock — until the co-founder saw competitors with indexed docs and llms.txt appearing in AI answers for the same category.

Audit findings: robots.txt blocking our best asset

Familiar for early-stage SaaS:

140 docs pages blocked — robots.txt from old setup protecting paths that became production docs.

Entity split — marketing said “Acme HR”, docs said “Acme Platform”, generic Organization schema without SoftwareApplication.

LCP 5.1s on pricing — unoptimized hero, blocking analytics scripts.

No llms.txt — in the AI search era, engines had no official map of features, pricing, and product boundaries. Competitors did.

CTO decision: two-week SEO foundation sprint before the next feature. Engineering resisted — “not core product.” Counter: unindexed docs = unnecessary support tickets + zero discovery.

What we ran (Oura Starter — technical lens)

Starter for SaaS is not marketing articles:

1. Robots & sitemap — open priority docs indexing, exclude internal API admin.
2. CWV — pricing LCP, lazy load, critical CSS path.
3. llms.txt — feature, pricing, FAQ, changelog URL map for AI crawlers.
4. Schema — SoftwareApplication + Organization + pricing FAQPage.
5. Entity merge — one product name, sameAs, consistent across marketing/docs/changelog.
6. GA4 AI-referral — developers from chatgpt.com/perplexity separate from paid.

Every production change through dry-run — CTO would not let SEO break the deploy pipeline.

How to read the numbers below

Typical patterns for B2B SaaS. Starter engagement patterns:

• Docs indexing usually moves 2–4 weeks after robots fix + sitemap submit.
• AI category citations need llms.txt + FAQ + comparison pages — months 2–3.
• Developer AI-referral often visible before competitive keyword rankings.

After foundation, Oura Content for comparison pages + Growth for OAVF — on docs machines can already read.