I Audited a Series C ESG SaaS Platform.
They Blocked the AI Crawlers Their European Buyers Use.

This company is one of the more cited names in mid-to-enterprise ESG and CSRD reporting. Fortune 500 customers across pharma, telco, banking, aerospace, retail, and energy. Three consecutive years of Gartner Market Guide recognition. A 2024 Technology Fast 50 award. A recent Series C of $37.7M CAD with a major Montreal venture firm leading and the existing investor base re-upping. The headline use of funds: European expansion ahead of the next CSRD reporting cycle.

And then I pulled their robots.txt.

The file actively blocks GPTBot, ClaudeBot, Google-Extended, CCBot, Bytespider, Amazonbot, FacebookBot, and Twitterbot. It allows ChatGPT-User, PerplexityBot, OAI-SearchBot, and Applebot. The intent is the sensible 2024 stance: protect the training corpus, allow live citation. The side effect in 2026 is that the company is suppressing its own visibility in the exact AI tools its European procurement buyers are using to build the shortlist.

The Scores

An overall weighted score of 50/100. The AI Discoverability score of 22 is one of the lowest I have given to a Series C SaaS, and the gap is entirely self-inflicted through the robots.txt configuration. Strong content, strong customer base, strong stack, and the AI front door is bolted shut.

Finding #1: The robots.txt Blocks the Training Crawlers Your European Buyers Use

When a sustainability director in Frankfurt or Paris with a CSRD obligation opens Perplexity and types "best CSRD reporting platform," Perplexity does two things at once. It retrieves live web pages with PerplexityBot (which this company allows, fine) and it consults its trained entity model of "the CSRD platform category" (which Perplexity, like every other foundation model, builds in part from Common Crawl, which this company blocks).

Finding #2: /llms.txt Returns the Homepage HTML

I requested https://[domain]/llms.txt. The server returned 200 OK with the full homepage HTML as the response body. To an AI agent, this reads as "the file does not exist." For a Series C SaaS in an AI-discovered regulated category, the missing /llms.txt is a daily lost-citation event.

A properly structured /llms.txt for this company would name the platform, the four product modules, the supported frameworks (CSRD, ESRS, SFDR, EU Taxonomy, GRI, SASB, CDP, TCFD), the named Fortune 500 customer roster, the Series C anchor, and the Gartner Market Guide recognition. That single file, served as text/plain at the root, materially changes how AI agents describe the company. Most competitors do not ship one. The ones that do get cited more. More on AI crawler directives and llms.txt.

Finding #3: No SoftwareApplication Schema. No FAQPage. No Person.

The homepage carries one JSON-LD block: WebSite, LocalBusiness (with the Montreal HQ address), BreadcrumbList, and an empty placeholder Organization block (no name, no URL, no logo). That is it.

The site has the strongest customer roster of any platform in its mid-tier AI shortlist. The schema layer that surfaces that roster to AI is the layer that has not been built. That is the single biggest reason AI keeps describing the company as "mid-market." More on schema markup for SaaS.

Finding #4: The Pricing Page 302-Redirects to the Homepage

The natural URL a buyer or an AI agent tries for the pricing page returns 302 to the homepage. There is no /pricing page anywhere on the site. AI tools repeatedly answer pricing queries with "custom quote, contact sales," which routes the buyer to whichever competitor has any disclosed pricing signal. A pricing page does not need to publish a number to be useful; it needs to disclose the model (per-entity, per-report, per-user, ARR floor) so the AI has something to cite.

Finding #5: Cited in 3 of 8 Listicles, Always as "Mid-Market"

The company appears in three of the eight major "best CSRD platform 2026" listicles I checked. In every one, the framing is mid-tier ("closest to a full CSRD solution in the mid-market") while platforms with fewer Fortune 500 customers are framed as enterprise leaders. The listicle authors are competitors or analyst-adjacent publications; the framing they use is what the AI then cites.

The fix is two layers. First, repair the entity graph (Findings #1 through #3) so the AI has a corrected source to anchor against. Second, outreach to the EU-focused listicles (KEY ESG, Tanso in Munich, Coolset, Safaqes) for re-positioning, supported by the named-customer evidence and the Series C announcement. Most listicle authors update their rankings on request when the supporting evidence is strong enough. More on review platforms AI trusts.

What is Actually Working

The Takeaway for Regulated-Category SaaS

CSRD reporting is the textbook AI-discovered category. The buyer has a regulator-imposed deadline. The platform decision is a board-visible procurement. The two-minute initial scan happens inside an AI tool before any human at the vendor knows the deal exists. Being absent from that initial scan, or being framed wrongly inside it, is the new version of being absent from the search results page.

This audit pattern repeats across regulated-category SaaS. Strong product, strong customer proof, strong content engine. The structural layer that decides whether AI recommends them is built for 2022 search behavior, not 2026 AI procurement behavior. The fixes are configuration and schema, not product. The hard part is already done. The missing piece is making the proof legible to the systems that decide the shortlist.

Frequently Asked Questions