Case Study: How a B2B SaaS Brand Increased AI Visibility by 340% in 6 Months

DataFlow is a B2B project management SaaS platform targeting mid-market technology companies with 50-500 employees. Despite solid product ratings on G2 (4.3 stars, 280 reviews) and healthy Google organic traffic, DataFlow's marketing team noticed something alarming: when they asked ChatGPT, Perplexity, or Gemini "what are the best project management tools for mid-market tech companies?", DataFlow's name never appeared. Not once, across any of the five AI models they tested. Their starting AI visibility score: 12 out of 100.

The starting point: nearly invisible to AI assistants

DataFlow's initial AI visibility audit revealed the scope of the problem. Across 80 category-level prompts tested on five AI platforms, DataFlow was mentioned in just 9 responses — a 2.2% mention rate. Competitors with similar product ratings and market position were appearing in 35-60% of the same prompts. The AI citation gap was severe.

The audit also revealed the root causes. DataFlow had no Wikipedia article — AI models literally had no high-authority entity definition to reference. Their Wikidata entry was empty. Their Organization schema was incomplete (no sameAs URLs, no category specification). Their website had 45 blog posts, but none were structured as definitional or FAQ content. And their press coverage, while existing, was concentrated in one niche trade publication rather than distributed across multiple authoritative sources.

The gap was large but the causes were clear. A structured GEO programme was designed in three phases.

The GEO audit: diagnosing the problem

The full GEO audit framework was applied systematically. Entity recognition check: AI models either said they had no information about DataFlow or provided a vague, sometimes inaccurate description ("a startup in the productivity space" — DataFlow had been operating for six years with $12M ARR). AI mention audit across 80 prompts, 5 models: 9 mentions total, all in brand-specific prompts ("what do you know about DataFlow?"), none in category discovery prompts. Sentiment: the 9 mentions were neutral at best — no positive framing. Technical audit: missing Organization schema, no structured data on any blog posts, robots.txt inadvertently blocking Bingbot (critical for Perplexity and ChatGPT browsing mode). Content gap analysis: zero definitional content, zero FAQ content, zero comparison guides. Off-site audit: no Wikipedia, incomplete Wikidata, only 1 significant press mention in 12 months.

The diagnosis was clear: DataFlow had an entity recognition problem compounded by a content authority problem compounded by a technical access problem. All three had to be fixed simultaneously — fixing content without fixing entity recognition would yield minimal results, and vice versa.

Phase 1 (months 1-2): entity and technical foundations

Phase 1 focused entirely on establishing the infrastructure that GEO requires. The actions taken:

• About page rewrite: The About page was rewritten from a brand narrative ("we believe in the power of collaboration") to a precise entity brief: "DataFlow is a B2B project management platform founded in 2019 that provides cross-functional workflow management for technology companies with 50-500 employees. DataFlow is headquartered in Austin, Texas, and is used by over 2,000 teams across 40 countries." Every word was chosen for entity clarity.
• Organization schema: Full Organization schema was deployed with sameAs linking to LinkedIn, Crunchbase, and G2. The schema included founding date, employee count, industry classification, and primary contact information.
• Wikipedia and Wikidata: A Wikipedia stub article was created, meeting notability guidelines with citations from DataFlow's existing press coverage. Wikidata was populated with entity attributes.
• Technical fixes: The robots.txt error blocking Bingbot was corrected. Canonical URLs were standardised. Page load times were reduced from 3.2 seconds to 1.1 seconds.
• G2 profile enhancement: DataFlow's G2 listing was updated with more precise category placement and improved attribute descriptions.

By the end of month 2, AI visibility score had increased from 12 to 21 — a 75% improvement from foundation work alone. Entity recognition checks showed ChatGPT and Gemini now describing DataFlow correctly as a project management tool for technology teams.

Phase 2 (months 3-4): content authority build

With entity foundations in place, Phase 2 focused on building the content library that AI systems could cite. The content strategy was prompt-first, derived from a bank of 60 AI prompts identified as high-priority for DataFlow's target market. Fifteen pieces of content were published over eight weeks:

• Five definitional articles: "What is cross-functional project management?", "What is a project workflow system?", etc. Each included FAQPage schema and comprehensive author bios.
• Three statistical reports: "State of Remote Project Management" (original survey of 400 tech managers), quarterly trend reports. These became frequently cited data sources.
• Four comparison guides: "Asana vs Monday.com for tech teams", "DataFlow vs alternatives for mid-market", "Best project management tools by team size", "Best tools for engineering-led companies".
• Three FAQ hub pages targeting top discovery prompts: "How to choose project management software", "Project management for remote engineering teams", "Cross-functional project management best practices".

All 15 pieces were published with Article schema, linked to the updated About page, and submitted to Bing Webmaster Tools. By month 4, AI visibility score had climbed to 38 — with DataFlow now appearing in 28% of category discovery prompts on Perplexity (which responds quickly to new content) and 18% on ChatGPT.

Phase 3 (months 5-6): off-site authority and press

Phase 3 focused on third-party authority signals — the external citations that give AI models confidence in DataFlow as a credible entity. The actions taken in months 5-6:

• Press outreach: DataFlow's "State of Remote Project Management" survey data was pitched to four major publications. TechCrunch published a 600-word article citing the research, with a link to DataFlow's report. Product Hunt featured DataFlow as a Product of the Week. Two SaaS-specific publications ran profiles.
• Podcast appearances: The CEO appeared on three B2B SaaS podcasts with combined monthly listeners of 80,000, all of which published full transcripts.
• Industry association: DataFlow joined the Project Management Institute (PMI) as a corporate partner, earning a listing in PMI's partner directory.
• G2 review campaign: A customer success initiative generated 47 new G2 reviews, improving review volume and adding more specific use-case descriptions.

By the end of month 6, AI visibility score reached 53 out of 100 — a 340% increase from the starting point of 12. DataFlow appeared in AI responses for 41% of category discovery prompts across all five models, with positive sentiment framing in 78% of mentions.

Results: 340% AI visibility increase in 6 months

The headline metrics from the 6-month programme:

• AI visibility score: 12 → 53 (340% increase)
• Category discovery mention rate: 2.2% → 41%
• Positive sentiment rate: 22% → 78%
• Perplexity citation rate: 0% → 54% (highest of all models, reflecting Perplexity's retrieval-based architecture and responsiveness to fresh content)
• Direct website traffic: +22% over the period (attributable to AI-driven brand discovery)
• Demo request volume: +17% (partially attributable to AI visibility improvement)

Key lessons for other SaaS brands

The DataFlow case study illustrates several broadly applicable GEO principles:

1. Fix entity recognition first: No amount of content investment will move AI mention rates if AI models don't have a clear, accurate entity definition for your brand. The About page and Organization schema are the starting point, not an afterthought.
2. Content structure matters as much as content volume: 15 well-structured, schema-marked, prompt-aligned articles outperformed DataFlow's existing 45 generic blog posts for AI citation purposes.
3. Perplexity responds fastest: Of all AI models, Perplexity showed the most rapid response to new content — citation rates improved within 2-4 weeks of publishing strong content. Use Perplexity as your leading indicator of content impact.
4. Third-party authority compounds: The press coverage in month 5 created a multiplier effect on all previous work. Entity recognition improved across models that hadn't updated quickly from the content phase.
5. Measure continuously: Without systematic measurement using Sight's AI visibility dashboard, it would have been impossible to know which actions were driving improvement and which weren't. Measurement enables iteration.

To replicate this programme for your brand, start with the GEO audit to establish your baseline and identify priority actions. Then build on the foundation with structured data, content strategy, and brand authority programmes. Use Sight to measure your progress every step of the way →