AI readability
Can AI read me?
Check whether important machine-readable and technical signals are accessible.
Crawler access
Rendered HTML
/llms.txt
Structured data
AI search visibility for B2B SaaS
Can AI read,understand and mentionyour SaaS?
Growthract checks the signals AI systems can use to understand your website, tests how your brand appears in a real response, and turns the gaps into evidence-backed next actions.
Discovery surfaces
Where software evaluation happens
ChatGPT
Perplexity
Claude
Gemini
Google AI Overviews
Microsoft Copilot
Reddit
YouTube
Quora
Growthract researches technical visibility across ChatGPT, Perplexity, Claude, Gemini, Google AI Overviews, Microsoft Copilot, Reddit, YouTube, and Quora. Displayed platforms do not imply partnership or endorsement. Search Operator’s live response test currently runs on Gemini only.
Original research
Evidence from 100 B2B SaaS websites.
Growthract inspected public technical signals across a frozen 100-company sample to see how common AI-search readiness practices actually are.
Original Research2026 Benchmark
2026 AI Search Readiness Benchmark: 100 B2B SaaS Websites
Original research across 100 B2B SaaS websites on llms.txt adoption, AI crawler access, JSON-LD, schema markup, canonicals, and technical readiness.
100
B2B SaaS websites studied
76.3%
llms.txt present · 74/97 determinable
99.0%
Tested crawler access · 97/98 known
78.4%
Homepage JSON-LD · 76/97 inspectable
Benchmark signals
n = 100AI search crawler access
97 / 98 known
Homepage JSON-LD
76 / 97 inspectable
llms.txt adoption
74 / 97 determinable
SoftwareApplication schema
21 / 97 inspectable
Deterministic observations from public website evidence. No proprietary readiness score and no claim that these signals cause AI citations.
Three questions
Start with what AIcan actually see.
AI understanding
Can AI understand me?
Check whether the site clearly explains the business, product and relevant entities.
Product category
Entity clarity
Content structure
Answer-ready information
AI visibility
Does AI mention me?
Run a point-in-time Gemini response test and inspect how the brand appears against competitors.
Brand mention
Competitor mentions
Website reference
Recommendation signal
See the product
From websiteto evidence to action.
Watch the workflow instead of reading another explanation of it.
Growthract product demo
The output
Know what was found.Know what to do next.
Growthract separates different kinds of evidence instead of compressing everything into one opaque score.
Technical findings
Crawler access, rendering, structured data and machine-readable files.
Content & entity findings
What the site communicates about the business, category and product.
AI-response findings
Point-in-time evidence from a tested Gemini response.
Prioritized actions
What deserves attention, why it matters and where the change belongs.
Evidence, not mystery
A recommendation shouldexplain itself.
The useful part is not the diagnosis alone. It is being able to trace the recommendation back to something observed.
Illustrative output format
Example of how a finding is structured
Clarify the primary product category.
Problem
The primary product category is not explicit enough in the evidence being reviewed.
Evidence
Relevant page copy and machine-readable signals do not consistently reinforce the same category.
Action
Make the category explicit and reinforce it consistently across the relevant signals.
Where
Homepage positioning, structured data and other authoritative product pages.
This is an illustrative format, not a finding about a specific customer or website.
How Growthract works
Evidence first.Action second.
Access
Check whether important technical and machine-readable signals are available.
Understand
Review how clearly the site describes the business, product and entities.
Test visibility
Inspect a point-in-time Gemini response for mentions, references and competitors.
Act
Prioritize the changes that are supported by the evidence collected.
Evidence, kept separate
Four evidence types.Never blended into one magic score.
A blocked crawler is a factual technical finding. A Gemini response is an observation. A content recommendation is an AI-assisted judgment. Search Operator keeps those distinctions visible instead of pretending they are the same kind of evidence.
Confirmed
Deterministic technical checks
Direct checks of robots.txt, /llms.txt, metadata, structured data, indexability, sitemaps, duplicate metadata, and image alt text.
Factual and repeatable
Observed
Gemini response evidence
A point-in-time check using three fixed Gemini prompts, showing whether the brand or a chosen competitor appears and exposing the response evidence available at scan time.
Observed at scan time
Directional
AI-assisted critique
Model-based assessment of positioning, content clarity, topic coverage, quotability, and which improvements deserve attention first.
Judgment, not a measured fact
Grounded
Connected search and traffic data
When Google Search Console or GA4 is connected, real clicks, impressions, average position, and sessions can ground prioritization alongside the website findings.
Optional real performance data
Why this matters: a technical failure should not be presented as an AI opinion, and an AI opinion should not be presented as a measured fact.
Observed crawler evidence
See which AI botsactually requested your site.
Crawler Logs parses raw access logs in your browser and identifies recognized AI crawler requests. It complements configuration checks by showing what appeared in the log window you provide.
Analyze crawler logs
Public Lab 02 · Live Gemini Evidence
A mention is notthe same as accuracy.
Gemini mentioned Growthract in 2 of 3 prompts. Both mention-containing snippets described a materially different business.
2/3
Gemini prompts with a mention
0/3
Displayed source citations
2/2
Mentions with inaccurate positioning
4/5
AI-readiness checks passed
Point-in-time Gemini test · 22 August 2026. Responses were exposed by the product as exact 280-character snippets, not full answers. AI outputs can change between runs.
Pricing
Start free.Pay when there's something worth fixing.
Start with the diagnostic. If the evidence points to deeper work, choose the level of implementation that matches the problem.
Start here
Search Operator — Free Diagnostic
$0
See whether AI systems can access, understand and mention your SaaS.
Public website diagnostic
AI readability checks
Entity and content signals
Point-in-time Gemini visibility test
Written report
Most requestedQuick Fix Report
$399 one-time
A written report ranking your top 5–8 AI-visibility issues by impact, with exact fix instructions for each. Delivered within 48 hours. No call required.
Top issues ranked by impact
Exact fix instructions
Delivered in 48 hours
No call required
I personally review every report before it goes out.
Not useful? Tell me and I'll refund it.
See the evidence methodology in action
Need deeper implementation across your whole site?
Tell me what you're working with and I'll scope it with you directly.
What we can prove
Separate what we controlfrom what we do not.
Verifiable
Technical evidence
Crawler access, rendered HTML, deployed files and structured data can be inspected and verified directly.
Directional
Third-party AI behavior
AI assistants control their own indexing, retrieval and response behavior. A test is evidence from that moment, not a guarantee of future citations.