DarkTraffiK's early access cohort has now been running for several months across a range of site types — B2B SaaS, content publishers, agencies, and professional services. Here's what the three-signal funnel data looks like in practice.
All data is drawn from the DarkTraffiK beta cohort. Figures are ranges, not averages, to reflect the variance across site types.
Signal 1: AI Crawler Frequency
- Median crawl events per site per month: 200–600 across all known AI crawlers
- Most active crawlers (by hit count): GPTBot, ClaudeBot, PerplexityBot
- Pages crawled as % of total site: typically 15–40% — AI crawlers are selective
- Most crawled page types: pillar content, FAQ pages, comparison pages, product pages
- Most common robots.txt error found: blanket
Disallow: /rules inadvertently blocking AI bots (found on 31% of sites in the beta cohort)
The ClaudeBot blocking finding
One early access tester had blocked ClaudeBot six months prior in a blanket disallow rule — and had no idea. Fixing it took five minutes. Their Anthropic citation volume doubled within three weeks. This was the most dramatic single-fix result we've seen.
Signal 2: Citation Rates
- Median citation rate (topics cited / topics crawled): 20–30%
- Median citation rate (topics cited / topics checked): 30–50%
- Most cited content types: direct-answer content, FAQ content, original data/research, comparison content
- Least cited content types: general overview content, content with high hedge-language density, content without structured data
Signal 3: GA4 Referral Data
- AI referrals as % of known traffic: typically 3–8% of all referral sessions in GA4, when filtered for AI domains
- Referrer stripping rate (estimated from Signal 1 vs Signal 3 comparison): 60–80% of AI-influenced visits arrive without a referrer
- Meaning: the GA4 number is typically 20–40% of the real Signal 3 figure
The Gap List: What We're Finding
- Median gap list length (crawled-but-not-cited topics): 6–14 topics per site
- Most common gap category: FAQ and definition content — AI engines frequently read FAQ pages but don't cite them if answers are vague or hedged
- Second most common gap: comparison content that doesn't include concrete data or specific named comparisons
- Fastest gap to close: direct-answer FAQ content with JSON-LD schema — median citation appearance: 3–6 weeks after publish
What Moves the Needle Fastest
Based on the beta cohort, the single highest-impact action for sites with an active gap list:
- Identify the highest-crawl-volume uncited page
- Rewrite the opening paragraph to be answer-first (direct answer in sentence 1)
- Add FAQ schema in JSON-LD
- Add one quotable, specific, verifiable claim with a source
- Publish and file an IndexNow ping
Median time to first citation appearance after this treatment: 3–5 weeks.