Shiftrank Answer engine optimization at the edge
Point your DNS at Shiftrank to see every AI crawler that reads your site, serve the machine version of your content, and decide who gets to train on it.
Observe mode records your first crawler visit within minutes of the CNAME resolving. No SDK, no origin deploy, and no change to how humans experience the site.
Observe
See which answer engines read your site
Every request through the proxy is classified: human or machine, which operator, which path, how many bytes. No sampling, no tag manager, no code change.
- Operators: OpenAI, Anthropic, Perplexity, Google, Meta, Common Crawl
- Split training crawlers from live answer fetches
- Path, status, bytes, and demand-signal breakdowns
2026-08-07T04:12:09Z GPTBot /pricing 200 14.2kB
2026-08-07T04:12:44Z ClaudeBot /docs/llms.txt 200 3.1kB
2026-08-07T04:13:02Z PerplexityBot /blog/aeo 200 22.8kB
2026-08-07T04:13:31Z ChatGPT-User /pricing 200 14.2kBEnhance
Serve the version a model can actually parse
Shiftrank rewrites responses on the way out. Structured data, an authoritative llms.txt, and markdown for clients that ask for it — from the edge, without a rebuild.
- Generated and hand-edited /llms.txt at the apex
- JSON-LD and meta injection per route pattern
- Markdown content negotiation on Accept: text/markdown
# Acme
> Acme is a billing platform for usage-based SaaS.
## Docs
- [Quickstart](https://acme.com/docs/quickstart): Ship your first meter
- [Pricing](https://acme.com/pricing): Plans and limitsControl
Decide who trains and who cites
Allow the crawlers that send you citations. Block the ones that only take. Rules evaluate at the edge in front of your origin, and a kill switch returns you to plain passthrough instantly.
- Allow, block, or redirect by operator or category
- Path-scoped rules with an explicit precedence order
- Observe mode records the decision before you enforce it
allow search ChatGPT-User, PerplexityBot /*
allow training GPTBot /blog/*
block training GPTBot /app/*
block scraper * /pricingFour surfaces, one proxy hop
Everything runs in the same request path, so measurement and enforcement always agree about what happened.
AI traffic analytics
Time series, operator breakdowns, and path-level detail for machine traffic your existing analytics never recorded.
Edge llms.txt
Author once in the dashboard and serve at the apex, versioned, with no origin deploy.
Crawler rules
Category-aware allow and block rules with observe-first rollout and an instant passthrough kill switch.
Demand signals
The paths answer engines keep re-fetching, ranked — the closest thing to keyword data for AI surfaces.
Live in an afternoon
-
01
Point DNS
Add a TXT record to verify and a CNAME to proxy.shiftrank.ai. Nothing in your application changes.
-
02
Watch in observe mode
Traffic passes straight through while Shiftrank records every crawler visit. First data lands in minutes.
-
03
Turn on enhancement
Enable llms.txt, schema injection, and crawler rules once you can see what they will affect.
From the blog
All posts-
Edge · August 1, 2026
Doing AEO at the edge, without rewriting your application
Why answer engine work keeps stalling in the backlog, and how a reverse proxy in the request path delivers structured data, llms.txt, and crawler control without an application change.
-
Measurement · July 25, 2026
Tracking AI referral traffic in GA4 and beyond
The referrers to segment on, why the numbers look small, why they undercount, and how to connect a referral back to the page the model actually cited.
-
AEO · July 18, 2026
An AEO checklist for SaaS companies
Twenty-two checks across access, content, structure, and measurement — ordered so the diagnostic work happens before the writing, and grouped by who on your team actually owns each one.