Skip to content
anserra
Run a free scanSign in

How it works

How does Anserra work?

Anserra reads your site the way AI crawlers do, asks the AI engines the questions your customers ask, sends a browser agent to try booking with you, and turns all three into one ranked fix plan. Here is each step in detail.

Updated:

  1. 1

    Scan

    You enter a domain. Anserra fetches the site the way AI crawlers do — with their user-agents, without running JavaScript first — and records what a language model would actually see: text, headings, structured data, robots.txt, llms.txt. Then it builds a prompt set for your niche and city, or uses the one you edited.

  2. 2

    See what AI says

    Each prompt goes to ChatGPT, Perplexity and Google AI Overviews through their APIs, several times. For each answer Anserra records whether you are named, in what position, what is said about you and whether it is accurate, who else is named, and which pages are cited. Numbers come with confidence intervals.

  3. 3

    See if an agent can book

    AnserraBot, a browser agent, opens your site and attempts five customer tasks: understand the offer, find the price, find the booking path, check the conditions, reach a human. It follows existing links only and never submits anything. You get the exact step where it stopped, with a screenshot, and a level from L0 to L3.

  4. 4

    Get the fix pack

    Every finding becomes a task, ranked by how much it unblocks. The technical ones come with the file written from your data: llms.txt, robots.txt, JSON-LD, and agent-service.json for L3. Next week's run marks what changed.

What is checked: the seven analyzers

Each analyzer contributes to the score and to the fix plan. A1–A6 read the site; A7 is the browser agent.

AnalyzerQuestion it answersWhere it shows up
A1 DiscoverabilityCan AI crawlers reach the site? robots.txt, sitemap, firewall rules for GPTBot, ClaudeBot, PerplexityBot and othersFix plan: robots.txt generator
A2 Machine readabilityIs the content in the HTML without JavaScript? Heading structure, clean textFix plan: rendering tasks
A3 Structured dataIs there correct schema.org markup for the business type, services, offers, FAQ?Fix plan: JSON-LD generator
A4 TransactabilityIs there a readable path to booking or purchase, prices as text, no CAPTCHA wall?Agent level L0–L3
A5 FreshnessAre pages dated, are hours and prices current?Fix plan: content tasks
A6 TrustReviews, licenses, consistent name–address–phone across the webFix plan: listing tasks
A7 Agent simulationCan a real browser agent complete five customer tasks?Agent Readiness report with screenshots

How the score is built

The score from 0 to 100 combines four components: mention rate across prompts and engines, accuracy of what the engines say, presence on the sources they cite, and agent readiness. Each component is shown separately, and the total always carries its confidence interval. The methodology page shows the weights and the sample math.

Engines

  • ChatGPT
  • Perplexity
  • Google AI Overviews
  • Claude · add-on
  • Gemini · add-on

Answers are collected through the providers' official APIs. Claude and Gemini are add-ons per business.

What you receive

Report

Score with interval, per-prompt and per-engine table, sources, agent run with screenshots. In the app, as a PDF, or as a shareable link.

Fix plan

Ranked tasks with impact, each linked to the finding that produced it. Reads as a work order for your developer or agency.

Generated files

llms.txt, robots.txt, JSON-LD and agent-service.json filled with your data, ready to paste. Plus an agent-ready badge at L2.

See a full sample report →

See what AI says about your business today

One business, one-time check. No account, no card. Results in about 60 seconds.

Run a free scanScan a client list