How scoring works

Every score starts at 100. Real issues take it down from there.

The Agent Readiness Score is a 0–100 number computed from a real crawl of your site, not a static checklist. It answers one question: can an AI agent actually find, understand, and act on this page?

How it's computed

One pipeline, four steps.

01
Crawl & extract
We crawl your site the way an agent would, and extract the same signals it would see — markup, text, metadata, and page structure.
02
Run compatibility checks
Every page, and the domain as a whole, is checked against the rules in each category below.
03
Issues subtract from 100
Each category starts at 100. Every issue we find subtracts points, sized by how much it actually breaks the agent's experience.
04
Categories blend into one score
The categories that apply to your site type blend into a single weighted overall score. Categories that don’t apply are left out entirely — never counted against you.
The seven categories

What we check, and why it matters to an agent.

Only the categories that apply to your site type count toward your overall score, and the weights are public — every rule, severity, and multiplier is published (see the full rule catalog), so a score is reproducible, like a benchmark. Where a category doesn't apply, it is excluded and the remaining weights renormalize.

Discoverabilityweight 17%
Before an agent can evaluate anything on your site, it has to find and reach your content without hitting an invisible wall.
  • Robots.txt allows AI crawler access
  • llms.txt published
  • Sitemap.xml present and valid
  • Pages render as clean markdown for agents
Higher weight
Structured Dataweight 18%
Machine-readable markup tells an agent what a page actually is, instead of leaving it to guess from rendered HTML.
  • schema.org Product, Offer & availability markup
  • Sensible heading structure
  • Page weight an agent can parse quickly
Higher weight
Consistencyweight 15%
Agents trust structured data over the visible page — when the two disagree, it reads as manipulation and the agent backs away.
  • Displayed price matches the price in markup
  • One canonical entity per page, not duplicates
  • Schema blocks on the same page tell one story
Average weight
Actionabilityweight 12%
Finding and understanding a page is wasted if the agent still can't complete the task.
  • Key actions reachable without a JS-only wall
  • Stable deep links to product, cart & checkout
  • A declared machine interface (e.g. MCP) for known tasks
Average weight
Trustweight 8%
Agents — and the people behind them — check whether a business looks safe to transact with before committing to anything.
  • Return policy easy to find
  • Shipping costs & timelines published pre-checkout
  • Clean citation metadata (OpenGraph)
Lower weight
UCPweight 12%
The Universal Commerce Protocol lets a compliant agent transact with your site directly, without scraping the UI at all.
  • /.well-known/ucp profile published
  • Profile schema is valid
  • Checkout capability declared
Average weight — commerce, booking, restaurant & SaaS sites
Citabilityweight 18%
When an AI answers a question by quoting a page, it picks passages it can lift cleanly and trust.
  • Self-contained passages
  • Answer-first structure
  • Concrete figures over vague claims
Higher weight — content sites
Severity ladder

Not every issue costs the same.

Critical
Blocks agents outright — the site fails at this step no matter what else is right.
High
Major friction that stops most agent journeys before they finish.
Medium
Makes the site meaningfully harder to use correctly, but an agent can often still get through.
Low
A minor rough edge, unlikely on its own to break an outcome.
Info
Worth knowing about — the smallest deduction of the five.

The point values are part of the open methodology: each finding deducts 40 (critical), 25 (high), 14 (medium), 6 (low) or 2 (info) points from its category before weighting. On top of the weighted blend, critical failures cap the overall score regardless of how good everything else looks: a site that actively blocks AI agents, or a commerce-tier site whose purchase simulations fail, cannot score as if agents could use it. The exact cap mechanics are part of the proprietary engine.

Transparent Score. Proprietary Engine.

Every Scovant score is explainable: the categories, their weights, every rule in the public catalog, its severity, and the severity point values above are all published — enough to understand any score, act on any finding, and cite a result like a benchmark. Every scan also records the exact scoring-model version that produced it, so a score change from a model update is never mistaken for a website regression.

What we deliberately keep private: the calibration layer — exact critical-cap interactions, per-rule normalization internals, and the anti-gaming heuristics that keep checklist tricks from inflating a score. Those exist so that a high Scovant score keeps meaning what it claims to mean.

Badges

Two thresholds, no exceptions.

Agent Ready
Score of 80 or higher.
Agent Compatible
Score of 60 or higher.

Scores also map to a grade — A at 90 or higher, B at Agent Ready (80), C at Agent Compatible (60), D below — the same ladder your scan reports show.

For commerce-tier sites, both badges have a hard cap tied to the real agent journey: if too few of the AI agents we run can complete the purchase, no badge is shown — regardless of the numeric score. A badge means the score and the real agent journeys agree.

Simulation-informed scoring

For commerce-tier sites, we don't just read the markup.

For commerce-tier sites (online stores, booking services, restaurants), we run several AI agents that each attempt the purchase journey in a live browser — finding a product, adding it to the cart, and working through checkout — stopping just before payment. They never enter real payment details or create an account. How many of them complete the journey feeds directly into your score: when the agents succeed it supports the Actionability category, and when too few can complete the purchase the overall score drops and the badge is withheld, as described above. One agent stumbling where the rest succeed is surfaced as a note, not a penalty.

FAQ

Common questions.

My score changed and I didn't touch the site — what happened?
A few things can move the number without a code change on your end: the page itself changed upstream (a CMS update, a new CDN or WAF rule that started challenging crawlers), or we refined the scoring model itself. Model updates are dated, so a shift right after one is expected — it isn't a regression in your site.
How do I actually raise my score?
Open the issues list on your latest scan and fix what's flagged, starting with the highest severity, then rescan. There's no shortcut: the score reflects what an agent can actually find and do on your site, so real fixes move it — chasing the number without changing anything won't.
How often should I rescan?
Rescan after any deploy that touches pages we score — content, markup, or the pages themselves. If you've connected CI, a rescan can run automatically right after each deploy, so a regression shows up immediately instead of days later.
Why does a category on my report show N/A?
Some categories only apply to certain site types — UCP applies to commerce, booking, restaurant and SaaS sites; Citability applies to content sites like blogs and documentation. If a category doesn't apply to yours, it's left out of your overall score entirely, rather than counted against you as a zero.

See where your site stands.

Run a scan and get your Agent Readiness Score, category by category, with the exact issues to fix.