Introducing the MPP Website Grader: including the score nobody else gives you

Today we are opening up the MPP Website Grader: the audit tool we have been running against client sites for months. It is free, it takes about a minute, and it is available to anyone.

You can run it right now. Below is what it measures and why we built a fourth score that most tools do not have.

Four scores, not one

A single "your site scores 73" number is a marketing device, not a diagnosis. We report four, because they fail independently and have different owners.

Speed. How quickly the page becomes useful: payload weight, render-blocking resources, image discipline, and the response characteriztics that decide whether a phone user on a weak connection stays. This is a conversion metric before it is a ranking one.

Code health. The structural quality underneath: valid markup, sane heading hierarchy, accessible images, sensible resource loading, and the accumulated debt of a site that has been edited by six people across four years.

SEO. The traditional layer: titles, meta descriptions, canonicals, headings, internal linking, indexability, structured data, and whether the important pages are reachable.

AEO. Answer Engine Optimization. This is the new one.

Why AEO is a separate score

A growing share of the questions your customers ask are now answered by an AI assistant rather than a list of links. Those systems read your website. They arrive without a session, frequently without executing JavaScript, and they decide in a few seconds whether your page can be parsed, trusted, and quoted.

A site can be excellent on the first three scores and invisible on the fourth. We see it constantly: a fast, well-optimized marketing site whose actual content is assembled client-side, so a model fetching the page sees an empty shell. Perfect SEO. Nothing to cite.

The AEO score checks the things that decide whether you exist to an answer engine:

• Content in the markup, not injected after load.
• A clear answer near the top of the page rather than buried under an anecdote.
• Structured data that states what the page and organization actually are.
• robots.txt posture toward AI crawlers: GPTBot, ClaudeBot, PerplexityBot and others. Blocking them is a legitimate choice; making it by accident is not.
• llms.txt: the emerging convention for telling models what your site is and where the good material lives.
• Factual, quotable structure: headings that match real questions, specifics rather than adjectives.

Traditional SEO asks whether you rank. AEO asks whether you get quoted. In an answer-first internet, the second question is starting to matter more.

What else it detects

CMS and stack detection. WordPress, Shopify, Wix, Webflow, Next.js, the hosting provider, and the analytics in use. Useful for us in a sales conversation, and genuinely useful to an owner who has inherited a site nobody documented.

The AI crawler audit. A plain statement of which AI bots your robots.txt currently blocks. Most businesses have never made this decision consciously.

The fix plan

Inside the portal, the team gets one more thing: Claude reads the full result and writes a prioritized action plan in client-friendly language. It says what to fix first, why it matters commercially, and roughly what it involves.

That last part is the difference between an audit and a deliverable. Every agency has handed a client a 40-page technical audit and watched nothing happen, because a list of problems is not a plan. Ordering the problems by business impact is the work.

Why we made it public

Two honest reasons.

It is a genuinely good first conversation. We would rather open a relationship with a specific, measurable statement about your website than a capabilities deck. If the grade is good, you have learned that and cost us a minute. If it is bad, you have a concrete list.

It demonstrates the thesis. We argue that most businesses are guessing about their web presence. A tool that replaces the guess with four numbers and a fix list makes the argument better than a page of copy about our philosophy.

What we learned running it across client sites

Images are still the single biggest speed problem. Not frameworks, not hosting. Unoptimized hero images uploaded at camera resolution.

Nearly every site fails something in the heading hierarchy, usually three h1s or a jump from h1 to h4 because a heading was picked for its size.

AEO scores are systematically worse than SEO scores. The gap is largest on the most modern-looking sites, which is exactly the inversion you would expect: the more the site depends on client-side rendering, the less an answer engine sees.

Almost nobody has made a deliberate decision about AI crawlers. The robots.txt on most sites was written years ago, or generated by a plugin, and has never been revisited.

Try it

Grade your website. No account, no sales call attached, results on screen.

If the results turn up something you would rather have someone own, apply for partnership. If they do not, you have a benchmark for the next time someone tells you your site needs work.