About the AI Scorecard
The way people shop online is changing
Why we built the Summit AI Scorecard, and what it measures.
Adam Preston
SEO Director, Summit
For two decades, performance marketing was straightforward: win the search result, get the click, and you've won the customer. But that traditional path is moving quickly.
AI agents and conversational search are quickly becoming the main way shoppers find products. These systems don't just browse; they ask, compare, and recommend items for the customer. Often, they do this without the shopper ever needing to visit a retailer's website.
If these AI systems can't easily read your site, you aren't even in the game.
We created the Summit AI Scorecard to help brands measure exactly where they stand in this new landscape.
We've done this before
We launched the original Summit Scorecard to rank the top 50 UK retailers across hundreds of criteria. We looked at everything from page speed and checkout flows to product photos and search visibility. We even mystery shopped every brand on the list to see the experience firsthand.
The idea was simple: most retailers knew their revenue, but they didn't know what they were losing to poor experiences or better competitors. Our league table made those gaps specific and visible.
Technology has moved on, but that core need for clarity remains the same.
What the AI Scorecard measures
Nine dimensions, audited across four page types: homepage, category, product and content.
- 01Crawl
- Whether AI crawlers can see your content at all, or if it's hidden behind code they can't run.
- 02Structured data
- The markup that gives AI the facts it needs to compare your products to others.
- 03Content
- Is your copy clear and detailed enough for an AI to summarise and quote?
- 04LLM signals
- Does your site look like a trusted authority, with the right expertise and credentials?
- 05Headings
- A logical structure that helps systems understand what each page is actually about.
- 06Metadata
- Titles and descriptions that help define your page's identity.
- 07Internal links
- How your content connects to the pages where people actually buy things.
- 08Alt text
- Descriptions that help AI "see" and understand your images.
- 09ARIA
- The background labels that tell agents what different buttons and menus actually do.
These factors combine into an AI Readiness score. We prioritise areas like structured data and content because they have the biggest impact on whether an AI can describe your product accurately.
Crawl access is the gatekeeper. If a retailer doesn't let a crawler in, none of the other work matters, because there's nothing for the AI to see.
But we don’t just measure what’s on your site
Next to readiness, we track how visible your site is within AI platforms as well as how your brand sentiment is viewed within AI platforms. This means we’re able to give a full rounded assessment of how your site is set up for AI, but also how it is currently performing.
Readiness
What you can control
A property of your site. Measurable, diagnosable, and fixable — schema, content depth, rendering, structure.
EarnedVisibility
Where you currently appear
Today this depends far more on brand fame than technical quality. Well-built sites go unmentioned; famous brands get named regardless.
InheritedWhy this matters now
Even the most beautiful website is invisible to an AI if the data behind it is thin or the content is shallow. We've audited household names that offer almost nothing readable to a crawler. These brands are being recommended purely on their reputation, for now.
In this new world, visibility is earned technically before it's won creatively. The foundations you build now will be your biggest advantage as AI shopping becomes the norm.
The good news is that readiness is something you can measure and improve. Once you know where you stand, things like data structure and content depth are levers you can pull.
What this is, and what it isn't
Readiness scores come from deep technical audits, but reflect a sample of key pages rather than every page on a site.
Visibility data is a snapshot, not a long-term trend: one point in time, on one platform.
Fifty prompts per retailer is enough to see a clear rank, but it's the beginning of the story rather than the end.
We'll run these tests regularly. Seeing how the numbers change over time is where the real value lies.
