AI Discovery – Through the Lens of a Supermarket Shopper
Imagine having an allergy and walking into a supermarket in a foreign country, one that you’ve never been to before. You’re hungry and looking for something specific to eat, maybe a protein-rich snack bar under 200 calories. You can recognise the packaging symbol used for products that are safe for you to consume.
With those requirements, you wouldn’t simply make your purchasing decision based on the front of the packaging alone. You’d flip it over and read the nutrition label on the back. That’s where the real information lives: ingredients, calories, allergy information.
Your website should work the same way, but doesn’t currently.
The front-end, what humans see when they land on your page, is like that attractive front product packaging. It’s designed to catch the human eye with beautiful images, compelling slogans, and strategic colours that represent the brand to which they belong.
But like intentional shoppers, AI doesn’t shop with its eyes. AI reads the nutrition label.
That “nutrition label” is your structured data: the behind-the-scenes code that tells search engines, and AI agents, exactly what your content contains.
Importantly, the nutrition label not only tells AI what your product is, it also tells AI where it belongs in relation to other products, in what category, near what related items. It’s both description and placement – delivering clarity and context.
These contextually relevant groupings, this context, drives discovery. It’s what we use to navigate through supermarkets. We don’t know exactly what frozen dessert we want tonight, but we know it will be frozen, so we head to the freezers. We don’t necessarily know exactly what we need to host a party this weekend, so we go to the party supplies section and let the shelves give us ideas.
Sometimes, we leave with items that weren’t ones we came to the store looking for. Instead, they were discovered and selected because they were in the right place at the right time to tempt you to buy.
As an example, imagine you came for a coffee table and left with coasters, candles, an area rug, and couch cushions. You had no intention of buying those other items, but they were contextually related to coffee tables and positioned nearby. Context is what led you to them, positioned them to be discovered by people who may not know they’re looking for them too.
Without structured data, your content has no label; AI doesn’t know exactly what it is or what shelves to put it on. Vague, ambiguous, and potentially misplaced, your content becomes increasingly likely to be missed or ignored by AI shoppers – it becomes digitally obscure.
When you get an AI Discoverability Assessment report, you’re essentially getting a quality control summary of your website and its “nutrition labels,” which suggests how accurately and comprehensively they describe your content, and how effectively your content is positioned for discovery.
Your VISEON.IO AI Discovery Assessment is the output from a data pipeline that collects all the structured data on your site and analyses it using an enterprise business intelligence solution that exposes your digital catalogue (knowledge graph, map), of your entire digital “supermarket”. This gives you insight into your store layout, your digital inventory by category, its completeness, and from a quality perspective how much of it is labelled properly for AI shopper satisfaction.
Next we’ll walk through each section of the report using our supermarket analogy, so you can understand exactly what’s being measured and why it matters.
Book a free consultation to gain access to your digital catalogue, discuss your AI Discoverability Assessment, and map out your data-focused strategy for online visibility in 2026.
AI Discoverability Assessment Detail
The Schema Section:
Are You Using the Right Label Format?
Do Your Products Have Labels in the First Place?
What it checks: Whether your website follows Schema.org standards, the universal language that search engines and AI understand, and which of your products have labels assigned to them.
The analogy: Imagine if every food manufacturer created their own label format. One lists calories in the top corner, another hides them on the side panel, and a third invents a new unit of measurement entirely. Chaos, right? Shoppers (and AI) wouldn’t know where to look or how to compare products. Schema.org is like the regulator’s standardised nutrition label format. It ensures everyone speaks the same language, in every country.
Data Types checks whether you’re using the right kind of information in the right places. If the label says “Calories: Yes” instead of “Calories: 150,” that’s a data type problem. AI expects numbers where numbers belong, dates where dates belong, and so on. Get this wrong, and AI might just ignore your label entirely.
Properties verifies that you’re not putting impossible information on your label. You can’t list “number of bedrooms” on a candy bar, just like you can’t assign the property “numberOfRooms” to a Person in your structured data. These mismatches signal errors that make search engines doubt the reliability of your entire label.
Context examines whether you’re signalling to AI that your structured data is aligned with Schema.org standards. If you’re not disclosing, in your structured data, that you are aligning with Schema.org, that would be like a nutrition label not sating that the food has been tested and approved by the FDA. Without this declaration, AI might not recognise your labels as legitimate.
The IDs Section:
Can AI Tell What You Are and Where to Put You?
What it checks: How you identify and reference entities across your website.
The analogy: Imagine shopping at a supermarket where the same cereal box appears in three different aisles. Each box’s nutrition label mentions a different number of calories per serving. Which one is accurate? Or, imagine a package of beef with a nutrition label that mentions a specific wine that pairs nicely with it. You go looking for the wine, only to be told that wine doesn’t exist.
Declarations define what each entity on your page is: a product, a person, an organisation, an event. Without proper declarations, AI is staring at an unlabelled mystery box. It might be cereal, it might be laundry detergent – who knows?
Declarations also check to see if entities have been declared in more than one place, which would be like giving the same products different nutrition labels. This makes it difficult to ensure that AI selects the ‘correct’, most complete, most up to date version of your product, since there are multiple, slightly different, versions of the same product scattered around the store.
References are how you connect related items. When a recipe card says “add 2 cups of organic whole milk,” that’s a reference. If done correctly, it points to the exact milk product elsewhere in the store (or on your website). But if that reference is missing, meaning the milk product doesn’t actually exist anywhere, shoppers get frustrated. If multiple products claim to be “organic whole milk,” which one does the recipe mean?
References are also how you link your products to broader entities, like the manufacturer, country of origin, allergens, key attributes, and certifications, they create the connections, the context, that determine placement. When your nutrition label says “Manufactured by Dannon,” that connects your yogurt to every other Dannon product in the store. “Product of France” groups you with other French foods. “USDA Organic” puts you in the organic section. These references don’t just describe your product; they determine what aisle you belong in and what you sit next to on the shelf.
Here’s where it gets powerful: these references can create multiple valid “shelf positions” for your product. High protein, French, organic yogurt by Dannon could legitimately appear in the organic foods section, the French imports section, the Dannon brand section, AND the high-protein foods section, all at once. Each reference creates a pathway for discovery.
The same logic applies to AI search. When you properly reference that your product is manufactured by a specific company, or originates from a particular country, or is part of a certain brand family, you’re creating multiple pathways for AI to find you. A user doesn’t need to search for your product by name. They might search for “sustainable coffee from Colombia” or “skincare products by Japanese manufacturers,” and if your references are properly connected, AI can surface your content based on those relationships.
Good ID practices help AI understand not just what individual entities within your domain are, but how they relate to each other and to the broader world. This contextual understanding is what allows AI to find and recommend your products accurately, even when users don’t know they are looking for you.
The JSON Section:
Is Your Label Printed Correctly?
What it checks: The technical quality of your structured data code.
The analogy: You’ve designed the perfect nutrition label with all the right information, but what if the printer malfunctions? The ink smudges, lines overlap, and half the text is illegible.
Graph checks if you have multiple versions of the same label printed on different parts of the package. Best practice is one nutrition label per product package. If there are three different labels, even if they all say roughly the same thing, which one should a shopper trust? Multiple knowledge graphs on one page create the same confusion for search engines.
Structure identifies printing errors: missing commas, misplaced brackets, syntax mistakes. In the world of code, even a single missing comma is like a nutrition label being printed without ink. These tiny errors can cause your entire structured data to fail, rendering it completely invisible to search engines. It’s not that the information is wrong; it’s that the format is so broken that AI can’t parse it.
The Google Rich Results Section:
Are You Earning Premium Shelf Space?
What it checks: How your structured data translates into actual visibility on Google.
Note: this section will appear empty, with zeros across the board. This is expected, as we require certain permissions from you before we can collect this data from your domain.
The analogy: In physical supermarkets, premium shelf space – eye level, end caps, featured displays – goes to whoever pays the most. Big brands pay to display for prime promotional real estate. The best product doesn’t always get the best spot; the biggest wallet does.
The digital supermarket offers us more value. You can earn premium placement by proving you have the best label: complete, accurate, properly formatted structured data that demonstrates you’re a reliable, trustworthy high-quality source with topical authority.
It’s one of the few meritocracies left in marketing. Small brands with excellent structured data can outshine corporate giants with messy markup.
Rich Results are the premium listings: the ones with star ratings, images, price tags, and other eye-catching details displayed directly in search results. These are like those end-cap displays or featured product stands. They get dramatically more attention (and clicks) than standard listings buried in the middle of the aisle, out of view.
Good structured data is your application for premium placement. The “Rich Results Pass” KPI tells you how many premium spots you’ve successfully earned, while “Rich Results Fail” shows where you could have been awarded premium placement but weren’t, usually because of missing information or formatting errors.
Indexing measures whether Google can even stock your products on their shelves in the first place. We consult Google and ask it if it knows about your pages. The ‘Index Pass’ KPI is a count of all pages Google can see, and the ‘Index Not Pass’ KPI is a count your pages that Google doesn’t know about, but should. Poor indexing means your pages might not show up in relevant searches at all – not because you don’t have good labels, but because your products literally aren’t in the store.
The Metadata Section:
Does Your Package Include All Required Information?
What it checks: The descriptive elements that help your content appear correctly when shared or discovered.
The analogy: Beyond the nutrition label, there’s other important information on food packaging: the product name, brand logo, tagline, and legally required warnings or certifications.
Metadata Tags are like the title printed on the front of the package and the short description on the side panel. When someone shares your website on social media, these OG (Open Graph) and Twitter tags determine what text and images appear. It’s the difference between a link showing up as “Untitled” with a broken image versus a compelling preview that makes people want to click.
Deprecated Sources are like using outdated food safety certifications that regulatory bodies no longer recognise. Maybe you’re still displaying a “USDA Approved” stamp from 1987 using the old format. It might have been valid once, but now it signals you’re out of touch with current standards, and that the rest of the information on your product may no longer be accurate.
Objects & Alt Text ensure that images, videos, and PDFs have descriptive text labels. Imagine a product with no name, just a picture of what’s inside. Sighted shoppers might recognise it, but anyone with visual impairment wouldn’t know what it is. Alt text serves the same purpose for AI, it describes non-text content so AI can understand and categorise it properly, improving both accessibility and search visibility.
The Reachability Section: Are Your Store’s Aisles Navigable?
What it checks: Whether search engines can navigate through your website effectively.
The analogy: Imagine a supermarket where half the aisles are blocked by fallen displays, some signs point to non-existent sections, and the doors to the stockroom are jammed shut.
Internal Reachability measures the health of links within your own website. These are like the pathways between aisles, directional signs, and “related products” suggestions. When internal links are broken, when you click “Learn More” and get a 404 error, it’s like following a sign that leads to a wall. Search engines rely on internal links to crawl your entire site and understand how pages relate to each other. A site with poor internal linking looks abandoned, poorly maintained, frustrates users, and keeps your content locked away, inaccessible to humans and AI crawlers.
External Reachability examines links pointing to other websites. These are like those small labels that say “Distributed by…” or “Recipe ideas at…” When you link to authoritative external sources, it signals credibility and helpfulness. But if those external links are broken, if you’re referencing a website that no longer exists, it’s a red flag. It suggests your site hasn’t been updated in years, damaging your perceived trustworthiness in the eyes of users and AI.
The Report: In Summary
When you receive your structured data analysis, you’re getting a overarching assessment of the quality of your labels. The report tells you:
- If your labels exist and are clearly understandable (SCHEMA)
- If your products are accurately described to AI (IDs/Declarations)
- If AI can find your products through association, without needing to know your product or brand by name (IDs/References)
- If technical errors are making your labels illegible (JSON)
- If you’re missing opportunities for premium placement (GOOGLE)
- If your products are represented properly across platforms (METADATA)
- If your site’s navigation is breaking down (REACHABILITY)
The good news? Unlike redesigning your entire front-end (redesigning the front of the package), improving structured data is often straightforward technical work. You’re not changing what your site says or how it looks to humans. You’re just making sure the nutrition label on the back is complete, accurate, and printed clearly.
What To Do Next
If you’ve received your structured data report and feel overwhelmed, start with the basics:
- Fix the critical errors first. Broken JSON syntax and major reachability issues. These are like having no label at all, or having aisles completely blocked off.
- Improve rich results eligibility. These are quick wins that can immediately increase your visibility and click-through rates.
- Address foundation issues. Incorrect properties and data types that make your labels unreliable.
- Refine your IDs and references. This builds the contextual understanding that helps AI see the big picture of what you offer, and paves new paths that can lead potential customers to you.
- Polish metadata and long-term reachability. The finishing touches that maximise your presence across all platforms.
How VISEON Can Help
For more detail regarding where your biggest pain points and opportunities are within your domain, schedule a free consultation with us by emailing [email protected], where we can guide you through your report, show you exactly which pages and links are the biggest culprits, and provide tailored, prioritised recommendations to get you started on your journey toward digital prominence.
During this consultation, we will also load your data into the VISEON app, which shines a light on where your errors are occurring and shows a full catalogue of your entities so you can identify any entities that are without labels, have full visibility of all information that’s been listed on each label, and ensure your labels are up to date, compliant with standards, and accurately implemented.
The VISEON app also visualises your structured data, so you can see how your entities are linked to each other and the broader web, helping you create maximum context, paving roads for potential customers to find you.
Why This All Matters:
The AI Shopping Experience
AI doesn’t browse websites the way humans do. A human might get attracted to your beautifully designed packaging and punchy slogan. AI doesn’t see any of that. It goes straight to the structured data, to the nutrition label on the back.
These labels used to be for regulatory compliance, seen as technical housekeeping – an afterthought. Not anymore.
Now, the purpose of these labels extends beyond compliance and into the world of marketing. Their quality can be the differentiator. It can determine whether your products show up in AI generated answers, appear in rich results, get featured in voice assistant responses, get recommended to the right people at the right time – or – collect dust on the bottom shelf.
We believe that businesses should be in control of how they are perceived online. We believe that being discoverable should not be a luxury only afforded to the highest bidder.
That’s why we’ve created VISEON, and why we are sending out free AI Discoverability Assessments to hundreds of organisations, helping business take control of their online presence by showing them that the attention of AI agents is not something that must be bought; it’s something that a holistic, data-focused approach can earn you.