Most brands have spent a decade optimizing PDPs for search. Keywords nailed, titles tightened, feeds structured. The pages convert.
Then AI search arrived, and the assignment quietly changed.
When someone asks ChatGPT for a product recommendation, the model isn't matching keywords. It's looking for pages that explain what the product is, who it's for, and when someone would choose it. Pages that read like a knowledgeable friend, not a product catalog. Most PDPs, even from the world's most valuable brands, don't do this.
That gap is a highly bankable, sales driving opportunity.
What we're seeing in the data
At Brandlight, we analyze citation patterns across AI engines. Two findings keep showing up.
First, product information is now one of the most influential signals in AI responses. Google Shopping is consistently among the most-cited domains in the unbranded queries we track across categories. Amazon shows up just behind. AI engines are pulling structured product data directly into answers, and the brands that surface are the ones whose product information is rich, specific, and machine-readable.
Additionally, third-party validation matters more than it did in classic SEO. Editorial roundups, social media, marketplaces, and review platforms are cited tens of thousands of times. The PDP is no longer a destination. It's a node in a network of signals that AI weights to form a recommendation.
The thing is, that LLM’s don’t just want details and data about your product, in order to answer real human questions, it needs to understand real human information about your product.
Let’s say someone asks an LLM “Which sunscreen is best for my surfing holiday?”. The LLM needs to source more than product information like SPF, ml and clinical trials. It wants to understand whether this product is the right fit. Have reviewers mentioned it’s easily packable to fit in a carry on? Have active people used it for hikes or biking? It’s not just about the product it’s about the people who use it and when it’s best used.
The six things AI is actually looking for
Across the high-cited PDPs we've analyzed, six patterns repeat:
- Semantic descriptions. Lead with what the product is, who it's for, and when to use it. Not poetic copy. Specifics. The model is matching meaning, not phrases.
- Ratings and reviews with thematic depth. AI doesn't read star ratings. It reads review text and looks for repeated patterns: use cases, occasions, sensory details. Volume helps. Recency helps more. Tagged themes help most.
- Contextual use cases. AI queries are situational. Your PDP needs to explicitly connect the product to scenarios. "Best for X" or “Can be used for” gives the model something to match against.
- Awards and certifications. Third-party validation reads as trust. If you have them, they should be on the page. If you don't, pursue them.
- FAQ content. AI queries are conversational. Pages with explicit Q&A become direct sources. Add the questions buyers actually ask, answer them, mark them up with FAQPage schema.
- Structured data. Product schema (price, SKU, availability, reviews) is how AI verifies facts before recommending. OpenAI has confirmed they use it. Google AI Mode pulls directly from Shopping cards built on it. Without proper markup, you're under-indexed to a layer of the stack.
The friction nobody's talking about
Here's what makes this hard for most consumer brands: you don't own the PDPs that matter most.
The retailer owns the page. The template, the schema, the FAQ module, the review system. Your brand team can submit content, but the retailer decides what gets published, in what format, and with what structured data.
This is where most AI visibility advice falls apart. "Add FAQPage schema" is straightforward when you control the page. It's a six-month conversation with a retailer's content team when you don't.
What to lobby your retailers for
The brands that will win the next two years are the ones treating retailer content teams as a strategic relationship, not an afterthought. A short list of what's worth pushing for:
- Structured FAQ modules with schema markup, not just customer Q&A threads
- Tagged review attributes surfaced as filterable themes
- Awards and certifications as structured fields, not body copy retailers strip out
- Richer product attribute fields for use cases and audience fit
- Brand-supplied contextual copy preserved in the published version
- Schema enrichment beyond price and availability, to include reviews, FAQs, and product specifications
This is white space. Most retailers haven't been asked. The brands that ask first will shape the templates everyone else inherits.
The Monday version
Pick your top three SKUs. Audit how they appear on your top three retailer PDPs. Compare to what AI is actually citing in your category. The gap between the two is your roadmap.
