Core tells you what a product is. Contextual tells you what people say about it, where they said it, and how many of them said it. The brand's claim and the shopper's reason are both kept, and never confused for each other.
Nothing here is inferred. Every value is a phrase someone published, and the record keeps the source and the number of independent readings behind it.
Verified reviews from every retailer that carries the item. We read the sentence rather than the star rating, so the record keeps the benefit the shopper actually named and how many separate people named it.
The queries and autocomplete language people use to find the item and the category around it. This is the vocabulary shoppers reach for before they know your brand, and it is rarely the vocabulary on your pack.
Claims, certifications and ingredient language the brand publishes itself. Kept in its own bucket, so an assertion the brand made is never quietly counted as an outcome a shopper reported.
Most teams infer demand language from a survey, a panel or a hunch. This is the same language taken from what shoppers published, at the UPC, across the whole category.
Build the shelf around the reasons shoppers gave, at the UPC, instead of the segments the syndicated hierarchy happens to ship with. The benefit that is pulling volume is a field you can sort on.
Answer a demand question in an afternoon. The language is already read, already counted and already joined to the UPC, so a rising trend is something you measure rather than something you commission.
Find the benefit shoppers keep naming that your pack has never said out loud. It is usually the claim that belongs on the next design, and the line that belongs in the next campaign.
At a leading snacks manufacturer, reading demand themes across more than 100,000 PDP records on over 5,000 products surfaced more than $300M in quantified revenue opportunity, including one theme worth $130M that the portfolio had barely entered.
Contextual is the second layer. Core holds what the product is, Custom holds the structure you define, and Contextual holds what the market says about it. Every attribute is coded into exactly one of the three, so the layers add up to the whole record and nothing is counted twice.
The same record, delivered three ways. Most teams start with one and add another once the record is in their stack.
Contextual attributes are the demand language your category already speaks. Read it at the UPC and the plan, the pack and the campaign all start from what shoppers actually said.
Say what shoppers are already saying, in the words they used, backed by how many of them said it.
Thirty minutes on your categories. How the record gets built, what it holds, and the questions your team could put to it.