The NPD Group saw what the checkout missed.

Consider a pair of shoes. The shop knows the size, the price and the moment the payment cleared. It may know the customer’s name. What it cannot see in that transaction is the pair bought somewhere else last week, or the customer who stopped visiting altogether. A checkout is an excellent witness with a restricted view. The NPD Group built a business by widening that view.

First, ask people to keep a diary

The beginning was remarkably domestic. In 1966, Henry Brenner founded NPD with a system for recording buying habits through purchase diaries. Ordinary shopping, written down consistently, could become something a manufacturer could use. The valuable ingredient was comparison. One household’s purchases were an anecdote. A structured record across households offered a way to study a market.

The methods expanded into retail tracking, consumer panels and receipt measurement. So did the questions. How much sold? Who bought it? Was the buyer returning? NPD’s work spread across categories including footwear, toys, beauty, technology and foodservice. Its product was the ability to put a company’s performance next to a wider picture of demand.

The missing denominator

Imagine a brand whose sales rise while its category grows faster. Its internal report looks cheerful. Its market share tells a less comfortable story. That is an illustrative example, but it explains the demand for outside measurement: the denominator lives beyond the company’s own books. NPD supplied data and interpretation for businesses making assortment, pricing, distribution and product decisions.

Different methods answer different questions. Point-of-sale records describe transactions at participating retailers. Consumer panels follow purchasing behavior across people and trips. Surveys can investigate motives that a barcode cannot explain. Circana now describes calibrating its consumer panels against retail sales data. Combining those perspectives helps researchers check a behavioral story against what actually moved through stores.

A book count with small print

BookScan makes the business tangible. Now offered under Circana, it tracks approximately 85% of U.S. trade print book sales. Booksellers can examine individual titles, construct bestseller lists and assemble custom collections. A publisher weighing a category or a bookseller considering an order gets a view beyond the titles moving through one shop.

The boundary matters as much as the number. Trade print books are a defined market, not every form of reading. An ebook is outside that print count. The American Booksellers Association also explains that BookScan supplies aggregate data rather than retailer breakouts. Precision about the question prevents a useful measurement from becoming an answer to something it never measured.

Then the grocery basket entered

By its 2021 acquisition announcement, NPD said more than 2,000 companies relied on its services. Hellman & Friedman acquired the business that year. On August 1, 2022, NPD and IRI completed their merger. The logic was complementary coverage: NPD brought general merchandise and foodservice expertise; IRI brought a business rooted in consumer packaged goods and retail measurement.

The new name arrived on March 7, 2023. Circana said it then advised almost 7,000 brands and retailers. That figure belongs to the combined company. Calling it NPD’s standalone growth would muddle the story. The combination’s appeal was a broader view of the shopper’s wallet, crossing the departmental boundaries that companies often use to organize themselves.

“We make the complex clear.”Circana’s published “Seek Clarity” behavior

The menu behind the measurement

The services remain more varied than a single dashboard suggests. CREST studies foodservice visits, including where people eat, what they order and the circumstances around the occasion. Complete Beauty combines mass and prestige beauty measurement. Liquid Data Go packages performance, consumer, pricing and distribution analysis for growing consumer packaged goods brands. The buyer chooses a research scope, not simply a chart style.

Commercially, this is business-to-business access to data and analytical tools, with research and advisory services alongside. Liquid Data Go describes a contract and setup process. A sensible evaluation starts with a decision: which stores deserve the next distribution push, for example? Then test whether the relevant categories, channels and detail are available at a cost the decision can justify.

The customer list also reaches beyond household names. Circana’s Liquid Data Go page features accounts from Lily’s Toaster Grills, Stoeffler and Partner Foods Group. Their comments describe retail meetings, sorting information and putting charts into presentations. Those are vendor-published testimonials, but the tasks are revealing. Research has to survive the trip from an analyst’s screen to a buyer’s meeting. The practical output can be a distribution argument or a category comparison that another person can interrogate.

The purchase becomes an advertising question

Circana’s recent moves carry the purchase record into media budgets. In April 2026, it launched Liquid Activation for building purchase-based audiences. In September, it announced a DoorDash partnership to validate incremental advertising sales, and added Google’s Meridian framework to Liquid Mix. The recurring question is whether marketing created demand or received credit for buying that would have happened anyway.

There are alternatives. NIQ also combines retail measurement and consumer panels. Buyers should compare actual coverage and analytical fit, rather than assume that joining datasets is unique to one provider. The useful distinction is which evidence answers the particular business question.

NPD’s transferable lesson is wonderfully practical: begin with a behavior, make it comparable, and know the boundary of the count. Apply it by pairing your own sales trend with a consistently defined market and investigating who returned. It loses force when coverage misses your channel, the sample cannot support your question, or nobody can act on the result. Even a crowded dashboard needs a decision waiting at the other end.