CPG market research: buy scan data last, call the category manager first

CPG market research is how a consumer packaged goods brand finds out whether its product will sell off a shelf and what will persuade a retailer to give it one. For a small or emerging brand, CPG market research comes down to four sources: retail scan data, shopper research, concept and package tests, and conversations with the people who decide shelf space. Do them in the order that matches the decision in front of you, and pay for syndicated data (the store sales reports vendors sell to every brand in a category) only when a buyer will ask for it.

The trap is treating CPG research like a big-brand insights department would. A big brand's insights team can test everything. A brand with two founders and a co-packer has to pick the one question a category manager will grill them on, and answer that first.

1. Write down the shelf decision before you buy any data

Start with the specific decision a specific person will make. "Will the Kroger natural pet buyer take two SKUs at the next category review?" is a research question. "How big is the pet treat market?" is a slide.

The reason is volume. USDA's Economic Research Service counted 21,435 new food and beverage products on US retail shelves in 2016, and that is food and drink alone, before beauty and household products. A buyer sees a steady stream of pitches, so the research that matters is whatever helps your pitch beat the others for a fixed amount of shelf.

Short version: name the retailer, the buyer's role and the review date, then work backward to the evidence that buyer will want.

2. Scan data shows how the category sells in measured stores, not why, and not everywhere

Retail scan data is the point-of-sale record that NielsenIQ and Circana collect from participating retailers and sell back to brands. It tells you units, dollars, price and promotion by product, retailer and week. It is the shared language of the shelf: your buyer is looking at it too.

Learn three terms before your first meeting. NielsenIQ's glossary of terms buyers expect brands to know defines them:

What scan data does not tell you matters just as much. It does not say why a shopper picked your bag or skipped it. It covers only the stores in the dataset you bought, so ask the vendor which retailers and channels are in your market before you pay, because a chain that does not share data is invisible in it.

It also misses most of your online sales. The Census Bureau estimated that e-commerce was 17.1% of all US retail sales in the second quarter of 2026, seasonally adjusted. That figure covers all retail, not only packaged goods, but it is a fair reminder that your Amazon listing and your own site sit outside the store data a grocery buyer reads.

For a small brand, the cheap entry point is NielsenIQ's Byzzer. Its product page says a free account includes three report pulls and a weekly alert on your brand, category and chosen competitors, with more data available as you pay; the page lists no prices. Use the free pulls to see your category's top sellers, their average price and how often they are on promotion. Pay for more only when a buyer asks for numbers you cannot pull for free.

3. Argue with velocity, because your dollar sales will always look small

A new brand loses every comparison on total dollars. It can win on velocity, which NielsenIQ's glossary describes as the way to compare how fast products sell relative to their distribution.

Here is a made-up example. Ridgeline makes freeze-dried dog treats. It is in 120 stores of a regional chain and sold 1,440 units in the last 12 weeks. That is 1,440 divided by 120 stores divided by 12 weeks, or one unit per store per week. The category leader in the same stores sells three units per store per week, and the median item in the set sells 0.8.

Ridgeline's founder should open the next review with "we outsell the median item in the set," and show the math. Being a third of the leader is normal for a newcomer. Being below the median is the number to fix before the next review.

Short version: compute units per store per week for yourself and for the items around you on the shelf, and bring that comparison to every buyer meeting.

4. Ask shoppers about the last thing they bought, and watch them in the aisle

Scan data says what sold. Shopper research says why, and it is the part a small brand can do cheaply and well.

Ask about a real recent purchase instead of a hypothetical one. "Walk me through the last time you bought dog treats" gets you the store, the price they paid, the brand they almost picked and the reason they did not. "Would you buy a premium freeze-dried treat?" gets you a polite yes. Our consumer research guide covers how to run these conversations, and in-depth interviews covers when one long call beats a survey.

Then go to the aisle. Stand near the set for an hour on a Saturday and count how many people pick up your package, turn it over and put it back. (The store manager will wonder what you are doing. Tell them, and ask what they see from the register.)

This method does not fit a brand that sells only online. If you are direct-to-consumer, your order data and returns are better shopper research than an aisle you are not in.

5. Test the concept and the package against the shelf, before the print run

Run a concept test before you commit to a flavor lineup and a package test before you commit to film and boxes. Both are cheap compared with a pallet of packaging you cannot use.

Test against competitors, not alone. A package that looks great on a white background can vanish next to six other brown kraft bags. Show shoppers a mock shelf with your item among the real competitors, and ask which they would pick and why. Our concept testing guide covers how to rank ideas on what buyers say and then check what they do, and price testing covers the price question that always comes next.

Be careful with launch-failure statistics. You will often read that 80% of new products fail, usually with no source. The numbers NielsenIQ itself publishes are narrower: in a 2025 piece for emerging brands, it says about half of new products keep growing in their second year, and one in four innovations reduces total brand sales, One way that happens is a new item taking sales from the brand's existing items, which is worth testing directly. Ask whether a shopper who picks your new flavor would otherwise have bought your old one.

6. Talk to the people who decide shelf space early, before you need a yes

The most useful CPG research is a conversation with the people between you and the shopper. They know the answer to questions no dataset holds.

Each one sees a different part of the decision:

• Category managers and buyers decide which items are in the set, and when the category is reviewed. Ask what made the last new item they added worth the slot, and what got the last one cut. Start with grocery category management experts.
• Brokers represent many brands to many buyers. Ask which retailers are open to new brands in your category this year and what velocity they expect after six months.
• Distributors decide whether you get into their warehouse. Ask whether the retailer you want requires them, and what their minimums and fees look like for a brand your size. UNFI merchandising leaders see this from the inside.
• Store operators and department managers see what happens after the planogram is set: out-of-stocks, damaged cases, the item that never gets restocked. Ask what makes an item easy for their staff to keep on shelf.

Have these calls three to six months before a category review, when nobody is deciding anything yet. A former buyer can speak more freely about what buyers want than a current one sitting across a pitch table.

This step matters less for a brand that will stay online. If wholesale is on your roadmap, people who have moved a brand from direct-to-consumer to wholesale can tell you what changes first.

7. Skip the big study if a buyer meeting can answer the question

A large brand with an insights team will run tracking studies, simulated test markets and household panels. A brand with fewer than a dozen retail accounts usually should not. The money is better spent on samples, a demo program and the calls in section 6.

The test is simple: if one 30-minute conversation with a buyer, broker or store operator would settle the question, have that conversation before you commission anything. Instant Expert can help you reach specific category managers, brokers and retail operators and invite each one to a short call or a written or voice answer; you pay only if someone books or answers.

Your next step

Pick the next retailer you want, find out when that category is reviewed, and compute your velocity against the items you would sit next to. Then call two people who have bought that category before, and ask them what number would get you in.