The age band was never the answer

The situation

A snack brand had a premium baked range ready for a regional pilot across grocery and convenience channels in six metro markets. The product team had a target: adults aged 25 to 34. It came from the last brand tracker and everyone in the room accepted it.

Insights did not. Two people in that age band can buy the same category for opposite reasons. One reads the nutrition label on the back of the bag. The other grabs whatever is on the endcap promotion. The brand had a 1,800-respondent survey from the previous quarter, full of attitudes and usage behavior. Nobody had run a segmentation on it, because that meant waiting for an analyst, and the analyst queue was three weeks out.

The question was whether the team could get a defensible segmentation from data it already owned, without a statistician in the room.

What they put in

One dataset: the quarter's category survey. It held 62 attitude statements, purchase frequency, brand repertoire, occasion data, and the usual demographics.

The team opened SegmentIQ and described the goal in plain English: find groups of snack buyers that differ on what they want from the category, so the premium range can be pointed at one of them.

What SegmentIQ built

Before any clustering, SegmentIQ assessed the file. It reported respondent counts, missing values, and the spread of each variable. It flagged two attitude items with almost no variation, and the team agreed to drop them.

It then sorted the remaining variables into two groups. Attitudes and needs went in as drivers. Age, metro market, income, and purchase frequency were held out as profiling variables, so they could describe the segments without creating them. Every classification came with a reason the team could read.

The 60 attitude items were reduced to seven underlying dimensions using PCA, and the dimensions went into the clustering run. SegmentIQ ran a four-segment solution and a five-segment solution and saved both. The team compared them side by side. The fifth segment split one group into two clusters that differed on almost nothing the business could act on, so they kept four.

The four segments:

• Label Readers: health-led, check ingredients, open to paying more
• Ritual Snackers: habit-driven, brand loyal, low interest in new formats
• Deal Hunters: buy on promotion, switch freely
• Social Sharers: snack for occasions, care about how it looks on the table

What they asked

They did not ask for a list of clusters. They asked what made each group different, what each one wanted from a premium snack, and where each one would be reached.

Then they pushed on the result. Was the Label Readers group large enough to matter? Did it overlap with any other segment on the attitudes that defined it? Which variables had pulled the segments apart?

SegmentIQ answered each question inside the same conversation and showed which variables sat behind every answer.

What the segments showed, and what they did not

The age band hid the real split.

Label Readers spread across every age group in the sample. Only 31% of them fell in the 25 to 34 band. The team's original target would have missed most of the people most likely to pay for the range, and included a large number of Deal Hunters who would not.

One segment had a natural reason to buy.

Label Readers described snacking as something they wanted to feel fine about. Claims like "baked, not fried" and "no palm oil" carried weight for them. Ritual Snackers liked the product idea and saw no reason to leave their usual brand. Deal Hunters read "premium" as a price problem.

Social Sharers were the surprise.

The team had not expected an occasion-led group to matter for a health-positioned range. The profile showed they cared about pack appearance and sharing formats, suited to tailgates and potlucks, and would respond to a different message on the same product. They were flagged as a possible second wave, not a launch target.

Willingness to pay was a gap.

The survey measured how often people bought and how they felt about the category. It never asked what they would pay for a baked premium format. SegmentIQ described price attitudes for each group and stopped there. It did not produce a purchase-intent figure the data could not support. The team took that as the instruction for what the next study had to measure.

What changed

Launch targeting moved from an age band to one segment: Label Readers. Social Sharers went onto the roadmap for a later pack format.

The marketing recommendations SegmentIQ generated for Label Readers became the starting point for the creative brief. The message led on ingredients. The channel plan followed where the segment already spent its time.

The brief for the next study also changed.

• Before: Test consumer response to the premium baked range among adults 25 to 34.
• After: Among Label Readers, how much more will they pay for the baked format, and which ingredient claim carries the price?

The team exported the clustered dataset and the segment profiles to share with the agency. The executive infographic went into the launch review without anyone building a slide.