Trial vs Repeat Purchase: The Most Expensive Way to Die in CPG

A brand can grow household penetration every quarter for a year and still be dying. This is not a paradox. It is the default outcome when trial acquisition outpaces repeat purchase, and most brands cannot tell the difference until the acquisition budget stops and the topline collapses in the same month.

The number that makes every board deck look good (new households reached) is also the number that hides the most dangerous failure mode in consumer packaged goods. Rising penetration does not mean the product is working. It means people are trying it. Whether they come back is the trial vs repeat purchase question, and most brands are not asking it.

Penetration growth without repeat is a treadmill

The math is simple and the implication is severe. A brand acquires 50,000 new households in Q1. If 35% repeat within the standard purchase cycle, the brand retains 17,500 buyers and needs to acquire another 32,500 just to hold flat in Q2. If 15% repeat (a number that looks fine buried in a quarterly review), the brand retains 7,500 and needs 42,500 new households to show the same topline. That is not growth. That is a treadmill that accelerates every quarter.

Bain & Company's research with Kantar Worldpanel found that across 10,000+ CPG brands in 35 countries, penetration (the number of buyers) is the primary driver of market share, not purchase frequency. Byron Sharp's How Brands Grow codified the same finding: brands grow by reaching more people, not by making existing buyers more loyal. This is correct and widely cited. What gets lost in the citation is the assumption underneath it: that a meaningful fraction of those new buyers stick.

When they do not stick, penetration becomes a lagging indicator of marketing spend, not a leading indicator of brand health. NielsenIQ's 2024 innovation report found that only 13% of new CPG products survive their second year in market. The primary failure mode is not lack of trial: most launches generate adequate first-purchase numbers through slotting, sampling, and introductory trade promotion. The failure is that trial buyers do not convert to repeat buyers, and the brand cannot sustain the acquisition cost indefinitely.

The repeat window reveals the product, not the marketing

Trial is a function of distribution, trade spend, and promotional execution. A $50,000 demo program at Whole Foods will generate first-time buyers. A BOGO at Kroger will generate first-time buyers. An endcap at Sprouts will generate first-time buyers. None of these events tell you whether the product earns a second purchase on its own merits.

Repeat rate (the percentage of first-time buyers who purchase again within a defined window) isolates the product from the promotion. McKinsey's consumer-practice research treats repeat purchase as a core signal of long-term brand viability in grocery. A 40% repeat rate within 26 weeks means the product has a consumer franchise: people who tried it, liked it, and came back without being bribed. A 15% repeat rate means the product or the positioning has a problem that more distribution will not fix.

The distinction matters because the interventions are opposites. Low trial with adequate repeat is a distribution and awareness problem: solve it with more doors, more trade, more visibility. High trial with low repeat is a product or value-proposition problem: solve it with reformulation, repositioning, or price architecture. Solving a repeat problem with more trial spend is the CPG equivalent of filling a leaking bucket by turning up the faucet. The water level looks stable until you run out of water pressure.

What a trial-heavy, repeat-light portfolio actually costs

The financial damage compounds in ways that do not appear on a standard P&L.

Acquisition cost per retained household escalates. If each new household costs $4.50 to acquire through trade and promotion, and only 15% repeat, the effective cost per retained household is $30. At a 35% repeat rate, the same $4.50 acquisition produces a retained household at $12.86. The brand with 15% repeat is paying 2.3x more per customer it actually keeps. That ratio worsens every quarter as the easiest-to-reach households are acquired first.

Trade spend efficiency collapses. Trade promotion spending runs 15-30% of gross revenue for CPG brands in conventional grocery. When that spend generates trial buyers who do not repeat, the ROI calculation changes fundamentally. A $200K promotional investment that generates 40,000 trial households and 6,000 retained households is a different proposition than one generating 40,000 trial and 14,000 retained. The trial numbers are identical. The business outcomes are not.

Retailer confidence erodes. Category managers track velocity after the introductory promotional period ends. A product that scans well during a launch promotion and drops 60% in the following period tells the buyer exactly what happened: the promotion drove trial, the product did not drive repeat, and the shelf space is underperforming. That velocity decay is the data behind the line review conversation where the buyer suggests "rationalizing" your slowest movers. NielsenIQ's innovation analysis points to post-promotion velocity decay as a leading indicator of second-year discontinuation.

The items that look great in month one

The pattern is specific and recognizable. A new item launches with strong trade support. First-scan velocity is high. The team celebrates. Twelve weeks later, velocity has declined 40-60% from the promoted peak. The explanation offered is that promotion ended and the item "needs time to build organic pull." The actual explanation is that trial buyers tried and did not return.

This is the item that gets a second promotional push, then a third, each time generating another spike-and-decay pattern. After a year of "successful" promotions, each one hitting its trial target, the item is discontinued because baseline velocity never reached the threshold the buyer set. The trial metrics were green the entire time. The repeat metric, which nobody was tracking at the item level, was red from week four.

A brand with five items in this pattern is spending six figures annually on promotions that generate trial for products that do not convert. That spend is not just wasted. It crowds out investment in items where the repeat rate indicates a real consumer franchise worth expanding.

Trial vs repeat purchase separates a brand from a promotion

The measurement is straightforward but requires household-level purchase data, not just POS totals. For each item and each period: how many households bought for the first time (trial), and of the households that bought in a prior period, how many bought again (repeat). The ratio between these two numbers, and the trend in that ratio, tells you whether the brand is building a customer base or renting one.

When trial is high and repeat is low, the diagnosis branches:

• Product issue. The consumer tried it and the experience did not justify a second purchase at the shelf price. Reformulation, not more marketing.
• Positioning mismatch. The promotion attracted the wrong buyer. A heavy-discount BOGO draws price-sensitive shoppers who will not pay full price on the next trip. The trial number was real; the audience was wrong.
• Price architecture. The product works but the everyday price is above the consumer's repurchase threshold for the category. The trial happened at a promotional price the brand cannot sustain.
• Purchase cycle mismatch. The repeat window is set shorter than the natural consumption cycle for the product format. A 26-week window for a sauce used monthly will undercount true repeat. This is the least common explanation and the first one brands reach for.

Each diagnosis leads to a different intervention. Treating all of them as "we need more awareness" (the default response to soft post-launch velocity) wastes money and delays the real fix.

What Leaky Bucket measures

Leaky Bucket is the fourth tool in the Cinderhaven household-penetration series, after Door Math (are you on the shelf), Spin Rate (are you selling), and Void Finder (where are you missing). It answers the question the first three cannot: of the people who tried you, how many came back?

The tool separates trial households from repeat households by item and period, flags items where the trial-to-repeat ratio indicates a product or positioning problem rather than a distribution problem, and surfaces the items that looked like successful launches but are running on acquisition spend with no organic consumer franchise underneath.

It runs on Cinderhaven Provisions: a synthetic ~$25M specialty food brand with household-panel data. The company is invented; the methodology is the same one NielsenIQ and Circana use on real panel data.

(Leaky Bucket is in development. Get in touch if you want to know when it's live.)

Thirty minutes: tell me your best-selling new launch, and I'll show you whether the penetration number is a brand or just a promotion.