Retention and churn benchmarks when there's no cancel button

"Our repeat rate is 38%. Is that good?" That question opens every FMCG review deck and cannot be answered as asked. Over what window, in a category bought weekly or twice a year, off what penetration base? Loyalty in packaged goods means something only as a set of four numbers: penetration, buying frequency, repeat rate, and the relationship between them that Ehrenberg-Bass researchers labelled double jeopardy. Here is that benchmark set, what "good" looks like by category, and the artefacts that make a healthy brand look like it is leaking.

Why subscription-style churn math doesn't work here

A cancelation is dated. A shopper who stops buying your shampoo tells nobody: she picks up a different bottle, and the gap between her purchases was never fixed anyway. A household may buy detergent every five weeks or every eleven, depending on promotions and cupboard space.

Retention here is inferred from repeat-purchase patterns in panel and loyalty-card data rather than observed, which makes one choice matter more than any formula: the length of the measurement window. Widen it and repeat rate climbs mechanically, because every household gets more chances to come back. A soup brand can honestly report 15% repeat over four weeks and 55% over twelve months. Both are correct, and neither is comparable to the other or to a rival quoting a different window. Pick one convention, rolling 12 months for planning and quarter on quarter for in-year tracking, then hold it.

Repeat purchase rate: the core retention proxy

Repeat purchase rate (RPR) is the share of buyers in period 1 who buy the brand again in period 2.

RPR = (Buyers who purchased in Period 1 AND Period 2) / (Buyers who purchased in Period 1)

Worked example: a shelf-stable soup brand sells to 100,000 households in Q1, and 38,000 of them buy it again in Q2. RPR = 38%.

Whether 38% is good depends on the category's purchase cycle, the average gap between purchases. Toilet paper cycles in weeks. A specialty marinade may be bought twice a year, which makes a quarterly window close to meaningless for it. Comparing RPR across categories without adjusting for cycle length is the most common analyst error in this area.

Benchmarks (estimates, as of 2025 to 2026)

Repeat rate, quarter over quarter:

• Packaged food staples (per Nielsen and Kantar panel commentary summarized in trade press): a healthy household brand often sits at 40 to 60%.
• Personal care (shampoo, skincare): often 25 to 40%. Switching costs are near zero and novelty-seeking is high.
• Private label: frequently above comparable national brands in value categories, because the switching trigger (price) is already resolved once a shopper adopts the store's own brand.

Frequency and penetration, over a year:

• In most categories the average brand is bought two to four times a year per buying household, and roughly half of a brand's annual buyers buy it exactly once. The light buyer is the norm, not the lapse.
• Only about 10 to 20% of a brand's yearly buyers are 100% loyal to it, and those sole-brand loyalists are mostly light category buyers who had few occasions.
• Penetration has a hard ceiling. Kantar's Brand Footprint work has repeatedly ranked Colgate as the world's most chosen brand, reaching on the order of 60% of households globally at around three to four purchases a year. That is close to the practical maximum; most brands in a category live in single digits or the low teens.
• Frequency ceilings are set by the category. Amul tops Kantar's India rankings on consumer reachreachThe number of unique people exposed to your message in a given period. Unlike impressions, reach counts each person once, no matter how often they see it.View full definition → points largely on frequency: milk, butter and curd get bought dozens of times a year, where toothpaste gets bought three or four.
• Share of category requirements rarely goes where marketers hope. Even category leaders typically hold a quarter to 40% of their buyers' category spend. Above 50% usually points to a small buyer base or a category with little choice.

Public reference: Ehrenberg-Bass Institute, whose "How Brands Grow" research is the closest thing FMCG marketing has to an empirical bible on penetration and loyalty.

Category-switching churn: the real leakage metric

RPR says someone rebought. It says nothing about where the missing 62% went. Category-switching churn splits that:

Lost buyers = Brand switchers + Category dropouts

• Brand switchers are a marketing problem: price, promotion, distribution, shelf presence, formulation.
• Category dropouts are structural: a life stage ended, the category is declining, a substitute took the occasion.

The split changes what you can do and what the household is worth. A switcher can be won back with a price or availability move. A dropout cannot, and any household value model, including the one the sibling lesson builds, quietly assumes a category duration that has just ended for that household.

Kantar Worldpanel and Nielsen build this read by following the same households across periods and coding what else landed in the basket. Rival soup in the basket is a switch. Nothing soup-shaped is a dropout.

Penetration versus loyalty: the diagnostic test

Most brand growth in FMCG comes from penetration, more buyers, rather than from existing buyers buying more. The double jeopardy pattern says bigger brands have both more buyers and slightly higher repeat rates, and the buyer-count gap dwarfs the loyalty gap.

The test:

1. Compute share of category requirements (SCR): of everything the household buys in the category, what share goes to you?
2. Compute penetration: what share of category buyers bought you at all, even once?
3. Read both against the category norm line, the expected relationship between the two at your size.

Illustrative:

• Brand A: 5% penetration, 55% repeat rate.
• Brand B: 30% penetration, 45% repeat rate.

Brand A looks more loyal. But a brand at 5% penetration is expected to have a *lower* repeat rate than one at 30%, so A is mildly overperforming its norm while sitting on a tiny base. Losing two points of B's penetration costs more volume than everything happening inside A.

Before you credit an above-norm repeat rate to affection, check availability. A brand with 90% distribution in three states and 4% nationally shows inflated loyalty among its buyers, because those buyers face fewer alternatives on that shelf. Deviations from the norm line are usually explained by distribution, pack range or a distinct purchase occasion.

Building it from data: a simple structure

If you're working with loyalty-card or panel-style transaction data, here's the shape of the calculation in pseudocode:

# household-level transactions, one row per purchase
# columns: household_id, brand, category, period

period1_buyers = set(df[(df.brand == "OurBrand") & (df.period == "Q1")].household_id)
period2_buyers = set(df[(df.brand == "OurBrand") & (df.period == "Q2")].household_id)

repeat_buyers = period1_buyers & period2_buyers
rpr = len(repeat_buyers) / len(period1_buyers)

lapsed = period1_buyers - period2_buyers
still_in_category = set(df[(df.category == "Soup") & (df.period == "Q2")].household_id)

brand_switchers = lapsed & still_in_category
category_dropouts = lapsed - still_in_category

Deliberately simple. Real panel analytics add weighting (panels are samples projected to the total market) and multi-period smoothing so one slow week does not distort the picture.

Watch out for these traps

• Seasonality masquerading as churn. Sunscreen "lapses" every winter. Compare year over year, not just sequential quarters.
• Pack-size effects. A household moving to a larger multipack buys less often and looks lapsed in a naive count while volume holds or grows.
• Promotion-driven false loyalty. Repeat rates built during deep discounting collapse when price resets. Read RPR at full price and on deal separately.
• Regression to the mean in loyalty programs. Target last year's heaviest buyers and they will buy less this year almost regardless of what you do, because extreme behavior in one period reverts in the next. Judge the program against a matched control, not against participants' own history.
• Budget aimed at the wrong tail. With half your buyers buying once a year, a retention plan designed for heavy users addresses a minority of volume, and the cost per incremental household the acquisition lesson prices is usually the fairer comparison.

🎬 [VIDEO: "How Brands Grow: Byron Sharp explains the laws of marketing" - youtube.com - a condensed explainer of the Ehrenberg-Bass penetration and loyalty findings underpinning this lesson]

Key takeaways

• Retention is inferred from repeat purchase rate, and the figure is meaningless until you state the window and the purchase cycle. Same brand: 15% over four weeks, 55% over a year.
• Directional repeat benchmarks: food staples 40 to 60% quarterly, personal care 25 to 40%, private label typically above national brands in value categories. Check current Kantar Worldpanel or Nielsen reporting before quoting them.
• Loyalty norms sit lower than most decks assume: two to four purchases a year per buying household, about half of buyers buying once, 10 to 20% sole-brand loyals, leaders holding a quarter to 40% of their buyers' category spend.
• Penetration is where the ceiling lives. Colgate's roughly 60% global household reach is near the maximum, and double jeopardy predicts small brands to have lower repeat rates, so read repeat and penetration jointly against the norm line.
• Split lost buyers into brand switchers (fixable with price, promo, distribution) and category dropouts (structural), and rule out seasonality, pack-size shifts, discount inflation and regression to the mean before calling anything churn.