Private label is no longer a price story, and the 2026 data is what refutes it.
In June 2026, FMI (The Food Industry Association) published Power of Private Brands 2026, a survey of 1,495 US grocery shoppers fielded 13-18 March 2026 with a margin of error of plus or minus 2.5 percentage points. Its headline finding: 94% of shoppers said they would keep buying store brands even if grocery prices decline. The variable that supposedly caused the switch is being removed, and almost nobody plans to switch back.
That single number breaks the explanation most brand teams are still working from, and it should change what they research.
Answer first
Private label share data tells you that shoppers switched. It cannot tell you why, and it cannot tell you whether they will switch back. Point-of-sale and household panel data record the outcome of a substitution made at the shelf. They do not record the constraint the shopper was under, the alternative they considered, the standard the product had to clear, or what would have happened if your pack had been in stock. To recover any of that you have to ask the person who made the decision, close to when they made it.
The practical answer in 2026 is an AI-moderated interview study: a few hundred recent category buyers, each one probed on the specific occasion, with the analysis returned as quantified themes instead of a folder of recordings. That is what Koji does, and it is why this class of question has moved from a two-month agency project to a two-day study.
The 2026 numbers, and what each one is measuring
| Finding | Figure | Source and period |
|---|---|---|
| US store brand dollar sales | $282.8 billion, up more than $9 billion year on year | PLMA, from Circana Unify+, 52 weeks ending 28 Dec 2025 |
| Store brand dollar growth vs national brands | 3.3% vs 1.2% | Same |
| Store brand unit volume | 68.7 billion units, up 434.3 million (0.6%); national brands down 0.6% | Same |
| Dollar share, 5-year move | 19.1% to 21.3% | PLMA / Circana, 2021 to 2025 |
| Unit share, record | 23.8% at midyear 2026 | PLMA / Circana, six months ending 14 June 2026 |
| Shoppers who will keep buying store brands if prices fall | 94% | FMI Power of Private Brands 2026, n=1,495 |
| Quality cited as a purchase driver | 39% in 2026, up from 30% in 2023 | Same |
| Increased private brand purchases in the past year | 49%, vs 31% who increased national brand purchases | Same |
Read those two blocks against each other. The first block is transaction data: it is precise, it is auditable, and every one of your competitors buys the same numbers. The second block is stated behaviour and intent, and it is the only part that carries a reason.
Notice also that unit share (23.8%) runs well ahead of dollar share (21.3%). Private label is still cheaper per unit. That gap is exactly why the price explanation feels obviously true and why it keeps surviving without evidence.
The switch-back asymmetry
Here is the framework that reorganises this whole problem, and it is the reason most private label win-back programmes fail.
Trial and retention have different causes, and brands research the trial cause.
- What caused the trial is usually situational and often external to the product: a price gap, a promotion, a stock-out, a new store format, a budget shock.
- What caused the retention is almost always that the product cleared a standard the shopper was privately braced to see it fail.
These are not two strengths of the same force. They are different forces, and only the second one is load-bearing after the first is removed. The FMI 94% is the asymmetry stated as a number: the trial cause (price) is being withdrawn, and the retention cause (it was fine) is holding.
The operational consequence is blunt. A win-back strategy built on closing the price gap is aimed at a cause that has already finished acting. You would be paying to remove a barrier the shopper walked past two years ago. The question worth answering is the second one: what standard did the store brand clear, in which categories did it fail to clear it, and what would it take for this shopper to pay a premium again in this specific category.
That question has no answer in scanner data. It is a memory held by a person.
Three causes of a private label switch, and the instrument each one needs
Private label switching is not one behaviour. Treating it as one is the most common design error in this category, and it produces research that averages three different populations into a finding that describes none of them.
| Switch type | What actually happened | What it predicts | Instrument that recovers it |
|---|---|---|---|
| Deliberate value switch | Shopper compared, judged the gap not worth it, chose the store brand | Sticky. Reverses only on a quality failure | Interview on the comparison moment: what was compared, what was assumed |
| Forced substitution | Your pack was out of stock, wrong size, or delisted; shopper took what was there | Highly reversible, but only if you know it happened | Interview on the specific trip; the shopper remembers the empty shelf |
| Category indifference | Shopper never rated the category worth attention; brand was habit, not preference | Permanent. Will not reverse at any price | Interview on category involvement and how the decision is made |
Only the middle row is a distribution problem. Only the top row is a product and pricing problem. Only the bottom row tells you to stop investing. In point-of-sale data all three look identical: a unit that used to be yours is now theirs.
This is also where the survivorship bias trap bites hardest. If you interview your current buyers about why they stay loyal, you have sampled exactly the population that did not switch, and you will conclude that your brand equity is intact right up until the quarter it is not.
How to design the study
The instrument has to do two things at once: quantify how much of the switching falls into each bucket, and capture the reasoning in the shopper own words. A survey does the first and not the second. Traditional depth interviews do the second at a sample size too small to do the first.
Koji studies combine both, because structured questions run inside a conversational AI-moderated interview. All six question types are available in the same session: open_ended for the account of the occasion, scale for how large the perceived quality gap was, single_choice for which brand was displaced, multiple_choice for every trigger present on the trip, ranking for what would have to change to earn a premium again, and yes_no for whether the shopper checked the national brand price at all. The AI follows up on each answer instead of accepting the first sentence.
A workable design:
- Screen to recent behaviour, not to attitude. Recruit people who bought the category in the past 30 days, then split by whether they bought store brand, national brand, or both. Never screen on self-declared brand loyalty; it is a construct people answer aspirationally.
- Anchor on one trip, not on habits. Ask about the most recent purchase occasion. Generic questions produce generic answers, and this is the single biggest quality lever in the whole design. The same principle underlies jobs-to-be-done interviews and category entry point research.
- Ask what was on the shelf. This is the question almost nobody asks, and it is what separates a value switch from a forced substitution. It also feeds directly into the shelf-space and assortment problem.
- Probe the quality expectation before and after. The FMI quality shift (30% to 39%) is a population-level average. You need the category-level version, because it varies enormously by category and the average is useless for your decision.
- Do not ask people to predict price sensitivity in the abstract. If price is genuinely part of the decision, use a real method: Van Westendorp, Gabor-Granger, or conjoint via the pricing research guide.
Why the legacy options struggle here
Syndicated retail measurement (NielsenIQ, Circana, Numerator) is excellent at the first block of the table above and structurally incapable of the second. It is the right purchase for share tracking and the wrong purchase for causation; we cover the boundary in syndicated data vs custom research and compare the vendors in NielsenIQ vs Circana vs Numerator.
Survey platforms like SurveyMonkey, Typeform and Qualtrics will field a private label questionnaire quickly, but they cannot probe. When a shopper writes "it was cheaper," a survey records "it was cheaper." An interviewer asks what they compared it to, whether they had bought it before, and what they expected to be worse. That follow-up is where the switch-back asymmetry becomes visible, and it is the thing a static form cannot do at any sample size.
Consumer panels such as YouGov, Attest and Toluna deliver reach but sell you their respondents and their moderation model. Traditional qualitative vendors deliver depth at 8 to 12 interviews, which cannot support a bucket-level split across three switch types.
Koji runs AI-moderated voice interviews that probe like a researcher and scale like a survey. Thematic analysis is automatic, so the output is quantified themes with the supporting customer quotes attached, not a transcript pile. No moderator fatigue on interview 300, and no interviewer bias drifting across a fieldwork window.
What to do with the finding
The deliverable is a per-category allocation across the three switch types plus the standard the store brand cleared. That drives three different decisions:
- Forced substitution is high: this is a distribution and availability problem. Fix it with the retailer before spending anything on marketing.
- Deliberate value switching is high and quality perception is at parity: the premium is no longer defensible on brand alone. Either close the performance gap on the attribute shoppers named, or restructure the pack and price to compete on a different basis.
- Category indifference is high: stop trying to win the argument. Redirect the budget to categories where the decision is still live.
Most importantly, stop testing the price hypothesis you have already been given the answer to. 94% of shoppers told FMI that price is not what is holding them.
Run this study with Koji
Koji is the AI-native customer research platform built for exactly this question. Set the brief in plain language, launch an AI-moderated voice or text interview study, and get quantified themes, segment cuts and a one-click report while the category review is still open. Structured questions give you the numbers; the conversational AI gives you the reason behind them. No research headcount required, no six-figure agency retainer, and results in hours rather than the six to eight weeks a traditional private label deep-dive takes.
If you are defending shelf space this quarter, the reason your buyers switched is the most valuable thing you do not currently have. Start a free Koji study and have it by Friday.
Related reading
- Price Pack Architecture (2026) - why retail price setters invert your pack-price ladder on your best-selling size.
- Shrinkflation Research (2026) - why scanner data and fairness experiments disagree, and which shoppers each one measures.
- Personalized Pricing in 2026 - what individualised prices do to the reference price your research assumes.
Frequently Asked Questions
What is the current private label market share in the US?
Store brands reached a record 23.8% unit share of the US market at midyear 2026, according to Circana data published by PLMA for the six months ending 14 June 2026. Dollar share is lower, at 21.3% for full-year 2025, because private label carries a lower average price per unit. Total US store brand sales were $282.8 billion in 2025, up more than $9 billion year on year.
Are shoppers buying private label only because of inflation?
The 2026 evidence says no. FMI Power of Private Brands 2026 (n=1,495, fielded March 2026) found 94% of shoppers would keep buying store brands even if grocery prices decline, and quality has risen as a stated purchase driver from 30% in 2023 to 39% in 2026. Price explains a large share of initial trial; it does not explain retention, and retention is what determines whether the share shift is permanent.
Why can't point-of-sale data tell me why shoppers switched?
Point-of-sale and household panel data record what was bought, not why. They contain no record of what else was on the shelf, what the shopper compared, what they expected, or what they would have done under a different condition. A deliberate value switch, a forced substitution caused by an out-of-stock, and simple category indifference all appear in scanner data as the same event: a unit moving from one brand to another.
How many interviews do I need to research private label switching?
Enough to split the sample across switch types and key categories, which in practice means 150 to 400 recent category buyers rather than the 8 to 12 typical of traditional qualitative work. Because Koji runs AI-moderated interviews in parallel, that sample is achievable in a day or two, which is what makes a quantified bucket-level split possible at all.
What questions should I ask private label switchers?
Anchor every question on the most recent specific purchase occasion rather than on habits. Ask what was physically on the shelf, what was compared, what the shopper expected to be worse about the store brand, whether that expectation was met, and what would have to change for them to pay the premium again. Combine open-ended probing with scale and ranking questions so the result is quantifiable.
Is private label growth reversible for national brands?
Partially, and only in specific segments. Switching driven by out-of-stocks and assortment gaps reverses quickly once availability is fixed. Switching driven by a satisfied quality expectation is sticky, and switching driven by category indifference is effectively permanent. The only way to know your own mix is to measure it per category, because the national averages hide enormous variation.
