Segmentation almost always sorts people. Shopper mission research sorts trips - and the evidence says the trip is often the better unit of analysis, because the same person behaves like three different customers in a single week.
This is not a small correction to segmentation. It is a claim that the unit most teams have chosen is the wrong one for a large class of decisions.
The short answer
A shopper mission is the job a specific trip is doing: a weekly stock-up, a mid-week fill-in, a top-up for tonight's dinner, a special-purpose hunt. Missions differ in basket size, price sensitivity, planning horizon, channel, and tolerance for substitution. Crucially, one household runs several missions, so any statement of the form "our customer is price sensitive" is an average across occasions where that is emphatically true and occasions where it is not.
What the corpus already covers, and where this differs
Our documentation guide to behavioral segmentation includes occasion-based segmentation as one of its eight types, and it is a good treatment: it groups customers by when they engage, such as a coffee app distinguishing a morning commute from an afternoon meeting.
This article takes the opposite position on the modelling question, and the distinction matters. Occasion-based segmentation as normally practised still assigns a person to a segment - it treats "morning commute shopper" as a kind of customer. Mission research says the person is not the unit at all. The same individual is a morning-commute shopper on Tuesday and a Saturday stock-up shopper four days later, and neither label describes them. If you assign them to one, you have averaged two incompatible occasions and produced a persona that describes nobody precisely.
The test is empirical rather than philosophical: does the variation within a person across trips exceed the variation between people? In grocery and most repeat-purchase categories, it very often does. Where it does, person-level segmentation is measuring the wrong variance.
The mission mix is moving, fast
This is not a stable background condition you can model once. Kantar data reported in mid-2026 shows the mix shifting substantially over five years.
| Mission | Share of trips (Q1 2026) | Share in 2021 | Direction |
|---|---|---|---|
| Fill-in (mid-week corrective) | 37% | - | Now the dominant mission |
| Stock-up (major weekly shop) | 33% | 39% | Down 6 points |
| Same-day consumption | 30% | 21% | Up 9 points |
Only one third of shoppers described their most recent trip as a major stock-up. The traditional big weekly shop that most category plans are implicitly built around is now a minority occasion.
Foot traffic data points the same way. Placer.ai found that visits under 15 minutes made up over 40% of grocery visits nationwide in 2025, up from 37.9% in 2022, with the biggest year-over-year gains going to the shortest trips. Between 2024 and 2025, grocery stores took an increasing share of short midday visits of under ten minutes.
A shopper on a nine-minute trip is not evaluating your pack architecture, comparing unit prices, or noticing your new claim. They are executing a retrieval task. Research designed around considered choice is measuring a decision process that a growing plurality of trips does not contain.
Even the measurement vendors concede the gap
NielsenIQ's own framing of trip mission segmentation says it "reveals the intent, urgency, and value behind each visit to the shelf or digital cart", and argues that "understanding what consumers buy is no longer enough". Their taxonomy names five missions: stock-up, fill-in, special-purpose, immediate-need, and routine or habitual.
That concession is worth sitting with. The largest retail measurement businesses in the world are telling their clients that the why is now the binding constraint - and point-of-sale data structurally cannot supply it, because intent is not a field in a transaction record. Notably, NIQ's public treatment names the missions but publishes no mission-mix percentages, basket sizes by mission, or quantified behavioural differences.
What actually changes between missions
| Dimension | Stock-up | Fill-in | Immediate need |
|---|---|---|---|
| Planning | List-driven, often at home | Partial list, mostly recalled | None |
| Price sensitivity | High, unit price compared | Moderate | Low, convenience dominates |
| Substitution tolerance | Low, will defer to next trip | High, takes what is there | Very high, any workable option |
| Pack preference | Larger formats, multipacks | Standard | Single serve, smaller |
| New product openness | Low, risk to the week | Moderate | Low, wrong moment |
| Time in store | Long | Short | Under 10 minutes |
Read the substitution row against your availability data and something uncomfortable follows. A stock-up shopper who cannot find your product defers and buys it next time, so you keep the sale and lose nothing visible. An immediate-need shopper takes whatever is there and may never return to your brand. The same out-of-stock event has completely different consequences depending on which mission was interrupted, and your availability reporting almost certainly does not distinguish them.
How to research missions properly
The methodological problem is that missions are episodic, and people are unreliable narrators of their own habits. Ask "how do you usually shop" and you get a tidy self-description of the stock-up mission, because that is the one people think of as shopping. The fill-in and immediate-need trips - now 67% of trips between them - are precisely the ones that do not register as events worth reporting.
The fixes are known:
- Anchor to a specific trip, never to a habit. "Tell me about the last time you bought this" outperforms "how often do you buy this" by a wide margin. See the critical incident technique.
- Sample trips, not people. Your sampling frame should be occasions. Recruiting 200 shoppers and asking each about one trip gives you 200 trips; recruiting 50 and asking each about four gives you 200 trips and lets you measure within-person variation directly, which is the whole question.
- Capture the trip close to when it happened. Diary studies exist for this reason, though they are slow and expensive in classic form.
- Ask what the trip was for before asking what was bought. Establishing the mission first prevents the shopper from rationalising the basket into a coherent plan after the fact - a reliable distortion covered in stated versus revealed preferences.
Where Koji fits
The reason mission research has stayed a large-agency product is cost. Getting 200 trip reconstructions through human moderators is weeks of scheduling and a five-figure invoice, and the trips have gone cold by the time you finish.
Koji runs AI-moderated voice interviews in parallel, so 200 trip reconstructions take hours rather than weeks, and can be triggered close enough to the trip that recall is still good. Every respondent gets the same brief and the same follow-up logic, which matters here specifically: a human moderator who finds the stock-up stories richer will probe them harder and manufacture a difference between missions that is really a difference in questioning.
The six structured question types let one session capture both the narrative and the numbers. Use open_ended to reconstruct the trip in the shopper's own words, single_choice to classify the mission, multiple_choice for every trigger that applied, scale for planning degree and time pressure, ranking to order what drove the store choice, and yes_no for the substitution counterfactual - would you have waited. Because the closed types mean the same thing regardless of how long the conversation ran, they give you a check on whether any open-ended difference between missions is real or an artifact of probing depth.
Run that design once and you can finally answer the question person-level segmentation cannot: not who your shopper is, but which occasion you are winning and which one you are absent from.
Frequently Asked Questions
What is a shopper mission?
A shopper mission is the job a particular shopping trip is doing. Common missions include the weekly stock-up, the mid-week fill-in, the immediate-need or same-day consumption trip, the special-purpose hunt, and the routine habitual visit. Missions differ in basket size, price sensitivity, planning, and willingness to substitute.
How is mission research different from customer segmentation?
Customer segmentation assigns people to groups and assumes the group describes them consistently. Mission research treats the trip as the unit of analysis, because the same person runs different missions in the same week and behaves differently on each. Where variation within a person across trips exceeds variation between people, person-level segments measure the wrong variance.
What percentage of grocery trips are stock-up trips?
Kantar data for Q1 2026 shows only 33% of shoppers described their most recent trip as a major stock-up, down from 39% in 2021. Fill-in trips are now the largest group at 37%, and same-day consumption trips have grown from 21% to 30% over the same period.
Are shopping trips getting shorter?
Yes. Placer.ai found visits under 15 minutes made up over 40% of US grocery visits in 2025, up from 37.9% in 2022, with the largest gains in the shortest trips. Grocery stores also captured a growing share of sub-ten-minute midday visits between 2024 and 2025.
Why can point-of-sale data not identify shopper missions?
Intent is not a field in a transaction record. Point-of-sale data captures what was bought, when, and at what price, but not what the trip was for. A small basket could be a fill-in, an interrupted stock-up, or a top-up for tonight, and all three look identical in the log. NielsenIQ itself argues that understanding what consumers buy is no longer enough.
How many interviews do I need for shopper mission research?
Sample trips rather than people. Asking 50 shoppers about four trips each yields 200 trip observations and lets you measure how much a single person varies across occasions, which is the central question. That within-person comparison is more informative for mission work than a larger sample of one trip each.
Related reading
- Market Basket Analysis: What Co-Purchase Data Proves and What It Cannot
- Cross-Shop Analysis: Why No Dataset Contains a Whole Shopping Trip
- Category Entry Points: How to Find the Buying Situations That Create Demand
- Private Label vs National Brands: How to Research Why Shoppers Actually Switch
- B2C User Research: How to Understand Consumer Behavior at Scale
