CPG Cross-Shop: The Five Shapes and What They Mean

Ask five CPG analysts for a cross-shop number and you can get five honest answers that disagree. Not because anyone did the math wrong, but because "cross-shop" names at least five different questions about the same shoppers. One is about adjacent categories, one about a rival brand, one about which stores a buyer uses, one about online versus in-store, and one about what lands in the same cart. Each moves for different reasons and each calls for a different response. Treat them as one metric and the response you fund is often aimed at a problem you do not have.

Why One Word Covers Five Questions

The word stuck because every version shares a skeleton: take a defined set of shoppers, look at what else they bought, and express the overlap as a share. What changes is the "what else." Swap the comparison object from another category to a rival brand, a different retailer, a different channel, or simply the rest of the receipt, and the same skeleton produces five different numbers. A brief that says "run a cross-shop on our salsa buyers" is underspecified until it names which of those comparison objects the team cares about.

The other source of confusion is grain. Some shapes count people: of the households that bought our brand, how many also bought X. Others count transactions: of the baskets containing our item, how many also contained X. Buyer-level and basket-level versions of the "same" question can diverge sharply, because one loyal household buying both items every week contributes one buyer but many baskets. A deck that switches grain between slides without saying so is the usual reason two cross-shop charts in the same meeting cannot be reconciled.

Shape 1: Category Cross-Shop (Adjacency)

Category cross-shop asks which other categories your buyers also purchase, either on the same trip or within a defined period. Salsa buyers who also buy tortilla chips is the textbook case, but the useful findings are usually less obvious: a premium coffee brand whose buyers over-index on a particular creamer segment, or a pet-food brand whose buyers also buy a specific kind of cleaning product. The point is functional or lifestyle relationship, and the question is whether a buyer of your category is more likely than an average shopper to buy the adjacent one.

That last clause matters. Raw overlap is not adjacency. If most households buy milk, most of your buyers will too, and the overlap tells you nothing about your brand. A meaningful category cross-shop compares your buyers' rate of purchasing the adjacent category against the rate for a relevant baseline population, so the index reflects a real tilt rather than a popular category. The decisions this shape supports are adjacency moves: bundles, shared displays, co-marketing with a brand in the neighboring category, and arguments to a category manager about where your product belongs on the shelf.

When category cross-shop falls, the likely causes are relationship-level. The adjacent category may be shrinking, a usage occasion may be fading, or your buyer base may be shifting toward households with a different lifestyle mix. The response is to revisit the pairing or the occasion, not to cut price.

Shape 2: Brand-Switching or Competitive Cross-Shop

Brand-switching cross-shop stays inside your category. Of the households that buy your brand, how many also buy a direct competitor, and which way is share of wallet flowing between the two? Overlap alone is a weak read here, because some duplication is normal in most categories: plenty of households keep two brands in rotation. The signal is direction and trend. A buyer whose category spend is steady but whose share for your brand is sliding toward the rival is a different case from a buyer who bought the rival once during a deep promotion and came straight back.

This shape needs buyer-level history with enough purchases per household to see share move over time, which is why it is usually built on item-level, receipt-linked data rather than store-level sales. It feeds competitive displacement campaigns, win-back targeting, and decisions about when a loyalty program should intervene. If you are building that intervention, the logic in lapsed buyer trigger design patterns applies directly: a buyer drifting to a competitor is an early-stage lapse, and the trigger should fire before the switch is complete. For teams comparing their position against named competitors over time, transaction-level brand-switching reads are a useful complement to the foot-traffic and share-of-visit reads in a competitive benchmarking program.

When brand-switching cross-shop rises in the competitor's favor, the cause is competitive: a rival promotion, a new pack size, a price gap that widened, or an out-of-stock that pushed your buyers to the next facing. The fix is aimed at the competitor and at the specific households drifting, not at the adjacent category.

Shape 3: Retailer or Banner Cross-Shop

Banner cross-shop asks whether the same shopper buys your brand across more than one retail banner, or buys it at one banner only. A household might buy your snack at a mass retailer on the big monthly trip and at a grocery chain on top-up trips, or it might buy your snack exclusively at the grocery chain while buying everything else at the mass retailer. Those two households look identical in brand-level sales and very different in this analysis.

The second household is the interesting one. A buyer who shops a banner regularly but never buys your brand there is a candidate distribution gap: either the item is not stocked at that banner, it is stocked but poorly placed, or it is stocked at a price point that pushes the purchase elsewhere. Banner cross-shop is also what gives a sales team evidence in a trade negotiation, because it shows how much of a retailer's shopper base already buys the brand somewhere else and could be captured in-store. It informs banner-exclusive promotion design as well, since a promotion aimed at buyers who are loyal to one banner for your brand has different economics from one aimed at buyers who split.

When banner cross-shop drops, the first suspect is availability, not demand. A delisting, a reset that moved the item, or a supply problem at one banner shows up here before it shows up anywhere else.

Shape 4: Channel Cross-Shop (Online Versus In-Store)

Channel cross-shop is a household-level question about where a brand is bought: e-commerce, brick-and-mortar, or both. The business question underneath it is whether online purchases are incremental or cannibalizing. If households that start buying your brand online keep their in-store purchase rate and add online orders on top, the online channel is growing the franchise. If their in-store purchases fall by roughly what they now buy online, the channel is relocating demand, and the economics of that relocation depend on margin, fulfillment cost, and retail media spend in each channel.

That distinction is what should drive the split of investment between retail media and e-commerce on one side and in-store merchandising on the other. A brand that sees online sales growing and shifts budget online, without checking whether those households simply stopped buying in-store, can end up paying twice to reach the same buyer. Reading this shape needs receipts from both channels tied to the same household, which is a coverage question to put to any data vendor early.

Shape 5: Basket Co-Occurrence

Basket co-occurrence asks which items show up in the same transaction as yours, regardless of whether they have anything to do with each other. It is the most mechanical of the five shapes and the easiest to compute, which is why it is the one most often presented as "cross-shop" when someone asks for cross-shop. It is also the one most likely to be over-read.

The problem is that a weekly shopping list puts a lot of unrelated items in the same cart. Bananas, bread, and dish soap co-occur with nearly everything because nearly everyone buys them on a routine trip. That co-occurrence reflects how people shop, not a relationship between the products. Association-rule mining, which grew out of supermarket transaction data, handles this with a measure called lift: the ratio of observed co-occurrence to what would be expected if the items were independent. A pairing that co-occurs often but only at the rate its popularity predicts has little lift, and treating it as an adjacency opportunity is a mistake.

Used with that caveat, basket co-occurrence is still valuable. It is the right input for cross-merchandising placement, secondary displays, and basket-building promotions where the goal is to add an item to a trip that is already happening. The earlier piece on what CPG brands can learn from transaction-level purchase data shows how basket composition can reveal retailer-specific shopper segments. What basket co-occurrence should not do is stand in for category cross-shop. One says the items ride together; the other says the same buyers are drawn to both. Those are different claims, and the second one is what justifies a co-marketing budget.

Diagnose the Shape Before Choosing the Fix

The practical cost of conflating these shapes shows up when a number moves. Suppose a brand team watches a blended cross-shop score decline over two quarters. The instinct is to respond with whatever the team usually does: a bundle, a co-marketing deal, a promotion. But each shape has its own failure pattern and its own remedy:

• Falling category cross-shop points to a weakening adjacency or occasion. Respond with bundling, co-marketing, or a revised pairing.
• Rising competitive cross-shop in a rival's favor points to competitive loss. Respond with displacement offers and win-back targeting aimed at the drifting households.
• Falling banner cross-shop points to availability. Check distribution, placement, and pricing at the banner where buyers stopped finding you.
• A shift in channel cross-shop points to demand relocating. Revisit media allocation before assuming growth in one channel is net new.
• A change in basket co-occurrence often reflects trip patterns, such as a shift toward smaller top-up trips, rather than any change in how shoppers relate to your product.

Apply the wrong remedy and the team burns a trade-spend cycle on a problem it did not have, while the real one keeps running. A co-marketing program will not fix a delisting. An adjacency bundle will not recover households who moved to a competitor on price unless it is rebuilt as a value offer aimed at that specific gap. And a promotion launched because basket co-occurrence dipped may simply be reacting to a change in how often people shop.

Decomposing the score is mostly a data-resolution question. Category, brand, and banner cross-shop all require knowing what a specific household bought, item by item, across stores and over time. Store-level sales and category totals cannot produce them. GSDSI's CPG transaction feed is built for that grain: item-level, receipt-linked purchase data covering roughly 71.5M daily active users and about 390M purchase signals per day, the resolution that category, brand-switching, and basket-level shapes need instead of store totals. Confirm banner and channel coverage for your specific categories before relying on any feed for the banner or channel shapes, since cross-retailer and online-versus-in-store linkage is a separate coverage question from item-level depth.

Three Questions to Ask Any Cross-Shop Report

Whether the report comes from an internal analytics team, a syndicated provider, or a dashboard built on a licensed feed, three questions sort out most of the ambiguity before anyone acts on it:

1. Which shape does this measure? Category, competitive brand, banner, channel, or basket co-occurrence. If the answer is "all of them combined," ask for the components.
2. Over what trip window? Same trip, same week, a rolling period, or a full year. Same-trip and period-level versions of category cross-shop can tell very different stories, and a window shorter than the category's purchase cycle will undercount real overlap.
3. Is it buyer-level or basket-level? Counting households and counting transactions answer different questions. Make sure the grain matches the decision, and that it stays the same from one chart to the next.

Two follow-ups are worth adding: what baseline the index is compared against, so popular items do not masquerade as relationships, and how much of the buyer's spend the underlying data actually sees across banners and channels. A cross-shop number built on one retailer's receipts cannot say anything about banner cross-shop, however confidently it is labeled.

Cross-shop is a useful word as long as everyone in the room agrees which of the five things it means that day. Name the shape, the window, and the grain, and the number becomes something a brand team can act on. To see the five shapes computed on your own categories, contact GSDSI for a sample.