Store-level vs. market-level data: choosing your altitude

Altitude is a choice, and most retail data arguments are two people at different altitudes talking past each other. Store-level data shows what happened door by door. Market-level data shows the shape of the whole. Neither is better. Each answers a different class of question, and using the wrong one produces confident nonsense.

The metaphor is more than decoration. Every dataset is an observation from some height, and every question has a natural cruising level. The skill is matching them on purpose instead of defaulting to whichever dataset happens to be open.

What does market-level data do well?

Direction and position. Category trends, brand share, channel health, benchmarks against the market. Aggregation is the feature: one store's odd week disappears into the whole, so the signal that remains is the durable kind.

Market-level is also the honest altitude for public reporting, since aggregation protects the confidentiality of individual stores. A monthly channel read like the NRS Insights same-store sales report lives here by design: it describes the independent channel's trajectory without exposing any single door.

Aggregate reads also travel well across an organization. A leadership team tracking channel momentum doesn't need door-level detail; it needs a stable, comparable series that means the same thing every month. That's what this altitude is for.

What market-level data can't do is locate anything. It tells you the tide moved. It can't tell you which boats.

What does store-level data do well?

Execution. A distribution void, meaning a store that could stock your product but doesn't, is only visible at store level, because a void is a property of an individual door. Same for out-of-stocks, for the spread between your best and worst doors, and for whether the new placement actually sold in the stores that got it.

Store-level data also exposes what an average conceals. A flat market read can sit on top of strong doors climbing and weak doors sliding, and the response to those two situations is completely different. When an aggregate behaves strangely, the explanation almost always lives one level down.

Velocity spread deserves special mention. Two brands can post identical average velocity while one sells steadily everywhere and the other lives on a handful of exceptional doors. The second brand's average is fragile, and only the store-level distribution shows it.

How do you choose the right altitude?

Match the altitude to the question:

• Is the category growing? Market.
• Is my share improving? Market.
• Which doors don't carry me? Store.
• Did the promotion execute where it was sold in? Store.
• Is a soft month broad-based or concentrated? Both: market to size it, store to locate it.

Practicality belongs in the decision too. Store-level datasets are larger and take real analytic capacity to use well. Market-level reads arrive ready to interpret. Plenty of teams run a monthly market-level rhythm and drop to store level when a specific question demands it, which is a sensible division of labor.

The altitudes also discipline each other. A market read that contradicts your store-level view usually means your doors aren't typical of the channel, which is worth knowing. A store-level finding that doesn't move the market number is real but small, also worth knowing before anyone builds a strategy on it.

One caution about mixing altitudes mid-argument: settle the question first, then pick the level. Sliding between store anecdotes and market numbers inside one narrative is how teams talk themselves into whatever they already believed.