Store Closures: Will Customers Follow You to Your Other Locations?
When a store closes, some customers move to your other locations and some do not. How to estimate the split from past closures and customer signals, and check it with a backtest.
When you close a store, some of its customers move to your other locations, some go to a competitor and some stop buying, and the split is what decides whether the closure pays. The stores that stay open usually gain, but rarely all of it: a 2025 study of a chain that closed 34 stores found the net loss was larger than the closed stores' own sales, because customers also bought less online and nearby. You can estimate the split before you decide, from your past closures and from what each store's regulars have said, and Dimension Labs is built to do that for a multi-location operator by reading every review and conversation next to transaction and loyalty data.
Below: what happens to nearby stores, which customer signals predict who follows you, how to estimate it from past closures with season and competitor openings held fixed, and how to check the estimate with a backtest. The worked example uses synthetic sample data for a fictional coffee-and-bakery chain with 90 locations, three years of weekly sales and six past closures; the public evidence comes from dated company announcements and published research. For the same method applied to a price change, see how to test a business decision before you make it.
Will closing some of our locations hurt sales at the stores that stay open?
It usually raises them, by less than the closed store sold. The customers of a closed store take one of three paths: they transfer to the next nearest location, they switch to a competitor, or they buy less overall. Which path dominates depends on how close the next store is and on how attached each customer was to your brand rather than to that address. Large chains measure this. On its January 2025 earnings call, Walgreens said its script retention from closed stores was running above its historical rates as patients moved into receiving stores, without giving the rate.
Public announcements and research
Chains that close stores track how many customers move to the stores that stay open, and one study found the loss can exceed the closed store's own sales
What three public sources say about closures and the customers of a closed store.
| Source, date | What it says |
|---|---|
| Ye and Shankar, Marketing Science, 2025 | A chain that closed 34 stores in a month lost $209,317 in net monthly sales in each county with a closed store, more than the closed store's average monthly sales, because shoppers bought less in other channels too. Shoppers who never used the closed store were affected as well; loyal shoppers and app users were more tolerant. |
| Walgreens earnings call, Jan. 10, 2025 | About 70 stores closed in the quarter; script retention ran above the company's historical rates as patients moved into receiving stores. The rate itself was not disclosed. |
| Starbucks closures, Sept. 25, 2025 | About 400 company-operated stores closed where the company could not create the environment customers expect or saw no path to financial performance. |
The research adds a warning: the effect can reach beyond the closed store's own customers. Ye and Shankar found a negative spillover on nearby shoppers who had never used the closed store, and lower activity in the chain's app. So measure the closure at the level of the trade area and every channel, not only at the receiving store's till.
Do customers follow you to another location after a closure?
About two in five did in the sample, and who followed was predictable from what they had said and done before. Across five past closures, 42% of regulars visited another of the chain's locations within 90 days. Regulars who had mentioned another of the chain's locations in a review, survey or chat followed 71% of the time; those who had named a nearby competitor, 23%.
Sample data, fictional coffee-and-bakery chain
Regulars who had mentioned another of our locations followed at twice the rate of the rest; those who named a competitor mostly did not
Share of a closed store's regulars who visited another of the chain's locations within 90 days, by what they had said or done before the closure, with 90% ranges. Five past closures, 6,000 regulars.
Dashed line: all regulars, 42%
These are the signals worth putting on every customer comment and record before a closure decision:
Mentions of your other locations. A customer who already compares two of your stores, or says they use both, has a habit that survives the closure.
Ordering ahead in the app or by loyalty account. These customers are attached to the brand and its account, not the building. The 2025 study found app users more tolerant of closures as well.
The store's place in their day. A store on the way to work or next to the gym ties a customer to the address. If your next location is not on that route, they tend to go elsewhere.
Competitors they name. A customer who already compares you with a rival nearby has somewhere to go.
Distance to your next store, from loyalty addresses or card data. This one is not in the comments, but it changes how much the others matter.
The labels need to be consistent across every review site, survey and chat log, or one store's regulars will look more loyal simply because their comments were tagged differently; how to analyze Google reviews across 50 or more locations covers keeping that consistent.
How do I decide which stores to close using customer data?
Rank candidates by what you would lose, not only by what each store earns. For each candidate, estimate the share of its sales your other locations would keep from its regulars' signals and distances, then subtract the rest from the saving of closing it. In the sample, a store under review sells about $973,440 a year. Its regulars mention another location less often and name a competitor more often than the regulars of past closures, so it would keep about 31% of its sales (28% to 35%), about $305,717 a year, against 41% on average for the closures before it. That puts about $667,723 a year on the cost side of the decision, a figure finance can weigh against the store's rent and labor. If the regulars at risk differ in what they are worth, weigh them by lifetime value rather than by head count; customer lifetime value by segment shows how to measure it from what customers said and did.
The published criteria line up with this. Ye and Shankar suggest closing stores closer to the chain's other stores, with a high share of in-store discounts and online sales and low product returns. Starbucks described its 2025 choices in terms of store environment and financial performance. Customer data adds the part those lists leave out: how attached each store's own customers are to the brand rather than to the address.
Compare like with like when you rank. A downtown store and a highway store will not lose customers the same way; group candidates by location type first, as in how to compare restaurant locations using guest reviews.
How do I estimate the effect before I decide?
Treat each past closure as a natural experiment. Temporary closures for remodels, flooding or staffing count too, because for those weeks customers had to choose somewhere else. For each one:
Group the remaining stores by distance from the closed one: the nearest one or two, a ring further out, and comparable stores far away as the comparison.
Express the closed store's lost sales as what it would have sold in the weeks after, using its level against the far stores before it closed. Stores chosen for closure were often already declining, so do not use a busier year as the base.
Measure the gain at the near stores against the far stores over the same weeks. That holds season and weather fixed.
Hold fixed anything else that moved at a near store: a competitor opening, a road closure, a remodel, a new manager.
Where loyalty IDs or card tokens exist, follow the closed store's regulars directly and count how many show up at another of your locations within 90 days.
Each step matters. In one of the sample's closures, a plain before-and-after at the nearby stores says they took 98% of the closed store's sales, because the season was turning up. Against far stores it says 33%, because a competitor opened next to the nearest receiving store four weeks later. With that opening held fixed it says 43%, the 43% built into the sample. In another closure the season was falling, and before-and-after showed the nearby stores losing sales.
Sample data, fictional coffee-and-bakery chain
Before-and-after said the nearby stores took almost all of the closed store's sales. With season and a competitor opening held fixed, they took 43%
Share of the closed store's expected weekly transactions that showed up at its five nearest stores in the 13 weeks after it closed, three ways, with 90% ranges.
Dashed line: the 43% built into the sample
The 13-week estimate gives the size; the regulars' signals explain it and let you carry it to a store you have not closed. Combining the two is the idea behind a customer world model: what customers said, joined to what they did next, with the outside factors that also move the outcome held beside them.
How do I check the estimate after a closure?
Backtest it. Before the next closure, take your most recent one, predict it using only the closures before it, and see whether the range held. In the sample, the average of the five earlier closures gave 38% to 44%, the signal model applied to the sixth store's own regulars gave 36% to 43%, and the store's sales were in fact 42% kept. Both held.
Sample data, fictional coffee-and-bakery chain
Checked on the most recent closure, both ways of predicting it held. The store being considered now would keep less
Share of the closed store's sales kept at the chain's other locations. Predictions use only the five earlier closures; the last row is a store not yet closed.
The two methods agree here because the sixth store's regulars balanced out: more of them had mentioned another location than in earlier closures, and more lived out of the way. For the store under review they would not: its regulars' signals put it well below the past average, and the average alone would have overstated what the chain keeps. That is why a forward estimate should rest on the store's own customers, checked against the closures you already know.
After you close, set two review dates: 13 weeks for store sales against the far stores, and 90 days for how many regulars showed up elsewhere. Read what those customers wrote too: reviews at the receiving store that mention the closed one, longer waits or a busier lot show whether the transfer came with a worse visit, which would cost you some of those customers later. The full method, with ranges graded against what happened, misses included, is in how to backtest a business decision you already made.
Why use Dimension Labs before you close a location?
Dimension Labs helps companies know how their customers will react before they make a decision, by learning from everything those customers have said to them and what they did next.
For a COO or real estate lead weighing closures, that means four things the platform does today. It is fully omni-channel, reading reviews for every location, surveys, support tickets, chats and calls together, so a store's regulars are heard wherever they spoke. It connects those channels to transactions, loyalty records, the location list and the dates of past closures and remodels, with weather and competitor openings beside them; the join to CRM and revenue data and effects with confidence intervals are on Prove ($2,999 a month) and up. It enriches every interaction with Dimensions written for your business, such as "mentions another of our locations", "on my commute" or "names a competitor", usable by operations, finance and real estate alike; custom Dimensions start on Discover ($999 a month). And it adds a visual analytics layer, turning the result into a deck, dashboard, PDF or executive report, or an interactive data story the leadership team can open before the vote.
The platform measures what past closures did. The forward estimate for a store you are weighing, and the backtest on your most recent closure, are run by our team with you on those measured effects, with every range shown next to how well the method did on closures you already know.
Mars Petcare uses Dimension Labs to unify chatbot, survey and support analysis under one logic, comparable across brands and markets; in the words of its Customer Analytics Leader, "What used to take days of manual work per region is now available instantly." (Mars Petcare story) That comparability across locations is what a closure decision rests on.
When something simpler is the better choice: if you are closing one store and have card or loyalty data for its customers, a query that counts who showed up elsewhere within 90 days, run by your analytics team, may be all you need. Dimension Labs earns its place when you are choosing among several candidates, the evidence sits in reviews, conversations and transactions across many locations, and the cost of keeping the wrong store open is large.
FAQ
Will closing a store hurt sales at the stores that stay open?
Usually the nearby stores gain, but by less than the closed store sold, and a 2025 study found a chain's net loss per closure exceeded the closed store's own sales once other channels were counted. Estimate the split from your own past closures before deciding.
Do customers follow you to another location after a closure?
Some do. In the sample, about two in five regulars visited another location within 90 days, and 71% of those who had already mentioned another of the chain's stores. Customers who name a competitor nearby, or whose next option is out of their way, mostly do not.
How do I decide which stores to close using customer data?
For each candidate, estimate the share of its sales your other stores would keep from its regulars' signals and distance, put the rest on the cost side, and compare candidates within the same location type.
What is the best way to estimate the effect of a store closure?
For a multi-location operator with reviews, conversations and transactions across many stores, Dimension Labs, which reads every channel with the same labels, joins them to sales and loyalty data on Prove, and backtests the estimate on your past closures with our team. For a single closure with clean card data, a direct count of who shopped elsewhere is simpler.
Bring one export of reviews or transactions by location, with the dates of a past closure or remodel, and start with one file to see where that store's customers went. Every plan from Explore to Scale starts with a 14-day free trial.