Your Closure Recapture Assumption Is Probably Too High

Closing an underperforming store is easier to justify when its customers have somewhere to go. If the nearest sister store keeps a third of the business, the closure pays back faster, and consolidating a market looks like an easy call.

That share is the recapture rate. It appears in a lot of closure pro formas as an assumption. It rarely appears with evidence behind it.

We wanted to know what the evidence says. Over the summer of 2026 we studied 20 years of Texas mixed-beverage receipts: monthly alcohol sales for every licensed bar and restaurant in the state, 3.78 million rows. For full-service chains like Chili's and Applebee's, alcohol is a steady share of sales (about 14% and 20% respectively), so it works as a stand-in for revenue across the chain.

What we measured

We selected 40 Texas closures of Chili's, Applebee's and Pappadeaux locations that had a same-brand store within 15 km (about 9 miles). 39 had usable data. For each closure, we compared the sister store's sales before and after with the change at similar stores elsewhere over the same months. That comparison removes seasonality and market-wide swings.

The two measured bars average the same closures in different ways. Weighting by the closed store's sales, so bigger closures count more, gives about 3%. Counting each local market equally gives about 14%. Only 23 of the 39 sister stores showed any gain at all, and the ranges are wide, so treat the exact figures as soft.

The direction is clearer. No estimate came close to 20%, let alone 30%. The ranges cannot rule out higher values, but the evidence points to recapture in the single digits to low teens.

That fits the academic work once you account for alternatives. A study of a fast-food chain found about 13% of a new unit's sales came from nearby sister stores (Pancras, Sriram and Kumar, 2012). A study of mass-merchandiser closures found over 90% recapture, but those shoppers had few other places to go (Shi, Inman and Gauri). Restaurant customers almost always have other places to go.

Why sister stores keep so little

When a store closes, its sales go three places: to your other stores, to competitors, or nowhere, because some trips simply stop. The lunch out becomes a sandwich at the desk. The Friday dinner doesn't happen.

The competition usually wins on numbers alone. In our Texas data, a typical closing chain restaurant had 0 to 3 sister stores nearby and between 18 and more than 30 restaurants in the same category. Customers spread across whatever is close and plentiful, and that is usually the competition.

We also modeled each closure with every restaurant a customer could reasonably drive to. Sister stores captured about 1% of sales on average after Applebee's closures and about 4% after Chili's closures. The highest was 17%, where a sister store sat very close with few rivals around it. These are model calculations, not measurements, but they point the same way as the measured results.

Your model's store list decides the answer

This is where closure pro formas most often go wrong. Many closure models include only the operator's own stores. A model like that has only one place to send customers, so it reports high recapture almost by construction.

To show how large the effect is, we built a hypothetical closure of 34 stores for a large multi-brand quick-service franchisee, using public store lists. Then we changed only which stores the model could see.

Same closures, same customers. The only change was the store list, and the share kept fell from about 77% to about 10%. Even the second version left out the major chicken, burger and coffee chains nearby, so the realistic figure is lower again.

The definition of competition matters as well. For the Texas closures, counting every bar and restaurant as competition gave average sister-store capture of 2.0%. Counting only casual-dining peers gave 6.4%. Neither definition is wrong, but a recapture figure means little until you know which list produced it.

Why your own sales can settle it

Better public data will not settle this for your network. The problem is noise. A single restaurant's monthly sales swing about 15% over six months for reasons unrelated to any nearby closure. In a dense market, a closed store's sales spread across so many nearby businesses that each one gains around 0.1%, roughly 250 times too small to detect.

The same limit applies to any vendor working from public sales or visit estimates. If someone claims validated per-store cannibalization accuracy, ask how noisy their data is and how many independent closures the claim rests on.

Your own weekly sales are a different story. They cover all revenue, not one category, and they are far less noisy. A franchisee with a dozen past closures and weekly sales for the nearby sister stores already holds enough evidence to check a recapture assumption instead of guessing.

The weekly-sales figures are projections until they are measured on real operator data. They assume three named sister stores per closure.

Five questions for your next closure meeting

1. Which stores does the model include? If the answer is "only ours," treat the recapture figure as an upper bound.
2. Can the model lose customers? If every customer must land at some store, the model overstates what your network keeps.
3. How many competitors sit near the closed store? Recapture shrinks as alternatives grow.
4. Is the recapture rate an assumption or a measurement? If it is an assumption, ask where it came from.
5. What did our past closures show? Compare sister-store sales before and after each closure, against similar stores elsewhere. About a dozen closures is enough to check a 20% claim.

What this research does and does not show

Every measured result here comes from alcohol-serving casual dining in Texas, using alcohol sales as a stand-in for revenue. Quick-service concepts were not measured, and their recapture may differ. The recapture estimates rest on 39 closures and carry wide ranges. The modeled capture figures are calculations, not measurements. None of this is an accuracy claim for any forecasting product, including ours.

Sources

• Texas Comptroller of Public Accounts, Mixed Beverage Gross Receipts, Texas Open Data Portal, January 2007 to June 2026.
• Pancras, J., Sriram, S. and Kumar, V. (2012). Empirical investigation of retail expansion and cannibalization in a dynamic environment. Management Science 58(11).
• Haans, H. and Gijsbrechts, E. (2010). Sales drops from closing shops: Assessing the impact of store outlet closures on retail chain revenue.
• Shi, Inman and Gauri. "Store Closed." Marketing Science Institute Working Paper 23-135.