Burlington is opening dozens of stores this year in spaces Joann used to fill. Nearly 40% of the chain's 2026 store openings sit in former Joann locations. They trace back to last year's bankruptcy auction, where Burlington Coat Factory walked into Joann's Chapter 11 and walked out with 45 leases. Not 5. Not 15. Forty-five specific addresses across 22 states.
I geocoded all 45 and pulled the demographics inside a 16-minute drive of each one. The numbers came back wildly different.
The smallest sits in Tupelo, Mississippi: 31,829 people, $69,677 median household income, 58.9% white, 3.5% in agriculture employment. The largest sits in Buena Park, California: 643,711 people, $105,263 median household income, 28.8% Asian, 41.6% Hispanic. It's the same retail banner and the same 25,000-square-foot store, dropped into a market twenty times larger and built for a fundamentally different shopper.
I expected the 45 picks to cluster around a demographic band the model favors, a trade-area type the chain repeats. They don't. They land across the entire distribution, which means either Burlington's selection is much looser than I assumed, or the variable doing the work isn't demographics at all.
Here's the part that stopped me. Take the two extremes, Tupelo and Buena Park, 20× apart in population. The average shopper in each one spends within 2% of the other at retail every year. That single number, per-capita retail spend, barely moves across all 45 sites while everything else swings 20× and more. It's the thread this whole edition pulls on. Let me show the work, then come back to it.
How Burlington Got the 45
Joann filed Chapter 11 in January 2025, the second time in roughly 10 months. By March, all 800 stores were headed for going-out-of-business sales. The real estate went to auction that April under the Delaware bankruptcy court, Case 25-10068. The winning bidder on each lease earns a designation right: they can step into the existing lease terms or hand it back.
These 45 weren't a fire-sale giveaway. The leases moved through a competitive, court-run auction with value and off-price retailers and landlords all bidding, and Burlington's in-house team won the ones that matched their prototype on price and designation flexibility.
Burlington's most recent 10-K shows roughly 1,200 stores against a long-term target of 2,000 in the U.S., with a unit cadence of about 100 net new stores a year. The 45 conversions add about 4% to the footprint in a single batch. That's half a year of organic expansion, taken in one proceeding, on court-negotiated terms.
The store compatibility is the obvious part. Joann ran mostly junior-anchor stores averaging around 22,000 square feet (the fleet spanned roughly 7,500 to 52,000) next to grocery, dollar, and value apparel. That's the exact strip-center band Burlington has built around. The conversion still costs real money. Demo of Joann's interior fixtures and mezzanine, new fitting rooms, racetrack-floor layout, updated HVAC for higher occupancy load, apparel lighting grid: trade press and lease-comp data put a typical Burlington conversion of a junior-anchor store at $40 to $80 per square foot in tenant improvements, plus inventory. The court did not eliminate the build-out. It compressed the rent dislocation, the spread between Joann's in-place rent and what a Burlington signing today would clear.
The less intuitive part is where the 45 sites are.
Where the 45 Sites Actually Sit
I ran a 16-minute drive-time polygon around each address, then pulled the 2023 ACS and 2025 ESRI projections inside it. Here is what the spread looks like across the seven variables that usually do the most work in a site-selection model.
That picture is what convinced me this isn't a typology pick. Five of the 45 sit above 50% Hispanic; another five are more than 80% white. Five lost population over the last decennial, while three are booming Texas suburbs growing north of 3%. The median age in Lady Lake, Florida is 68.6, because the trade zone wraps around The Villages retirement community; in Flagstaff it's 28.6, because Northern Arizona University anchors the zone. These are not adjacent demographic worlds.
Joann's old positioning worked across that spread because the merchandise pitch was universal. Quilting, back-to-school, and holiday décor exist in every census tract. Burlington's pitch is universal in a different way: value-priced apparel and home goods at a $20 to $50 average ticket. It survives the same demographic spread, and that part I already believed.
The per-capita math told me why the spread doesn't break the unit economics.
The Per-Capita Number That Made This Click
Back to the number that stopped me, with the math under it now. Tupelo's 31,829 people each spend, on average, $11,220 a year at retail in the trade zone. That's the $357M of total retail spend divided by population. Buena Park's 643,711 people each spend $11,053. The two are within 2% of each other, on opposite ends of a 20× population gap.
Run that across all 45 trade zones and the band tightens, but it does not collapse. The full range is $7,663 (Hesperia CA) to $19,480 (East Walpole MA), with a median of $11,651. 27 of 45 fall between $9,400 and $13,500 per shopper. The absolute zone size varies 20×; the retail intensity per shopper varies only about 2.5×.
That is the variable Burlington's selection process is actually clearing. Not "is this trade zone large enough." Not "is this trade zone wealthy enough." Just: does the per-shopper spend cross the threshold the prototype needs to hit four-wall margin? For an off-price store at the kind of sales per square foot Burlington runs, that threshold is low enough that 45 wildly different zones all clear it.
Three things doing the work, in order:
- Bankruptcy-court rent dislocation. Joann's leases were signed in 2018-2023 at rents that, in most of these markets, are now meaningfully below today's signing rent. Burlington steps in at the old rate.
- A store that needs ~$50/sf in TI plus inventory versus a ground-up build that runs $200 to $400/sf. The capex saving is real.
- A merchandise category that doesn't lose to demographics. Apparel, home goods, and baby gear hold up across all 22 states.
Burlington's Q1 2026 release showed a comp of +6%, gross margin up 0.3 points to 44.1%, and total sales up 14%. They raised full-year guidance to roughly 115 net new stores, with the ~45 Joann conversions folded inside that number. Take Joann out of the picture and Burlington's 2026 expansion program is nearly 40% smaller. The conversions aren't a side project.
One caveat I should name. The 45 picks are a forced-distribution sample. They're whichever Joann sites Burlington bid hardest on, not 45 independent demographic experiments. If you ran the same exercise on Burlington's organic 90 stores opening this year, the demographic spread would almost certainly be tighter. The Joann 45 tells you the floor the format can clear when supply is dislocated. It does not tell you the format would have picked these zones organically.
What Burlington's Own Model Would Have Flagged
The forced-distribution caveat opens the obvious follow-up: what would a portfolio-aware model say about these picks? So I loaded Burlington's existing footprint and ran two reads. The first checked cannibalization: how much each conversion overlaps stores Burlington already operates. The second checked organic fit: whether a chain-specific market model would have surfaced these markets on its own.
Cannibalization is the rule, not the exception
Most of the 45 conversions land inside meaningful trade-zone overlap with an existing Burlington store. That checks out with the underlying placement logic: Burlington bid hardest on Joann sites that sat in retail corridors Burlington already validated, which means most of the picks have neighbors. The three sharpest cases on the cannibalization heat map look like this.
That is real cannibalization on the high end, and meaningful overlap across the broader 45. A site-selection committee running this portfolio through internal review would have flagged most of the picks.
A theory on why they took them anyway
When I pulled up Burlington's existing footprint, I overlaid a flat 1/3/5-mile radius on every store by market type: a mile in dense urban cores, three in suburban corridors, five in rural. What jumped out wasn't the circles, it was the spacing: the stores sit far enough apart that overlap stays low under exactly that radius assumption. Pure speculation, but their footprint looks sited to minimize overlap on a flat-radius logic.
If that's how Burlington's own real estate team thinks about spacing, then the cannibalization their internal model sees on these 45 picks is materially lower than what the foot-traffic-based analysis shows. The 44.69 percent Denton overlap in our heat map could read as a single-digit number in theirs because the two stores' flat-radius rings barely intersect even though the actual foot-traffic catchments largely do.
That leaves two readings. Either Burlington is comfortable with the cannibalization and bid for the sites anyway, or they're underwriting these picks on a circular-radius model that systematically under-counts overlap and may not see what the foot-traffic data does. Both readings produce the same operational consequence: the 45 conversions sit closer to existing Burlington stores than a foot-traffic-based committee would have approved. The Q3 same-store comp number on the existing stores nearest the conversions is the data point that resolves which reading is right.
What the market-fit read shows
The second check: how does Burlington's own portfolio profile rank these markets? The market-fit read scores every area in the country against the chain's existing footprint. Warm = high fit, cool blue = low fit. It's a read on which zones look most like the chain's best-performing stores.
It's one signal, not a verdict on the market. But across these three sites the read is consistent: the model sees these zones as low-fit relative to the rest of Burlington's footprint. The conversions came with that score and Burlington still bid for them.
Put the three reads together. Most of the 45 stack foot-traffic cannibalization against the existing portfolio, the chain-fit layer scores several of the markets low, and Burlington's own internal lens looks circular-radius rather than foot-traffic-based, which would systematically under-count overlap. Whatever combination of those signals the underwriting team saw on their screens, they bid for the sites. The bankruptcy-court rent dislocation was worth more than the picture would have suggested. Off-price unit economics absorb both the overlap tax and the market-fit penalty.
If you're underwriting expansion against your own portfolio this cycle, that's the pattern. Court-supervised supply is bidding hard enough to override (or invisibly absorb) cannibalization and market-fit signals at the same time, and the foot-traffic-based view is the one that surfaces what the radius-based view misses.
Where the Other 755 Joann Sites Went
Burlington took 45. The market took the rest in pieces. Burlington's was the largest single block any retailer claimed out of the Kroll-administered docket. A few other chains stepped in for small numbers of sites (Hobby Lobby and Boot Barn among them), while Michaels bought Joann's private-label IP rather than any stores. But no national tenant came forward at Burlington's scale.
The other 755 went in every direction at once: a handful to those chains, the bulk back to landlords, to local and regional tenants, or simply dark, concentrated in tertiary markets where no national bid and the owner has to find a single-tenant or sub-divide solution.
The rent-comp picture on the re-leased sites is not one number. Conversations with tenant-rep brokers tracking the wind-down suggest:
- Tertiary markets: re-leases are clearing at -15% to -25% below Joann's in-place rent. Landlords are accepting the haircut to avoid 12+ months dark.
- Secondary markets: roughly flat to -10%. Joann's leases were signed at fair-market in 2021-2022; the gap is mostly tenant-quality re-pricing.
- Sun Belt growth markets (the Austin/McKinney/Round Rock band on Burlington's list): re-leases are clearing at or slightly above Joann's rent, where demand from value and off-price tenants is strongest.
If you're an off-price competitor looking at the sites that went dark, the opening is real but uneven. The ones in your prototype's size range have already cleared lease friction through court oversight. The transaction work is done. Burlington moved fastest on the bidding because they had an in-house real estate team and the balance sheet to absorb 45 at once. Smaller chains can still pick from what's left over the next two quarters, but the bidding dynamic is geography-specific.
A Different Underwriting Threshold
If you're modeling any off-price expansion against your own portfolio, the read is straightforward. The variable that holds across the 45 is per-capita retail spend, not population size or median income. The 16-minute zones clear at a median of $11,651 per shopper, with 60% of sites between $9,400 and $13,500. Across all 45, per-shopper spend varies about 2.5× end to end, while the population and income spreads run 20× and 2.6× across the same sites.
A site-selection tool that searches for trade zones matching the cluster of top stores will not surface a list like Burlington's 45. It will favor median demographics and penalize outliers like Tupelo and Lady Lake and El Paso. The Joann sample is dispersed precisely because the supply was court-driven, not selection-driven, and the dispersion held up against the floor.
That's the deeper shift. The right question for a court-driven supply shock isn't which zones look like the top of your portfolio. It's which zones clear the per-format floor. The 45 Joann sites are a real-world read on what that distribution looks like when the supply side is dislocated.
If you want to run your own per-shopper-spend cut against the picks, I exported the full dataset: all 45 sites, lat/lon, 16-minute drive-time population, median HHI, median age, home value, retail spend, industry mix, and growth rate. Download the CSV (45 rows, 22 columns, no email gate).
What I'll Be Watching by Edition #39
Three threads I'd track over the next two weeks:
- Who backfills the craft category. Michaels picked up Joann's private-label IP but not its stores, and Hobby Lobby took only a handful of sites. The craft-retail vacuum Joann left is largely unfilled. Worth watching whether any chain moves to reconstitute it at scale, city by city.
- The Joann sites that went dark. First re-leases set the comp for the segment. If tertiary markets keep clearing well below Joann's in-place rent, expect another wave of landlord give-backs and value-tenant bargain-hunting. If Sun Belt growth markets keep clearing flat or up, expect the larger stores there to move fastest.
- Burlington's Q3 comp on the existing stores closest to the conversions. This is the data point that distinguishes between "they accepted the cannibalization" and "they didn't see it on their internal lens." If the nearest existing Denton, Mesquite, and South-Austin stores comp meaningfully below the chain's portfolio comp in Q3, the foot-traffic-based view was right and the conversion picks ate share their underwriting did not predict. If they comp in line, the off-price economics absorb overlap better than the foot-traffic model suggests.
If you're in an acquisition committee this quarter and your underwriting threshold is anchored to portfolio averages rather than format economics, the Burlington picks are the stress test. Run your own list through a per-capita-spend floor instead of a cluster match, and see how many of your past rejects come back into range.