In April 2024, Foxtrot closed all 33 of its stores on the same day. Upscale convenience and café, beautiful curation, a loyal following in four cities, and the parent company was in bankruptcy a few weeks later. When co-founder Mike LaVitola brought a handful of locations back a year on, he described the fix as going back to "curation, a simplified model, and customer service," growing "at a more measured pace," and walking away from leases where rent had climbed 30 to 40 percent.
Notice what's not in that list. Nobody said the smoothies weren't good enough.
Meanwhile Erewhon has grown to eleven stores in Los Angeles and is opening in New York. Everyone has an opinion about why. Celebrities. TikTok. The $20 smoothie. Here's the number almost nobody talks about: when Erewhon took growth capital in 2019, it disclosed that prepared food was already more than 40 percent of sales. Think about what that means. A store where nearly half the revenue is made in-house, from perishable ingredients, on a daily production schedule, is not a grocery store with a café. It's a food manufacturer with a retail front. You cannot run that on vibes.
I've spent my career in the back of the store. Costco, where a case that comes in short is a case you don't pay for. Fairway in New York, as Director of Retail Technology and the person handed the labor, inventory and process projects nobody else wanted, where I watched a beloved brand go through bankruptcy twice because the back of the house couldn't carry the front. Quick commerce. A restaurant in Southampton, where I learned hospitality by paying for it. And then I built a premium grocer in New York from the ground up and ran it as COO.
All of it taught me one thing, and it's the thing this whole category gets wrong. Everything the customer sees is paid for by discipline the customer never sees. The $20 smoothie is the headline. The receiving dock is the story.
Everything the customer sees is paid for by discipline the customer never sees
Two ways to fail
Premium and experimental grocery is the same business seen from two sides. Premium is the position: better ingredients, tighter curation, real hospitality, priced accordingly. Experimental is the engine that keeps the position alive: the drops, the collabs, the twenty percent of the floor that's always in beta. A premium store that stops experimenting is just expensive. An experimental store without premium discipline is a pop-up that never figured out how to make payroll. And both exist for one reason. The customer isn't buying groceries. She's building a wellness lifestyle she can actually afford and live, and she wants a place that helps her build it and keeps giving her something new to add.
The demand is real and it's structural. So why do so many of these stores flame out?
In the stores I've run, consulted for and competed against, I've seen it happen for exactly two reasons. Either the operators never knew what the job actually required, so they built a beautiful store on top of nothing. Or they knew and stopped doing it, usually somewhere around month fourteen, usually because the founder got busy with store two or the one person who held the standard left.
Ignorance or decay. That's it. I have never seen a premium grocer fail because the product wasn't good enough or the store wasn't pretty enough. I've seen plenty fail with gorgeous stores and great product because nobody checked the truck at 5 a.m., nobody knew which department was bleeding, and nobody could say what was actually in the walk-in.
The stores that scale aren't smarter and they aren't luckier. They know what the disciplines are, they install them before opening, and they refuse to let them slide when it's inconvenient. The daily flash P&L by department, on the desk before the doors open. The weekly review where decisions come out of the room, not observations. Receiving as the first margin gate, every case checked, every credit taken that day. A planned exit for every perishable before the PO is cut. Pricing managed by department role on a monthly calendar, not a markup somebody set in year one. Labor scheduled to the traffic curve, not the manager's anxiety.
Theater in front. Warehouse in back. The customer should never feel the control environment. The staff should never be allowed to forget it.
The thing almost nobody in grocery does
Here's the one that surprises people. Most grocers don't actually know what's in the building. I mean that literally.
Perpetual inventory, a live item-level count the shelf actually agrees with, is standard in warehouse clubs, apparel and electronics. Grocery as an industry gave up on it decades ago. Thirty thousand SKUs, random-weight produce, DSD vendors stocking their own shelves, ingredients walking into a sandwich with no transaction anywhere. So the industry runs periodic inventory instead: a count every quarter or two, a margin estimate in between, and a shrink number that is really just the plug that makes the books balance. The best study on this looked at nearly 370,000 inventory records at a single retailer and found 65 percent of them were wrong. That was a general merchandiser. Picture a produce department.
Most premium operators inherit that assumption and run the store on counts, gut and a plug. That's a mistake, and it's a mistake for a structural reason they've been handed and never noticed. You're not running 30,000 SKUs. You're running 4,000. That's a Costco number, a Trader Joe's number. At 4,000 items, perpetual accuracy stops being a physics problem and becomes a management decision. The curation that makes the store premium is the exact same thing that makes it countable.
And without it, everything else is a guess. You can't order perishables to velocity if you don't know what's in the walk-in. You can't attribute shrink if shrink is a quarterly plug. You can't know which items earn their shelf space if cost of goods is a department average. You can't sell from the store online if the site promises what the shelf doesn't have. And you have no business opening a second store if the first one runs on feel.
It takes three things, in this order: defined processes, management that enforces them, and tools that make the first two cheap. Almost everyone who fails at it bought the tool first. The processes fit on a page. Receive what arrives. Count what's on hand. Adjust what's lost. Transfer what moves. Report what changes. Nobody creates an item at the receiving dock. Receipts stay pending until the office matches purchase order to receipt to invoice. Adjustments carry a reason code and an approval. Counts run on a calendar, not a mood. The management part is harder: someone owns it by name, accuracy sits on the daily flash next to sales, and leadership holds the line the Saturday before a holiday, because accuracy decays the first week anyone looks away.
The tools used to be the excuse. ERP territory, six figures, a controller to feed it. That excuse is gone. Your POS already records every sale at item level. What's missing is the layer on top that turns the data into a short list of exceptions a manager can act on before open: negative on-hands, stock with no sales, sales with no receipt, cost changes against agreed price, today's count list. I've built most of it myself over the years, and it costs a few hundred dollars a month. Less than one day of unrecorded shrink in a single perishable department.
Facts, not feelings
One rule sits over all of it. Decisions in this business get made on data and facts. Not on how the store feels. Not on what the founder is excited about this week. Not on which vendor took you to dinner.
I know that's the hardest thing to hold in a category that runs on feeling. The whole product is emotional. The founder is usually a person of taste whose instincts were right about the concept, and that is exactly why those instincts are dangerous once the doors open. The same gut that said "this neighborhood needs us" will say "this honey is beautiful, bring in six more from the same maker," and "we don't need to mark down the salads, people love them," and "I know Tuesdays are busy, I can feel it." Every one of those is a feeling dressed up as a decision, and every one has a number that could have settled it in thirty seconds.
So: instinct picks the question, data makes the call. Have the hunch. Then go look. If the number agrees, move fast. If it doesn't, the number wins, and you go find out why your gut was wrong, because that's where the learning is. I'm not asking anyone to take the taste out of the business. Taste is why the customer came. I'm asking that taste be tested before it's funded, and that when it fails the test, it loses. The stores that do this look no less magical from the front. They just stay open.
It's learnable
The good news in all of this is that "done right" is definable. It isn't magic, it isn't celebrity, and it isn't taste. It's a spine: the control cadence, the inventory discipline, the pricing governance, the labor model, the authority levels in your systems, the reporting that feeds all of it. It can be written down, installed before opening day, trained, and carried intact into the next store. I've done it as the operator, as the technology director inside a chain that lost its way, as a restaurant owner, and as the consultant brought in after the fact.
If you want to know whether a premium store is a business or a very expensive set, don't look at the Instagram. Ask to see yesterday's flash by department, the receiving log, and the notes from the last weekly review. If any of the three doesn't exist, you have your answer.
I'm writing the long version of this as a book, The Truth Behind Premium Grocery. In the meantime I've put the questions I ask on day one of any engagement into a short diagnostic, twenty yes-or-no questions that tell you where you stand. It's at exklusiveconsulting.com/diagnostic. If you're building one of these stores, or you've got one and month fourteen is coming, that's the place to start. The ones who call before they open spend a lot less than the ones who call after.