Two Zimbabwean grocery stores, sketched with no labels and handed to an AI with one question: what can you tell about these businesses? The architecture predicted which one is closing branches — before a single financial report was opened.
A floor plan is a profit-and-loss statement written in physical form. Where a store puts its space, its doors and its tills is a costed bet on who walks in, how long they stay, and what they buy.
The test was simple. Two grocery layouts, drawn roughly to scale, no labels, no brand, no context. Handed to an AI. One question: what can you tell about these businesses?
The conversation went further than expected. It read the demand curve off the doors. It found the load-balancing mechanism hidden along one wall. It predicted what would happen if one operator bought the other — and named the timeline. Then the published numbers landed in exactly the same place, from two completely different starting points.
Store A. Separate entry and exit force a one-way route. Produce takes the centre. A strip of prepared-food counters runs down the right wall. Dry goods are squeezed into five aisles in the middle. Tills sit small, near the exit.
Store A is built for a long visit. Separate entry and exit doors force a one-way route, so a shopper crosses most of the floor before reaching a till. A welcome zone — a picnic tree, flowers — sits at the door, the way Food Lover's and TM's better branches greet you. Produce gets prime central real estate. A butcher, a bakery, a deli buffet, a sushi counter, a juice bar, a burger station, biltong, cheese, wines and bulk spices ring the edges. Dry goods exist, but they are confined to five aisles in the middle. They are a category, not the category.
Store B is built for a fast one. One door in, the same door out — grab-and-go. Seven parallel aisles of dry goods run the length of the floor and define the whole proposition. Tills stretch across the entire back wall, which is what you build when you are betting on throughput. Candy sits on the till strip, pure impulse placement. Fresh is present but compressed: one vegetable block, two narrow frozen-meat aisles, one deli, one bakery strip, eggs. No wines, no sushi, no juice bar, no cheese counter.
| Signal in the drawing | Store A — Specialty | Store B — Conventional |
|---|---|---|
| Doors | Separate entry & exit, one-way route | Single shared door, in-and-out |
| Centre of floor | Fresh produce, oversized | Dry-goods aisles, seven of them |
| Tills | Small, near the exit | Full back wall, high throughput |
| Prepared food | Deli, sushi, juice bar, burgers, bakery | One deli, one bakery strip |
| Implied dwell time | 30–45 min | 10–15 min |
| Revenue occasions per day | Many — breakfast to dinner | Few — peak windows only |
Every row was inferred from the unlabelled sketch alone. None of it required knowing whose store it was.
Ask which store an informal trader can replicate, and the drawing answers immediately.
Store B is exposed on almost everything it sells. Mealie meal, rice, cooking oil, sugar, soap, washing powder, canned goods, tea — shelf-stable, no cold chain, no skilled handling, a tight basket of fast-movers. That is exactly what a tuckshop stocks two blocks closer to home, in USD cash, with no VAT and no corporate tax. The cost gap is structural, not temporary.
Store A is mostly out of reach. A tuckshop cannot run a butcher, a bakery, a sushi counter, a deli buffet or a juice bar. It cannot hold fresh produce at scale — the cold chain, the daily turnover, the waste management are beyond it. The five dry-goods aisles in Store A are the only part open to tuckshop pricing, and by the time a shopper reaches them they are already inside, already spending on higher-margin fresh.
This is why one store is reportedly full all day while the other fills only at peak. Store A holds breakfast, mid-morning, a genuine lunch trade off the deli, an afternoon trickle, and an evening prepared-meal rush. Many of those visits have nothing to do with a grocery shop — they are meal occasions. Store B has one occasion. The quick top-up trips that used to fill its off-peak hours — milk at 11, bread at 2 — are precisely what the tuckshop took.
The strip of counters down Store A's right wall looks like a set of high-margin extras bolted onto a grocery store. It is not. It is a load-balancing mechanism for the cost base.
A supermarket pays for labour, lighting, refrigeration and rent across a 12-hour day. If it only sells in three of those hours, fixed costs eat the margin and there is no way to staff a 12-aisle floor with a skeleton crew at 11am and a full one at 5pm. The deli, sushi, burgers, bakery and juice bar pull demand into the dead hours. They flatten the curve. They turn the building into a quasi-restaurant for the stretches when nobody is doing a grocery shop, and that continuous flow is what lets the whole cost base amortise.
Put the two together. If Store B's operator buys Store A, the acquirer's own metrics will read the asset as inefficient everywhere. The deli has worse food cost than a dry-goods aisle. The butcher has worse labour productivity than a packaged-meat shelf. The bulk spices turn slower than ketchup. The flowers and the picnic tree are pure overhead with no measurable return. Daily fresh deliveries cost more than twice-weekly. The sushi chef earns more than a till operator.
So the integration playbook runs in sequence: centralise procurement, which breaks the small fresh-producer relationships that cannot meet centralised terms; rationalise to fast-movers, which strips out the discovery; close or outsource the kitchen, which kills the lunch trade and the all-day footfall; swap skilled staff for general retail labour, which ends the butcher, baker and sushi counter; cut delivery frequency, which dulls the produce. Every decision is defensible on a spreadsheet. The sum of them hollows the store out within 12 to 18 months.
The deeper trap is that the business case mandates the destruction. Acquisitions get approved on synergy numbers, and synergy means cost integration. The document that justifies the deal to the board is the document that dismantles the asset.
All of the above came from two unlabelled drawings. Two recent Zimbabwean analyses, working from real data, arrived at the same conclusion from different directions.
An analysis at Money & Moves, by Tinashe Mukogo with data from Injecta Analytics, counted vehicle traffic in the Greendale catchment between 6:30am and 7pm. Of more than 8,700 vehicles, 50% — 4,340 — were heading to Honeydew Shopping Centre, where Food Lover's is the anchor. In a catchment with Spar, TM Pick n Pay and Bon Marché all competing for the same shopper, half the area's traffic flowed toward one store. That is not a footfall metric. It is gravity.
The reporting also supplied the M&A test case the drawing predicted. A piece in the Zimbabwe Independent made the case that single-site excellence is now beating networked scale: Food Lover's Greendale, owner-run by the Willcox family for over 40 years, holds a queue six days a week. Buried in the same piece — the two Food Lover's outlets in Avondale and Borrowdale, run under OK Zimbabwe's umbrella, have both closed permanently.
Same format. Same brand. Different operator. Opposite outcome. The fresh model did not fail in Avondale and Borrowdale. The operator running it did — exactly as the floor plan said it would, because the model only works when the food court does, and the food court only works when an owner is running it.
Fresh produce runs on a feedback loop. A store known for fresh has customers buying often, which means quick turnover, which means frequent restocking, which means nothing sits on the shelf, which means it can buy the freshest stock — which reinforces the reputation. Competitors are stuck in the mirror image: slower sales, longer shelf life, staler produce, slower sales again.
That is why integration is so dangerous. The moment turnover slows even slightly — centralised procurement, fewer deliveries, less skilled merchandising — the loop starts running backwards. Customer perception lags reality by a couple of months, so the operator sees no damage until the loop has already compounded. By the time it shows in revenue, the produce really is less fresh, because turnover really has slowed, because the customers really have noticed. There is no recovery path from inside the loop.
Kenya already ran this experiment. Nakumatt and Tuskys both collapsed under rapid expansion. Naivas and Quickmart, the chains that replaced them, grew slowly and protected quality at each site before opening the next. The continent's retail graveyard is full of chains that tried to be everywhere. The survivors earned each store.
Store B competes on unit price for commodity goods against traders who pay no VAT and no corporate tax. That is a 30–50% gap that cannot be closed. Any floor plan a tuckshop can fit inside is a floor plan under structural attack.
Store A is defended because the informal sector is structurally excluded from cold chain, skilled production and prepared food — not because it carries more lines. The right question for any retailer here is not "can I beat the tuckshop on price?" It is "is there a category they cannot enter at all?"
Sixty-nine stores is sixty-nine supplier networks, sixty-nine staffing problems, sixty-nine buildings on an unreliable grid, and sixty-nine points where one failure contaminates the brand. The model that wins is depth in one location, run by an owner, not breadth across many.
A floor plan is a P&L written in physical form. The only question that matters when you draw one: is it a shape a tuckshop can fit inside?
If yes, you are competing on the one axis you cannot win. If no, you have time. You can see the moat before you see the financials.
This analysis began as an unlabelled sketch and a conversation with an AI. It was checked against, and credits, two prior pieces that reached the same conclusion from data:
· Tinashe Mukogo, with data from Injecta Analytics — Money & Moves: "Why Food Lover's Thrives While Others…"
· The Zimbabwe Independent: "Small is the new big — future of retail in Zim"
Financial figures (OK Zimbabwe revenue, operating costs, rights offer; the TM Supermarkets writedown), the Avondale and Borrowdale closures, the 8,700+ vehicle count and 50% capture, the 30–50% tuckshop cost advantage, the 76% informal-economy share, and the Kenya precedent are drawn from those two pieces and the public company disclosures they cite. The floor plans are schematic reconstructions of two store types, drawn roughly to scale; they are illustrative, not surveyed measurements. Deep Dive Data covers Zimbabwe capital markets and financial products, and produced this analysis independently.