Case study 01 · Equipment analysis + website
O to go: about 30 convenience stores with a coffee counter, Baku
O to go runs roughly 30 convenience stores across Baku. Each one sells counter coffee next to the till. Footfall is high, queues are short and nobody is hired just to make coffee, so the margin on a cup has to cover the labour behind it.
- Client
- O to go · ~30 stores, Baku
- Machine chosen
- Franke A400
- Saving projected to December
- ≈ $300,000
- Volume modelled
- 250–300 cups/day per store
- Payback on the machine
- ≈ 13 months
The situation
When counter coffee was introduced across the stores, an entry-level Dr. Coffee bean-to-cup machine was installed so that nobody had to be paid to staff a coffee counter. At real trading volume the cup was weak and inconsistent, the machine stalled under load, and customers noticed. The company concluded that automation had failed and started pricing traditional barista espresso machines and the staff to run them.
The trap: both obvious answers were wrong
The machine on the counter was rated for roughly 30 cups a day. The counters were pulling 250–300, close to eight times its rated duty. Automation had not failed. A home-class machine had been put on commercial duty, and no equipment survives that. The barista line only looked like the quality answer because nobody had priced a barista's wages across three years.
- Duty rating. The cheapest machine is the one specified above your real volume, not below it.
- Labour. A loaded barista costs about AZN 700 a month per store. Across 30 stores that bill comes every month. A machine is paid for once.
- Downtime. A stalled machine at 8am is a lost sale, and a customer who buys coffee somewhere else tomorrow.
- Consistency. A super-automatic controls brew pressure, water temperature, extraction time and dose on every cup. A person holds those loosely.
- The floor, not the ceiling. A convenience brand does not need the best cup possible. It needs every customer to get the same cup.
The analysis
The cost model used Azerbaijani operating data instead of imported assumptions: loaded barista cost, the current business electricity tariff and the exchange rate. It was run across three store formats. Below is the medium format, 300 cups a day. Figures marked client came from O to go's own payroll and trading data. Figures marked est. are researched market estimates, and the client saw them labelled as estimates too.
| Line | Dr. Coffee incumbent |
Barista line machine + staff |
Franke A400 chosen |
|---|---|---|---|
| Rated duty | ~30 cups/day spec | Limited by staffing, not machine | Commercial, 300+ cups/day spec |
| Actual demand | 250–300 cups/day client | 250–300 cups/day client | 250–300 cups/day client |
| Dedicated staff | None, counter staff | 1 barista, ~AZN 700/mo loaded client | None, counter staff |
| Annual running cost | Stalling under load; unserviceable at this duty | ≈ $20,000/yr est. | ≈ $3,900/yr est. |
| Annual difference | n/a | + ≈ $16,100/yr, almost entirely labour | baseline |
| Labour cost per cup | n/a | materially higher | ≈ $0.14 on a cup sold at ≈ $2.20 |
The decision: what it is worth to the business
O to go selected the Franke A400. Taste did not decide it. Unit economics did: the same cup every time, at a labour cost of roughly 6% of its selling price, with no extra staff in any store.

“What we had experienced was not a failure of automation but a tiering error: a home-class machine placed on commercial duty. He argued against his own recommendation where the evidence required it. That is not always our experience with people selling us equipment.”













