Januna went into its first real shop in January 2025. A year later it is in a handful, they are all still using it, and almost nothing on our original roadmap survived contact with a Saturday afternoon.
- sales completed
- 99.98%
- sales lost to an outage
- 0
- features requested
- 11
including offline and unverified
one power cut, one ISP failure
in twelve months, across every shop
Eleven requests, and what they had in common#
We expected feature requests. We got recovery requests. Nine of the eleven were some version of “something went wrong and I need to fix it in front of a customer” — a mistyped amount, a wrong item, a refund on a sale from yesterday, a receipt that has to be printed again.
| Asked for | What it really was |
|---|---|
| “Undo” | Recovering from a mistap without voiding a sale |
| “Reprint” | The printer jammed and the customer is waiting |
| “Change the price here” | The catalogue is wrong and the queue is not |
| “Yesterday's sales” | Reconciling the till against a bank statement |
Offline was the normal case, not the edge case#
We built offline support as resilience. It turned out to be the operating condition: shop wifi in a stone building with a metre of wall between the till and the router is not a network, it is a suggestion. Terminals spent whole afternoons partitioned and nobody noticed, which is the highest praise the feature ever got.
The most appreciated change of the year#
Not a feature. We stopped the receipt printer from guessing. It used to attempt clever layout for long item names; it now truncates predictably and prints the same shape every time. Three separate shops mentioned it unprompted, which is three more than mentioned anything we were proud of.
What we would build differently#
Would you still build your own sync?
Yes, but smaller — the model we landed on is the third. The first version tried to merge everything; the version that works gives every kind of data one owner and merges almost nothing. We could have started there.
Hardware?
We were too flexible too early. Supporting three printer families in year one cost more than the sales it unlocked. Pick two, support them properly.
Was the offline work worth it?
It is the product. Everything else is a catalogue with a keypad.