We designed uNobo Connect on reasoned assumptions, but assumptions nonetheless. With the network in real production, we have spent months looking at the operating system’s aggregated, anonymised telemetry. Some assumptions held. Others did not. Here is what the numbers say, without dressing anything up.
The business never closes
The starting assumption was that automated retail wins its edge outside business hours, in the night stretch where no one is there to open a shift. The network data confirms it with a concrete figure: 46.6% of monthly sales happen between 22h and 7h. Nearly half of the network’s revenue occurs in the exact window a staffed store is simply shut. This is not one store’s anecdote: it is the aggregate behaviour of the entire network. The model’s edge is not selling cheaper by day. It is selling at night too.
The friction was not where we were looking for it
We assumed the hard problem for an app-based business would be retention: getting the customer to come back. The data says otherwise. 98.5% of users who made a second purchase in the app keep buying, averaging 94.4 purchases per user, with 47.7% of those users exceeding 50 purchases. The average gap between purchases for a returning customer is 7.1 days, essentially weekly. Once a customer clears the hurdle of installing the app and making that first purchase, repeat behaviour sustains itself: it does not need pushing with discounts or aggressive notifications. The real problem is not retention. It is getting to that first purchase. The product effort we used to pour into loyalty has moved to cutting entry friction instead.
What looked impossible without staff, wasn’t
The most repeated objection against automated retail selling age-restricted product is that verifying age without an employee behind the counter is, in practice, unworkable. The network has processed over 3,100 biometric age verifications on aggregate, through integrations like Veriff or Bouncer Digital depending on the market. This is not a proof of concept: it is real, sustained volume, with no human intervention. What required a person in the room five years ago is a software layer today.
What these numbers do not prove
Being honest about this data means also saying what it does not prove. This is a network still under construction, with a single active Country Partner in a single market, Spain. These are not projections or a theoretical model: they are aggregated, anonymised data pulled from uNobo Connect’s real telemetry. But extrapolating Spanish consumer behaviour to any European market would be the same intellectual laziness we criticise in anyone selling automated retail on figures nobody can audit. Aggregated uNobo Connect network data as of 17/07/2026. Metrics anonymised. Auditable under NDA.
Why we’re telling you this
We could keep these numbers to ourselves. The reason we publish them is the same reason we designed uNobo Connect not to depend on one manufacturer or to charge a transaction commission: the more information whoever is weighing whether to lead the network in their country has, the better the decision they make, and the better the decision we make about who we partner with.
If you operate unattended retail in your market and these numbers say something about yours, let’s talk. Write to info@bonoboservices.com or use the form at bonoboservices.com.
Román Suárez, Founder of Bonobo Services OÜ