What I learned mapping data silos in East African SMEs

It's easy to say 'SMEs have siloed data' as a starting pitch. It's much harder and much more useful to say precisely what that means in a way that leads to a specific product decision.

Researching this before writing any code for Dataloom meant looking closely at how a mid-sized SME's data actually lives day to day: a point-of-sale system for transactions, a spreadsheet for inventory, WhatsApp threads for supplier coordination, and paper logs for anything that didn't fit neatly elsewhere.

The failure mode that showed up repeatedly wasn't a lack of digital tools plenty of SMEs had several. It was that the same concept meant different things in different tools. 'Customer' in the POS system wasn't guaranteed to match 'customer' in the spreadsheet a sales rep kept. That's a semantic inconsistency problem, not a connectivity problem, and it changes the shape of the solution.

A tool that just pipes data from one system to another doesn't fix that it just moves the reconciliation problem downstream. What's actually needed is a normalization layer that can map inconsistent fields and formats onto a shared internal schema before anything gets reported on.

That distinction connectivity versus semantic reconciliation became the core of how Dataloom is positioned. It's a more specific, and more defensible, claim than 'we connect your data,' and it's one that came directly from spending time with the actual mess before designing around it.