On Sundays, customers paid Nina Volkov to do manually what her software had promised to automate.
The model worked with clean sales data. Most shops had handwritten invoices, duplicate product names, and cash purchases recorded days late. After two years, only six used it unaided.
At a market stall, a grocer handed Nina a paper ledger and asked how many apricots to order. She spent an hour cleaning his records and gave him a number. He paid fifty euros and returned the next week.
“We raised money for software, not bookkeeping,” said her cofounder, Felix.
Their investors agreed. They offered a final tranche only if the company stopped services and pursued larger retailers with better data.
Nina visited one large prospect. Its data was clean because twelve employees maintained it. The retailer wanted a discount, custom integration, and proof from customers Nina did not have.
She declined the tranche. Felix left with the software and tried selling it separately under a licensing agreement. Nina kept the customer contracts and two support employees.
They began offering monthly inventory reviews, using scripts where useful and spreadsheets where not. The work did not scale neatly. Each new shop required conversation. Yet customers stayed because Nina’s team fixed records before predicting anything.
Two years later, the service employed eight people and earned less revenue than the original pitch deck had forecast for month eighteen. Felix’s software company had closed.
At the Sunday market, the grocer asked whether Nina was still a technology founder.
“On weekdays,” she said.
He handed her the ledger. Apricot sales had risen, but half the fruit spoiled during a heat wave.
Nina crossed out the model’s recommendation and wrote a smaller order. The most valuable part of her old company, she had learned, was knowing when its answer was wrong.
