The Funnel Hill Creamery.

An earlier venture had taught me that ambition without unit economics rarely ends well. Starting a new brand meant approaching the numbers very differently this time.

Every decision began with the numbers: what a plate costs, what a location can yield, and what an outlet should return before opening the next one. Growth followed from the model working, rather than the other way around. That discipline is a big part of why we could raise $200K on the strength of the first outlet and grow to four.

With four outlets, 250+ SKUs and a customer expectation that every visit tastes the same, consistency needed to live in the system, not in any one person.

We standardised recipes, processes, supply chain and quality routines so that consistency didn't depend on individual heroics. Inventory of 250+ SKUs was managed across outlets with quality holding steady day to day. That system is what made expansion feasible in the first place.

Retail footfall has a natural ceiling, and a brand that lives in only one channel is more exposed than it looks, something 2020 made very clear.

We opened HORECA channels (hotels, restaurants, catering) that added roughly 20% to revenue on top of retail. When COVID closed dining rooms, a hygiene-first delivery campaign on Instagram lifted engagement by 60% and delivery sales by 30%, enough to keep the brand steady through a very difficult period.

Hospitality lost staff everywhere during COVID, and losing the team would have meant losing the experience customers came back for.

We hired and led 70 people across front and back of house. An employee incentive programme helped us retain 93% of the team through the crisis, and customer satisfaction moved from 80% to 92% in the same period. The people who stayed were, in effect, the experience.

With four outlets, 250+ SKUs and thin F&B margins, gut feel stopped scaling. We needed to know what was selling, what would sell next, which customers were coming back, and what each product should cost, before committing cash to stock and expansion.

I brought the data habits from my Cognizant years into the business. We analysed our finances, supply chain and procurement to see where the money was actually going. We also built machine learning models to forecast sales and, using delivery and loyalty data, to predict customer churn. Product-level analysis showed us the top sellers and the trends behind them. Predictive modelling of customer behaviour fed straight into pricing, so every SKU was priced to be fair for the customer while still earning its margin. It is also what pushed me to retrain formally in data and AI afterwards.