A model with 62 tabs. Under two months of cash. The founder knew neither.

The situation

A founder-led producer was planning an expansion. The plan rested on a forecast model that had grown to sixty-two tabs over several years and that nobody in the company could audit end to end — including the person who had built it.

What the numbers showed

Rebuilding the model into a normalised base produced fourteen documented data-quality defects: broken formula references, scenario tabs that were empty, date headers stored as serial numbers, vendor and channel fields mixed in the same column, and duplicate fragmented sheets carrying different versions of the same series.

Two findings changed the conversation. Cost of goods had exceeded net revenue in the prior agricultural cycle — a timing effect of the crop calendar, but one nobody had isolated or explained. And trailing burn implied a cash runway under two months. There was also no approved budget for the year, which meant every variance being discussed internally was being measured against prior-year actuals rather than against a plan.

The decision

Expansion came off the table. Visibility came first.

What changed

The company moved from a model nobody trusted to a normalised base covering a decade of history and forecast, with a documented method for monthly variance and a runway figure stated explicitly rather than inferred. The expansion question could then be asked properly — with an answer that meant something.

What this engagement was not

Not an accounting or audit engagement, and not a restatement of the company’s books. The rebuild produced a management view for decision-making, sitting alongside the statutory records rather than replacing them.

Anonymised at client request. Figures are stated as ranges or ratios where absolute values remain commercially sensitive.