Financial modelling in a team means someone else catches your errors. Working alone means building processes that substitute for that review.
The first structural mistake
I started by modelling outputs before inputs. It felt productive to see revenue projections forming. When I later realised my cost assumptions were not linked to drivers, I had to rebuild significant sections. The lesson was to define all input assumptions on a single sheet first, even if it feels abstract at the start.
Tools that helped most
Google Sheets over Excel for one specific reason: version history is automatic and accessible without any setup. After losing work twice to corrupted Excel files, the cloud-based backup became a non-negotiable requirement.
For sensitivity analysis, a simple data table in Excel remains faster than any dedicated tool for the scenarios I typically needed to run.
Checking your own work without a second pair of eyes
One method that holds up: model the same output two different ways and compare. If your revenue projection using a top-down market share approach matches your bottom-up unit sales estimate within a reasonable range, the logic is likely consistent.
Deliberate redundancy in a model is not inefficiency. It is the closest substitute for a reviewer when you are working alone in a quiet room.