Thornbury is a ninety-five-person digital agency in Bristol running around ninety client projects at any one time. Two people handle the entire finance cycle.
The situation
Agency economics are decided at the project level, and Thornbury could not see the project level until the month was closed and costed — which happened around the ninth working day, by which point a project that had gone wrong in week one had been going wrong for five weeks.
- A bookkeeping tool with no concept of a project.
- Personal cards for anything urgent, reconciled from receipts weeks later.
- A project-cost spreadsheet rebuilt monthly from the other two.
The spreadsheet was the real problem. It was accurate, it was maintained by one person, and it was the only place in the business where a client’s true cost existed. Everything the founders decided about pricing and staffing came out of a file that took four days a month to produce and was already two weeks out of date when it appeared.
The change
The project code moved onto the card. That is close to the whole change: freelancers, software, travel and client expenses all now carry the job they belong to at the moment they are spent, so project cost is a running total rather than a monthly reconstruction.
Three weeks, and the hardest part was agreeing the project-code list — which was a conversation the agency needed to have anyway and had been avoiding for two years.
The spreadsheet was excellent. That was the problem: it was so good nobody questioned that it took four days and arrived too late to use.
Tobias Lang · Founder, Thornbury
The outcome
The close finishes on the second working day. The costing spreadsheet no longer exists, and the four days a month that went into it went back to the two people who were producing it.
The change the founders talk about is not the close, though. It is that an overrunning project now surfaces in week two of the job rather than in the following month’s numbers — early enough that something can still be done about it.
