Mathew·Blog

Part 5 of 6 · The one-person finance team

How do you run a budget when the goals move?

Every team justified its spend from zero, against a burn line the CEO drew. Then the goals moved, and the budget stopped working as a scorecard and started working as an agenda.

In this part

  • Every team built its budget from zero, starting with headcount: over 80% of monthly costs.
  • Each corporate card became a budget line, with its limit set to that line's budget, so routine approvals stopped coming to finance.
  • When the goals moved mid-quarter, the budget stopped working as a scorecard and started working as an agenda.

New to the series? It follows one finance lead rebuilding a startup's finance function with Claude. Start with Part 1.

In early August, the Product lead posted in the team's budget channel. The team's Claude bill was growing, and the lead wanted help managing it. A teammate suggested a second seat instead of paying for overage.

It was worth a shot, and it fit their pre-approved budget. I replied: "Love the creative solution. Works." That was the whole approval.

Three months earlier, the same request would have gone by Slack to the chief of staff. By my third week, I'd been looped into a few of these requests a day (Part 1).

What changed in between was a budget.

A budget is a decision made in advance, so the next hundred decisions don't need you. The most common way to build one is to roll last period's spend forward and grow it with headcount, revenue or customers.

But leadership wanted some measure of return. Here's your team's goal, here's the budget to hit it. Did you hit it, yes or no?

That's the promise of a zero-based budget: every line justified from scratch, against what it's meant to achieve. But the textbook process is a heavy lift. Here, eight teams were hiring and moving fast. We didn't have time, so instead I built off two principles: start every line from zero, and tie it to a goal.

Figure to come: The functional finance stack. Part 5 builds budgets, on the third rung.

1. Start from zero, and from headcount

In early July, the CEO and I agreed the frame for the second half of 2026: an overall quarterly target for each team, with spend budgets as caps. "But I want to end the year with at least [redacted figure] in cash reserves," he told me.

Salaries were over 80% of the company's monthly costs, so I asked for headcount needs first: get the hiring roadmap right, and the budget was most of the way to reliable before we discussed a single software license.

Everything else was built up from zero: marketing and events, localization, user testing, security certifications, recruiter costs, and so on. Each line carried a method, an input and a note, so anyone could see how its number was built. Over five days, each team lead worked with me to shape their resource request.

Figure to come: Redacted from the Q3 budget workbook. Lines marked * make up the team lead's discretionary budget.

Once the CEO had reviewed and approved the quarter's overall investment, each lead got a budget plus a discretionary pot. As I put it to one lead: "You can allocate it across your team as you see fit to meet your quarterly objectives."

2. Make the budget the control

All eight sheets rolled up into the one view the CEO needed: if every team spent to plan, how far under the burn cap would we stay, and how much cash would be left at year end?

Figure to come: Redacted from the Q3 budget workbook and the model's quarterly summary.

The policy fit on a page:

Then, in August, the corporate card system went live, and my message to every budget channel was: "Each card is one of your budget lines, and its limit = that line's budget."

The plan was to judge spend by what it bought. A monthly finance sync put budget against actuals for every team, and growth goals against actuals for the commercial teams.

After month one, each lead got a one-to-one review. And the right question started to come from the teams themselves. Weighing a trip to showcase the product to a client, one commercial lead asked: "How tangibly is this going to ladder up into our upsell goals?"

3. Measure against a moving target

Then in month two, the goals moved. A planned market expansion was canceled, customer success reorganized into new pods, and hiring shifted to a manager-led model.

To adapt, I gave every planned role an on/off switch so leadership could evaluate the impact of hiring decisions on runway in real time.

Figure to come: Redacted from the Q3 budget workbook (version of 5 Oct).

By the end of the quarter, the budget-versus-actuals retrospective needed five columns for every line: the original budget, the final one, the change, the actuals, and the budget-to-actuals variance.

By the time spend could be measured against a goal, the goal had moved, and "Did you hit it, yes or no?" no longer had a clean answer.

As a measure of ROI, the budget had lost most of its value. But as a discussion starter, it had never been more useful. It asked the whole company: are we investing in the right things?

Every switch flipped and every line moved put that question on the table, in the monthly sync and in each one-to-one.

How Claude helped

This was a largely human, stakeholder-driven process. Claude helped draft messages and check the model, but on work this high in judgment, I expect my AI copilot to inform, not decide.

What I'd do differently

The goals were always going to move for a company at this stage. Rather than set up the budget as an ROI evaluator, I should have built the on/off switches in from the very beginning so it could be an alignment tool between the CEO's target burn rate and a team lead's resource requirements.

Takeaway: When the goals move every month, a budget makes a poor scorecard, but a great agenda. Use it to ask whether you're investing in the right things.

Now that we had a budget, we needed a spend management system that could enforce it.