The dashboard: decide on data, not gut feeling

Too much data, not enough decisions

Once the stack is in place, you're buried in numbers: every tool shows its statistics, charts, and tables. The trap is to look at everything and decide nothing. A good dashboard does the opposite: it isolates the three or four indicators that actually steer your business, and ignores the rest.

The question to ask isn't "what can I measure?" but "what decision will this number change?". An indicator that triggers no action is a vanity metric. Remove it from the dashboard.

The numbers that actually matter

A few universal indicators, to adapt to your model:

Available cash and runway. The first reflex, the vital sign from Chapter 6: how much I have, how many months I last.
Revenue (or MRR if recurring), and its growth. Not in absolute value only, but as a trend.
Real margin. Revenue minus direct costs. Selling more while losing on each sale is a classic trap; margin protects you from it.
Conversion rate of quotes/invoices and average payment delay (DSO). If your customers pay later and later, you'll see it here before cash tightens.
For a recurring business: churn and lifetime value (LTV), compared to acquisition cost (CAC).

Three to five numbers are enough. Better to watch five indicators every week than fifty once a year.

Building the dashboard, from simplest to most advanced

Several levels depending on your maturity:

Level 1 — the spreadsheet. Re-enter the key numbers by hand, once a week, in a Google Sheet. It's imperfect but forces you to know them. An AI (ChatGPT, Claude) can generate the template and formulas from a description of your activity.
Level 2 — your tools' native modules. Pennylane, Stripe, your business bank already display dashboards. Often enough to steer a small structure without building anything.
Level 3 — the light BI tool. When data comes from several sources, tools like Google Looker Studio (free) aggregate your sources and produce a visual, shareable dashboard. Metabase or Databox go further for those with a database.

Don't skip levels. Many entrepreneurs build a sophisticated dashboard they never consult. A spreadsheet looked at every Monday beats a forgotten Looker Studio.

AI as a financial analyst

Generative AI tools transform analysis for those uncomfortable with numbers. Concretely:

Export your data (invoices, transactions, MRR) as CSV from your tools.
Give it to an AI (ChatGPT, Claude) with a precise question: "What are my three rising expense lines this quarter?", "Which customers systematically pay late?", "At this rate, when does my cash drop below €5,000?"
Ask for a plain-language summary rather than a table. AI spots trends a hurried eye misses.

AI doesn't replace judgment or the accountant for regulatory matters, but it democratizes analysis: in two minutes you get a reading you wouldn't have produced alone. Beware, however, of sensitive data: anonymize what should be, and check the tool's privacy policy before feeding it your numbers.

The ritual that makes the difference

The tool is useless without a ritual. Effective steering comes down to one simple habit: a fixed, short, regular appointment with your numbers.

Every Monday (15 min): cash, unpaid invoices to chase, forecast to adjust.
Every month-end (30 min): revenue, margin, MRR/churn, comparison to the previous month.
Every quarter (1 hr): underlying trend, structural decisions (pricing, hiring, investment).

It's these appointments, more than the tool itself, that get decisions made in time. Discipline beats sophistication.

In practice

For this week: list the three to five numbers that, in your activity, trigger a decision. Build a minimal dashboard (spreadsheet or the native module of a tool you already have) that displays them. Set your weekly and monthly ritual in the calendar. And test the AI analysis once on an export of your data. You move from steering by gut feeling to steering by numbers — the mark of an entrepreneur who lasts.