What your accounting system already tells you about the funnel
- 4 days ago
- 3 min read
Updated: 2 days ago
Introduction
Businesses looking for funnel data usually start by considering new software. The better first step is to open the accounts, because they already contain a complete, dated, reconciled record of every sale for several years, kept accurately because a tax authority requires it.
That is a better dataset than most deliberate marketing measurement ever produces. It is complete rather than sampled, it does not depend on anybody remembering to log something, and it stretches back further than any tracking you might install today.
It only covers the bottom of the funnel. But the bottom of the funnel is where the money is, and almost nobody interrogates it.
1. What your accounting system already tells you starts with new versus repeat
The first split.
Sort customers by first invoice date and the two groups separate immediately. The proportion of revenue from repeat business is one of the most important numbers in the company. Very few owners can state it. Calculate it for the last three years and the trend is the real finding.
2. The interval between purchases
The second finding.
For any customer with two or more invoices, the gap between them is your natural repurchase cycle. The median across all of them tells you when to make contact. This is a genuinely actionable number. Contact slightly before the median rather than after.
3. The distribution of job values
The third.
Not the average but the spread, which shows whether your revenue depends on a handful of large jobs. That changes how you read every other measure. It takes one sort to see. A long tail of small jobs and three large ones is a different business from an even spread.
4. Seasonality, properly
The fourth.
Two or three years of monthly totals shows the real annual shape rather than the remembered one. Memory exaggerates recent bad months. The ledger does not. Plot the twelve months on one row and the pattern is obvious.
5. Customers who have stopped
The fifth and most useful.
Anybody whose last invoice is older than one and a half repurchase cycles has effectively left, and nobody noticed. This list is usually longer than expected. It is also the cheapest source of work available. These are people who already chose you once.
6. Concentration risk
The sixth.
What share of revenue comes from the largest three customers. This is a funnel question as much as a finance one, because it says how much new acquisition you actually need. It takes two minutes to calculate.
7. Time from quote to payment
The seventh.
Where quotes are recorded in the same system, the gap to the invoice is a complete elapsed-time measure for the back half of the funnel. Few systems make it obvious. Most will export the dates.
8. Which services sell together
The eighth.
Line-item data shows what customers buy alongside each other, which is a better guide to what to offer than any opinion. This is available without any additional collection. Look at it once a year.
9. Join it to the enquiry source
The step that completes the picture.
If the recorded source travels from enquiry to invoice, the accounts become a revenue-by-channel report. That one join makes everything else in this list more useful. It is worth the effort to set up properly.
Be careful about reading accounting data as marketing data without adjustment. Invoice dates are not decision dates, staged payments can double-count a customer, and refunds and write-offs need removing before any of these figures are trustworthy.
Conclusion
Open the accounts before buying anything to measure your funnel.
Split customers into new and repeat and calculate the share of revenue from each, find the median interval between repeat purchases, look at the spread of job values rather than the average, derive the real seasonal shape from two or three years of monthly totals, list the customers who have quietly stopped, check your concentration in the top three, and join the enquiry source to the invoice so the accounts become a revenue-by-channel report.
.png)



Comments