Revenue is an output. To improve it, you need a model of the activities that produce it. “More marketing” is too broad to tell a team where to work or whether the work helped.
- 01Keep the units consistent
- 02Locate the weak stage
- 03Choose one experiment
- 04Reconcile the model
Keep the units consistent
For a simple online store, start with visits multiplied by purchase conversion rate multiplied by average order value. For a service business, use enquiries, qualification rate, close rate and average contract value. Do not mix visitors with leads or bookings with completed sales.
A hypothetical store with 10,000 visits, a 2% purchase rate and a $100 average order produces $20,000 in sales before returns and costs. This is arithmetic, not a forecast. It shows which assumptions need evidence.
Locate the weak stage
A traffic problem and a conversion problem can produce the same disappointing sales total. They require different interventions. If qualified prospects reach the offer but cannot understand the next step, buying more traffic sends more people into that confusion.
Repeat purchase deserves its own model. Count customers who were eligible to buy again, how many returned and what they spent. Avoid treating repeat orders as new customers when assessing acquisition costs.
Choose one experiment
Record the current stage rate, the proposed intervention, its cost and the period you will observe. Change a checkout, a qualification question or a follow-up sequence with a specific reason. Keep the adjacent stages visible so an apparent improvement does not conceal a decline elsewhere.
Reconcile the model
Compare the model with actual completed transactions. Account for cancellations, refunds, taxes and payment timing where relevant. Revenue growth that creates less contribution or overwhelms delivery may not improve the business. The model is useful when it helps you choose a better decision, not when it gives a dashboard more numbers.