If You Can't Name What Produced Your Last Ten Customers

Here is a question that ends a lot of marketing conversations early. Where did your last ten customers come from?

Not roughly. Individually. Which channel, which page, which conversation. If the answer arrives as a shrug, an educated guess, or a gesture toward the analytics dashboard, then every budget decision made after this point is a guess wearing the costume of a decision.

This is the fifth stage of the revenue system, and it is the one businesses most often assume they have because they are collecting data. Collecting data and being able to trace revenue are different conditions, and the gap between them is where a great deal of money quietly goes.

Activity is not revenue

Most marketing dashboards report activity: sessions, impressions, reach, click-through, engagement, form fills. All of these are real, all of them are measurable, and none of them are money.

The problem is not that activity metrics are useless. It is that they are available, which makes them the default subject of every report and every conversation. A channel that produces a great deal of visible activity and no traceable revenue will keep its budget for years, because the thing it produces is the thing being measured.

Meanwhile the channel that produced four of your last ten customers through a route nobody instrumented gets described as "hard to measure" and funded accordingly.

The test is not whether you have numbers. It is whether any of your numbers are denominated in money and attached to a source.

Three things that break tracing in real businesses

Revenue closes offline. Someone finds you online, calls, and buys over the phone or in person. The transaction leaves no digital trace connected to its origin, so the channel that produced it appears to produce nothing. In service businesses and local businesses this is not an edge case. It is the majority path.

Consideration is long and multi-touch. A prospect finds an article, forgets about you, sees a post two months later, asks a colleague, searches your name directly, and converts. Last-click attribution credits the direct search. The article that started it gets nothing, and if you optimize toward what the report shows, you defund the thing that actually worked.

Nobody records the answer. Many businesses do ask how a customer heard about them. Fewer write it down anywhere structured, and fewer still can retrieve it later against a revenue figure. An answer given at intake and typed into a note field is not a measurement system.

The smallest version that works

The instinct at this point is to buy a bigger analytics stack. That is usually the wrong first move — it adds resolution to a system nobody is currently reading, and complexity that no one has time to maintain produces less usable truth, not more.

Start with something almost embarrassingly small.

Ask, and record it in one place. One required field at the point of sale or intake: how did you first hear about us. Free text is fine to start. The discipline is that it is required and that it lives somewhere you can sort, not in a notes field.

Attach a number to it. The source is only useful next to what the customer was worth. A source list without revenue attached tells you where volume comes from, which is a different and less important question than where money comes from.

Review it monthly, out loud. Ten minutes, the same question every time: what produced revenue this month, and what did we spend to get it. The value is less in the precision than in the fact that the question gets asked on a schedule by someone who can change the budget.

That is the entire system. It is imperfect, self-reported, and it will be more useful in the first quarter than the analytics platform you are already paying for, because someone will actually look at it.

What honest attribution cannot do

Worth saying plainly, because the industry is not always honest about this.

You will not achieve complete attribution. Some revenue arrives through routes that cannot be traced — a recommendation in a group chat, a conversation at an event, someone who remembered you from two years ago. Self-reported sources are approximations; people misremember, and they tend to name the last thing they touched.

The goal is not certainty. It is directional confidence sufficient to make better budget decisions than a coin flip. A system that is roughly right about your three largest sources beats a system that is precisely right about nothing, and it beats a sophisticated one that nobody opens.

Be suspicious of any report that resolves this into a single clean number. As we cover in what a Growth Analysis actually measures, a figure that feels precise and cannot be traced to something measured is a figure that was generated rather than found.

Why this stage comes before the others

There is an argument that measurement should come last, after the system is built. The opposite is true, and the reason is sequencing.

Without revenue visibility, you cannot tell whether any change you make to the four stages upstream worked. You will improve the site and see a number move, or not, and have no way to connect the two. You will test a channel and draw a conclusion from noise. Every subsequent decision inherits the uncertainty of the one before it.

With even a crude tracing system, the whole sequence becomes navigable. You can find your constraint, fix it, and know whether the fix produced money. That is the difference between running a marketing program and running an experiment you cannot read the results of.

Where to start

Pick up your customer list and try the exercise for real. Last ten customers, one line each, source and value. If you can complete it from records rather than memory, this stage is functioning and your constraint is somewhere else.

If you cannot, you have found the highest-leverage fix available, and it costs a form field and a monthly calendar entry rather than a budget. Acquisition work done without it is spending against a hypothesis — and as the order of operations suggests, scaling before you can read the result is how a channel gets blamed for a problem that was never located.

If you would rather have your system examined properly, a Growth Analysis reports what measurement is actually in place across each of the systems we build, what it can and cannot see, and what to fix first. Where nothing is found, that is not an empty section of the report. That is the finding.