The usual follow-up dashboard shows the same table: messages sent, open rate, click rate. Three numbers that go up with volume, and prove nothing.
The business owner's question is not "how many messages went out". It is: did it bring in more than it cost? Answering it means linking a euro taken to the action that produced it. That is attribution, and it is trickier than it looks — not technically, but ethically.
The bias that makes every tool look good
Imagine a follow-up sent to 200 customers. In the three weeks that follow, 40 of them come back to the shop.
Did the follow-up produce 40 sales? No. Some of those 40 would have come back anyway: it was their rhythm. Attributing them to the follow-up is taking credit for the passing of time.
This is the fundamental bias of sales measurement, and better tools don't make it go away: rules set in advance, and kept, do. Without those rules, any follow-up tool looks impressive, because it mechanically captures activity that would have happened without it.
Three levels of proof, not to be mixed
There are three ways to link a sale to an action, and they are not worth the same.
The link carrying an identifier. The message contains a unique address — for booking, paying or signing up — that identifies the recipient and the action it came from. If the appointment is booked through that link, the connection is established. It is the only level at which there is no doubt.
The time window. The customer didn't click, but came back in the days after the message, and no other action claims the sale. It is plausible; it is not proven. This attribution is legitimate if — and only if — it is marked as indirect and counted separately.
The customer's word. The customer says they came because of the message. Rare, valuable, and unverifiable. Treat it as a clue, not a measurement.
The everyday sin is to add all three into one big total without telling them apart. The number you get is flattering and useless: nobody knows any more what it contains.
The window is set before, never after
An attribution window is a length of time: how long after the action can a sale still be linked to it?
It has to match the trade's cycle. Seven days for a service booked again every six weeks. Sixty days for a building quote. Six months for a training course whose funding requires an application.
The rule that matters is not the length itself, it is when you choose it. A window set before the campaign is a measurement. A window stretched afterwards, because the results were disappointing, is a justification. The difference is invisible in the final table — which is exactly what makes it dangerous.
What you should refuse to claim
Honest attribution always leaves money on the table. That is the price of its credibility.
- A sale outside the window is not attributed, even if you "can tell" it came from there.
- A sale claimed by two different actions goes to the closest one in time, not to both.
- A sale with no identifiable link stays in the "organic" column. It exists, it is welcome, and it belongs to no one.
This discipline has an unexpected effect: it makes the figures usable for decisions. When you know the total is more likely understated than inflated, you can use it to allocate a budget. When you know it is generous, you can't do anything with it.
What it changes in practice
Once attribution is in place, questions that were impossible before become simple:
- Which type of follow-up brings in the most per message sent?
- Which customer segment responds, and which one uses up the channel for nothing?
- How much does filling empty slots bring in, in euros, over twelve months?
- Has the tool's subscription paid for itself, and through which lever?
None of these questions can be answered with an open rate. All of them can with a "revenue attributed" column and a stable rule behind it.
The tracked link: the centrepiece
In practice, everything rests on a detail that looks harmless: the link.
A message that points to the website's general address makes the journey invisible. A message that points to a unique address — specific to that recipient and that action — makes everything else possible: the booking knows where it came from, so does the invoice that follows, and the total builds itself.
It is also what makes honesty effortless: whatever didn't go through the link isn't counted at the same level, without anyone having to remember to separate it.
The reverse: measuring isn't enough
One last trap, less obvious. Perfect attribution on a mechanism that doesn't run is useless. It is easy to end up with impeccable tracking for three follow-ups a quarter.
The useful order is this: get the mechanism running, then measure it, then improve it. Not the other way round. A dashboard has never called a customer back.
RevenueOS builds both together: every approved offer goes out with its identifying link, the window is set per type of action before sending, and the revenue comes back attributed — keeping what is proven separate from what is probable. How it works details how opportunities are valued upstream, and the industries page shows which levers apply to each type of business.