Why your channel reports and your P&L disagree
Add up what every channel claims and the total comes to more than you made. This is the reporting we put in instead.
Every advertising platform ships with a metric designed to prove the platform is worth paying for. That is not a conspiracy. It follows from the fact that each system can only see the conversions it touched. Finance can see all of them, which is why the marketing report and the P&L so often describe different quarters.
If you add up the revenue your channels claim and the total is bigger than what you actually made, you do not have an attribution problem. You have a reporting model nobody can make a decision with.
How return on ad spend goes wrong
It fails in a specific and predictable way. The easiest way to improve it is to take credit for demand that already existed.
Narrow the targeting to people searching your brand name. Retarget visitors who were coming back anyway. Bid on your own trademark. Each of those will lift reported ROAS and do close to nothing for the business, because you have not created demand. You have intercepted it and paid a toll for the privilege.
That is how an account looks excellent for a year while contribution margin drops every quarter. Nobody is lying. The metric just cannot separate demand you caused from demand you caught.
The three numbers we install instead
Blended cost per customer. Total sales and marketing spend divided by total new customers for the period. It is crude, which is the point: it cannot be gamed by moving credit between channels because it does not care which channel did what. If this is rising while per-channel ROAS improves, part of your reporting is fiction.
Contribution margin. Revenue minus cost of goods, delivery, payment fees and acquisition. This answers whether the growth is worth having. A fair number of companies discover at this point that their best-selling product loses money at current acquisition costs, which is considerably more useful to know than any campaign insight.
Payback period. How many months it takes to earn back acquisition cost out of gross profit. This is the earliest reliable warning you get. Payback deteriorates months before revenue does, because it registers the moment you start buying lower-quality demand.
Working out what a channel actually caused
Once you report on blended numbers, the next question is how to split budget between channels without per-channel attribution to lean on.
The honest answer is testing. Geo holdouts, where you switch a channel off in matched regions and measure the gap, will tell you more in three weeks than a year of multi-touch modelling. Scale tests, where you push budget in one channel while the others hold flat, do the same job for headroom.
Both cost real money in forgone revenue. Both are cheaper than eighteen months of overfunding a channel that was harvesting demand something else created.
What usually changes
Three things come up again and again once clients switch to this view.
Brand search spend gets cut back or re-scoped. Retargeting budgets shrink and the money moves to prospecting. And email, SMS and retention finally get funded properly, because they rarely look impressive in a last-click report but raising repeat rate directly raises what you can afford to pay for a first customer.
None of this needs new software. It needs agreed definitions, a monthly reconciliation against finance, and the willingness to watch a number get temporarily worse because you have started measuring it honestly.
Where to start
Take one month that has already closed. Calculate blended cost per customer and contribution margin from finance data rather than platform exports. Compare that with what your channel reports claimed for the same period.
The size of the gap is your real starting position. In our experience it changes the conversation with the board more than any campaign result ever has.