The owner asked a fair question: what are the emails actually making us? His platform had an answer on every campaign screen. Add the screens up for June, July and August and you get $1.5 million. It is a real figure. It is also about five times what the emails can honestly claim, and the reason is worth five minutes of your time, because your platform almost certainly counts the same way.
Where the extra million comes from.
Most email tools attribute revenue with a window. Somebody receives a send, and if they buy anything in the next thirty days, that purchase is credited to the send. On its own that is a reasonable rule. The trouble starts when you send more than once a month.
This client sends an email a week. A customer who came in on the 20th had received the email of the 1st, the 8th and the 15th, and every one of those campaigns claimed her repair order in full. Three screens, three credits, one order. Across one summer, 600 orders were credited to the emails, and 450 of them were claimed by more than one campaign. Add the screens together and you are not adding revenue. You are adding claims.
The platform is not lying. Each screen is true on its own terms. The lie arrives when a person, usually the one writing the monthly report, puts the screens in a column and totals it. That total has no relationship to the money in the till, and an owner who has seen his own bank statements will know it before you have finished the sentence.
The five-minute test.
You do not need our software to find out whether your own number is inflated. Pick last month and do this.
- Add up the revenue on every campaign screen for the month: that is the figure the platform would let you report.
- Count the emails a customer would have received in the thirty days before a purchase: with a weekly send it is four, with a fortnightly send it is two.
- Divide the first number by the second: that is a rough ceiling on what the emails can claim, and a quick way to see how inflated the screen total is. It is a sanity check, not a revenue figure. The figure comes from counting orders, below.
- If your platform lists the converting customers by name under each campaign, do it properly: count each purchase once, however many emails preceded it, and the sum takes care of itself.
If the sum lands anywhere near a number that feels like your whole month, it is not a coincidence. It is the same customers being counted once per email.
Counting it once.
We took the customer list behind every campaign, one row per person per purchase date, and kept each purchase exactly once. Then we gave the credit to the last email the customer received before they came in. That is called last-touch attribution, and it is not perfect, but it has the one property a report needs: the emails add up to the month, and the months add up to the year.
Counted that way, three months came to $300,000 from 500 repair orders, an average of $600 each. One month alone came to $110,000 from just under 200 orders, where the screens had said $340,000. The appointments-booked count fell the same way, from 230 to 70, because a booking had been claimed by every email in the window too.
The smaller number went to the owner. He preferred it. A figure you can defend in front of your own accountant beats a large one you cannot, and $300,000 in three months from an email a week is not a number anyone needs to inflate.
What it is still not.
Even counted once, this is influenced revenue, not caused revenue. Some of those customers would have come in with or without the email. The regulars always do. If you want to know what the emails caused, there are two honest ways to get closer, and neither needs new software.
The first is to set the counted-once figure beside the shop’s total revenue for the month. If eighteen percent of a month came from customers who had an email in the previous thirty days, that is a statement an owner can weigh. The second is the only test that measures cause: keep a random slice of the list out of one month’s emails, then compare the purchase rate of the people who got them against the people who did not. The gap, multiplied by the size of the list, is what the emails made. One month is enough for a first read, and it is the test we suggest to every client who has a list worth arguing about.
What to do on Monday.
None of this means the emails were not working. A customer who buys within thirty days of hearing from you is the reader you wanted. It means the report had been describing them four times over, and a report that flatters you is a report the owner stops reading.
- Find the setting: look up your platform’s attribution window and whether a purchase can be credited to more than one send. Most can, and most do not say so on the screen.
- Report per order, not per campaign: if the platform will give you the converting customers, count each order once and credit it to one email under one rule you write down, such as the last email before the purchase. If it will only give you campaign totals, the true figure is not in them. Get the order list from the till and match it yourself.
- Label the number: call it revenue within thirty days of an email, not revenue from email. The second claim needs a holdout to back it up, and you do not have one yet.
- Keep the screens for reconciliation: the platform’s own figure is still useful for spotting when a send misfires. It is just not the number that goes to the owner.