We stopped reporting first order ROAS to clients, and it changed the budget conversation

Every number on the dashboard is green, and one thing is missing

A client's Meta dashboard is doing that thing where every number is green. First order ROAS on the new campaign looks superb. Cost per acquisition is falling, and the founder's in the group chat asking whether to double the budget. A brilliant day one number makes for a short meeting.

Nobody in that meeting is looking at the discount code attached to every one of those orders, a fifth off, applied automatically through the ad's landing page. The return is real. The margin behind it isn't what the slide implies, and nobody yet knows whether a single one of those customers will buy again. A big enough discount will always move product. That's arithmetic, not marketing skill.

First order ROAS measures one transaction and calls it a customer

First order ROAS divides one transaction by one cost. That's the whole calculation. It can't tell you whether the order was profitable once the discount, the shipping and the payment fees come out, and it says nothing about what happens next because there is no next yet. It's day one.

It becomes the headline anyway. One clean figure, and the one every ad platform puts in front of you first, so budget decisions get made on it. But it measures acquisition efficiency, not customer quality. Different jobs.

That's why we stopped leading client reports with it. A brand can look brilliant on first order return while quietly bleeding margin, because every one of those customers needed money off to say yes and most never come back at full price. ROAS still matters. It just needs a second number sitting beside it.

Ninety day LTV by channel shows who actually comes back

Ninety day LTV by acquisition channel tags every customer by where they first came from. Then it tracks everything they spend across their first ninety days, instead of their first transaction. It's the number our lifecycle work gets judged on. Run it by channel and campaign, and the picture stops flattering anybody.

We've watched this flip a channel ranking more than once. Paid social wins day one on the back of an always-on discount for cold traffic, while email converts subscribers off a smaller welcome offer and nurtures them through full price product. By day ninety the email cohort is spending more, because far fewer of them needed paying to come back.

Day one, paid social is the hero. Day ninety, it's the channel teaching customers to buy only on discount, while email builds people who come back at full price. None of that shows up on a ROAS dashboard.

Two offers can post the same day one return and build different customers

Think about two acquisition offers running side by side. A heavy welcome code pushed hard through paid social, and a smaller, targeted offer on one bundle. Both post a similar first order ROAS. Split the budget evenly on that evidence and you've flipped a coin.

Track ninety day value against each and they come apart fast. The heavy code pulls in people who were never going to pay full price, so the second order is discounted again or never happens at all. The smaller offer pulls in people who wanted that product, took the nudge, then bought again without being paid to.

Cutting the discount channel outright isn't automatically the answer. A discounted front door can still be the right one for a brand that needs volume or reviews quickly. What changes is the question you ask of it: how many customers worth keeping did it hand over. See how that reads across real accounts.

The gap between order one and order two is where email does the work

The acquisition channel decides who walks in the door, and on what terms. What happens between order one and order two is decided almost entirely by what lands in their inbox in the weeks after. Whether they ever pay full price. Whether they come back at all. That's why the channel that loses on day one keeps winning by day ninety.

A discounted first order isn't a loss if the weeks after it do their job. That's what the email systems we build are for: a post-purchase sequence that sells the product they just bought instead of the next code, and a second order nudge timed to when it runs out. Put money off in every subject line after that and you've taught them full price is a mistake other people make.

Build the report in an afternoon, not as a data project

None of this needs a data team. The version we run for clients is deliberately plain, because it has to survive being read every month by somebody who isn't an analyst. A report nobody opens is worse than no report. Four steps cover it.

1. Tag every order with its acquisition channel and campaign, using UTM parameters that survive checkout
2. Pull first order value and ninety day cumulative value into one view, split by that channel tag instead of blended
3. Flag which first orders used a discount code, so discount dependency gets its own column instead of hiding inside revenue
4. Put ROAS and ninety day LTV side by side in the same deck every month, so nobody can quote one without the other

The discount flag is the column that starts arguments, in a good way. It separates the revenue you earned from the revenue you bought. It's also the earliest warning you'll get that a channel is building a discount habit instead of a customer base.

What changes in the meeting when both numbers share a slide

The conversation changes because the question changes. Double the paid social budget turns into something harder: double the channel that teaches customers to expect a code forever, or fund the work that turns those same customers into full price repeat buyers. Put it that way and most founders answer it themselves.

It changes how wins get claimed, too. A cheap first order stops being the end of the story and becomes an opening position. Budget follows the channel that builds buyers, not the one with the prettiest screenshot.

One honest caveat, because we won't pretend this suits everybody. A brand that's three months old, or one selling a genuine one-off, shouldn't chase this yet. If ninety days barely covers one repeat cycle, the number is noise you can't act on. Get the volume and the purchase frequency first, then switch.