ROAS is the wrong number for a consumables brand

Here is a scenario that plays out on nearly every consumables account we audit.

The agency reports a 4.2 blended ROAS. Everyone is pleased. Spend gets increased. Three months later the ROAS is 4.4, spend is up 30%, revenue is up 12%, and the subscriber base has not moved. The account is more efficient and the business is not growing.

That is not a failure of execution. It is what happens when you optimise the metric.

Why ROAS drifts upward while the business stalls#

ROAS is revenue divided by spend. Nothing in that definition asks whether the revenue would have happened anyway.

Branded search is the clearest case. Someone types your brand name. They were going to buy. You bid on the term, win the click, pay for it, and book the revenue against ad spend. That campaign will show a ROAS of 8, 12, sometimes 20. It is almost pure attribution theatre, and it drags your blended number up beautifully.

An honest acquisition campaign — non-branded, category terms, competitor conquesting, upper-funnel video — will show a ROAS of 1.5 on a good day. If you are managing to a blended target, the algorithm inside your agency's head does the obvious thing: shift budget toward the campaigns that hit target. Which is to say, away from the ones that grow the business.

The arithmetic that should be driving the account#

Take an account we manage. An established Subscribe & Save subscriber is worth roughly six times what a one-time buyer is worth. That figure has risen about 25% since we audited the account, because subscribers are now staying longer and ordering more.

Six times. Which means the price you can rationally pay to acquire a subscriber is nothing like the price a single month of ROAS would suggest.

The trade is explicit: you give up ad efficiency in exchange for a growing base. That is the right trade for a consumable brand, and it is very hard to make if the number at the top of the weekly report is ROAS.

What replaces it#

Four numbers, in this order.

1. Active subscribers, and the direction of travel#

Not subscription revenue — subscriber count. Revenue can hold flat while the base erodes underneath it and the survivors order more. Count is the leading indicator.

2. True Portfolio CAC#

Acquisition cost calculated on the spend that actually wins new customers. Strip branded search. Strip conversions from people who have bought before. What is left is what you are really paying, and it is usually two to four times the blended figure.

Set a CAC ceiling from lifetime value with a margin of safety, then let bids move freely underneath it. Add a stop-loss if CAC approaches LTV — that is the one condition under which you genuinely should pull back.

3. A TACoS target, set from your margin#

Total advertising cost of sale — ad spend against all revenue, organic included. It is the number worth managing to, for two reasons. It cannot be flattered by shifting budget into branded search, because branded revenue sits in the denominator either way. And it moves in the right direction when organic rank improves, which a campaign-level ROAS reading is blind to.

Set the target from the economics rather than from a category benchmark: take COGS, fees, returns and promotions, work out what you actually keep on an order, and decide how much of that you are willing to spend on growth this year. That gives one number to hold the account to.

The point of doing the margin work up front is that you then stop having to look at it. A campaign can hit its ROAS target and still lose money once a 15% return rate and a coupon are counted — a TACoS ceiling derived from real margin closes that gap without putting a P&L in front of you every week.

4. ROAS, as a floor#

Keep it. A 1.0 floor stops genuinely indefensible spend. But a floor and a target behave completely differently: a floor stops the worst, a target pulls everything toward the middle.

What this looks like when it works#

On the account described above, the change was in what got reported, not in how hard anyone worked. Subscriber acquisition went to the top of the weekly report. ROAS moved down the page. Budget shifted out of upper-funnel video into lower-funnel manual campaigns, then AMC audiences, dayparting and retention targeting were layered on.

Nine months later the base is 59% larger year over year, Subscribe & Save revenue is up 48% year to date, and subscriptions account for about 40% of revenue against the 30 to 35% most consumable brands manage.

Ad efficiency, meanwhile, is worse than it was. On purpose.

How to have this conversation internally#

The obstacle is rarely analytical. It is that someone has already committed to a ROAS number in a budget meeting.

Three things make the switch easier:

Model it before you propose it. Take last quarter's actuals, strip branded search, recalculate CAC, and put it next to your best estimate of subscriber LTV. If the ratio is comfortable, you have your case in one slide.

Agree the measurement window up front. Two to three purchase cycles, written down, before spend moves. Otherwise month two produces a panic that undoes the whole thing.

Keep reporting ROAS. Do not remove it, just move it. Nobody has to lose an argument for this to work — the number simply stops being the one that drives decisions.