That's the textbook answer. The more interesting question about ROAS, as with most numbers that make a meeting feel good, is who's doing the counting.

Start with who's doing the counting

Say you spend R50,000 on Meta in a month and R200,000 of sales are credited to those ads. Your ROAS is 4, the slide goes green and everybody gets back to their lunch.

Now ask where that R200,000 came from. Meta decided which sales its ads caused, using its own rules inside its own system, and Google did exactly the same with many of the same customers. Your agency, which may well be paid on performance, then reports the figure back to you with an arrow pointing reassuringly upwards. It's a bit like letting each of your children mark their own exams and then being pleasantly surprised by the results.

Nobody in that chain is lying, exactly. They all simply prefer the number high, and none of them is ever asked to show that the sale wouldn't have happened anyway. That last part, as it turns out, is the whole game.

How to build a beautiful ROAS by next month

If you wanted ROAS to look spectacular in four weeks, you wouldn't need much talent, only a willingness to spend money on people who were already on their way to you. Move budget into retargeting shoppers who have something sitting in their basket. Bid harder on your own brand name, so that people typing your company into Google can be charged for the privilege of finding it. Send existing customers a discount and tag the resulting sales as paid.

The number will rise every time, and the business will barely notice, because most of those people were coming back regardless. You've paid, in effect, to stand next to the till and accept the applause.

ROAS rewards being present at the moment of sale, while growth tends to come from being present much earlier, long before anyone has decided to buy, which is precisely where attribution is weakest and the ratio always looks worst. Run a team tightly to ROAS and it will, quite rationally, starve the work that creates demand in order to feed the work that harvests it. The ratio improves, the top line flattens, every dashboard in the building glows green, and somehow the year is still a disappointment.

A ratio, however handsome, has never paid a salary.

Why your ROAS dropped (and why that might be good news)

When the number falls, the explanation is usually more mundane than the panic in the room suggests.

• You're spending more. The first rand reaches people who were nearly ready to buy, and the ten-thousandth reaches people who'd never heard of you. Ratios nearly always fall as reach grows, which makes it less a problem than a law of nature.

• The creative has worn out. The same people have seen the same ad often enough to have stopped seeing it at all.

• Tracking has thinned. Consent banners and browser privacy settings mean platforms see fewer of the sales they caused, so your ads may be working exactly as well as before while reporting worse.

• The auction got busier. It's peak season, or a competitor has found a new budget and a sense of urgency.

If ROAS dropped because you're finally reaching people who didn't know you existed, that may be the best news in the report, even if it arrives dressed as the worst.

What to run the business on instead

ROAS is perfectly good for housekeeping, comparing ads within a campaign and catching the ones that have quietly broken. The business itself is better run on numbers that can't be flattered by moving money from one pocket to another.

Total revenue against total marketing spend. Some call this blended ROAS, others MER, and the name matters less than the principle: every sale counts, whoever claims it. Shifting budget into retargeting buys you nothing on paper here, which is rather the point.

Cost per sale. All the spend, divided across all the sales. If it falls while volume rises, the efficiency is real, and you can say so without anyone in finance reaching for a pen.

Incremental revenue. The sales you wouldn't have had otherwise. You find it by switching the ads off for a matched region or audience for a few weeks and measuring the gap. It's slower and far less glamorous than reading a dashboard, and it's the only test that tells you whether the money actually did anything.

What it looks like when the top line moves

Since 2021 we've run an embedded media and analytics team for a national retail group, across nine brands and 12 channels. In that time, digital revenue went from R289 million a year to R1.2 billion in FY2026, which is R911 million of new annual revenue, while cost per sale came down by 78%.

The FY2026 figure is the one worth reading twice: revenue up 29.2% on media spend up 16.2%. The top line outran the budget, and it did so in total revenue, a place where no platform gets to mark its own homework.

What to do on Monday

1. Work out break-even ROAS for each major product line. They'll differ more than you expect, and the single target you're using now will start to look a little arbitrary.

2. Add up every sale each platform claimed last month and set the total against what the business actually sold. Most people are surprised by how big it is.

3. Take your biggest retargeting or brand-search line and run a four-week holdout.

4. Put total revenue against total spend at the top of your report, and move ROAS further down, where it can still be useful without being in charge.

Where we stand

We'd rather show a client a lower ROAS on a bigger business than a gorgeous one on a business that isn't going anywhere.

Armada is a founder-run consultancy working with retail, financial services and B2B businesses in South Africa and the UK. Theo Bili, our co-founder and chief analyst, runs media and analytics himself, and every account and its data are set up in the client's name, so the learning stays with the business long after any agency, ours included, has moved on.

If the dashboards are green and the year still feels flat, that's usually where we start. Talk to us, or see the retail group work.