How Much to Spend on Ads

Most ad budgets are a guess dressed as a decision. A percentage of revenue, a number a competitor once mentioned, whatever was left after the other bills — and then it sits there, unchanged, for months. The trouble is that a fixed budget is almost never the right one. It is either too small, leaving demand unbought, or too large, quietly buying turnover at the cost of profit. The monthly report can't tell which, because it shows the average, and the average hides the one number that actually sets the right budget.

The question "how much to spend on ads" has a real answer. It just isn't a percentage. It's a level — found, held, and moved as the market moves.

Two Ways to Be Wrong

Spend too little and the paying demand goes unbought. There are customers who would have converted at a cost that still made a profit, and a small budget never reaches them. A competitor does instead.

Spend too much and the opposite happens. Past a certain point every extra dollar reaches a colder, more expensive prospect. The added spend still produces some sales, so the campaign looks alive, but each new dollar brings back less than it cost. The overall return can look perfectly healthy while the most recent increase is losing money.

On a monthly report the two mistakes look identical — a number that seems fine. That's why so many budgets stay wrong for so long.

The Last Dollar Is What Matters

The useful question is not "does the whole budget pay?" It's "does the last dollar still pay?"

Advertising has diminishing returns, and the reason is plain. The first dollars reach the most eager buyers, the ones already looking, and they convert cheaply. As the budget rises, the pool has to widen — to people who need more convincing, at a higher cost each. So the return on the first dollar and the return on the ten-thousandth are not the same number. The average blends them together and tells no one where the line is.

Total return can read 3× while the last budget increase came back at cents on the dollar. The decision lives at the margin, not the average. The right budget is the one where the last dollar added still returns more than a dollar — and not a step beyond.

Every Channel Has a Ceiling

Plot returns against spend and the shape is always the same: a threshold the auction won't let anything through below, then a steep stretch where each dollar pays well, then a plateau. On the plateau the reachable audience is already bought; more spend is pure overpay. The best budget sits at the edge of that plateau — the knee — where the channel is fully worked but not yet wasted.

And the edge moves. Competitors arrive and bid it up; they leave and it falls back. Seasons shift demand. A budget set correctly in March is wrong by June, not because the decision was bad but because the ground moved under it. Finding the right number once is not the job. Finding the edge and holding it while the market drifts is.

Recover, Grow, Harvest

The right budget isn't a single figure. It's a path with three stages, and knowing which stage a channel is in decides the next move.

Recover. Until a level has earned back what went into it, there is nothing to reinvest. A channel that hasn't paid its own cost yet isn't a growth engine — it's a test. The move is to hold, keep it small, and give it a cycle or two to show whether it pays at all before feeding it.

Grow. Once the level plainly pays, the whole return goes back in — not a cautious ten-percent bump, the full amount that came back. That is what compounds. A level returning two-for-one, reinvested each cycle, roughly doubles the budget every cycle until it reaches real numbers, and that is exactly the stage most budgets never reach because they were set once and forgotten. What keeps it safe isn't a speed limit; it's that only money that actually came back gets reinvested, so a cycle that got lucky simply returns less the next time and the budget follows it down. The one thing to watch is the last dollar — reinvest the whole return for as long as it still pays.

Harvest. When the last dollar stops paying, the ceiling has been reached. The move is to hold at that edge and take the surplus as profit — putting in less than comes back. Harvest is not the end. If a competitor leaves and the level starts paying again, the same channel drops back into grow. The switch runs both ways, continuously.

Finding the Level in Practice

Day-to-day numbers are mostly noise, so the reading happens in cycles — a week is a sensible default. Each cycle records two things and only two: what went into ads, and what came back. Everything else is decoration.

The budget changes in deliberate steps, holds for a cycle or two, and the last dollar is read again. A change too small to measure — a five-percent nudge — disappears into the noise and teaches nothing, so steps are kept large enough to see. And the reading is anchored to money returned, never to clicks or impressions. A cheap click that never becomes a sale is not efficiency; it's a cheaper way to lose. What counts is the sale and the margin behind it.

Done this way, the eternal question stops being a guess. The budget isn't a number chosen once and defended. It's the edge of the plateau, tracked cycle by cycle, climbed while it pays and harvested when it doesn't — which is the only honest answer to how much to spend on ads.

Common Questions

What percentage of revenue should go to ads?

No fixed percentage answers it. A rule like '10% of revenue' ignores the only thing that matters — whether the last dollar added still comes back. Two businesses at the same revenue can have very different right budgets because their channels saturate at different points.

How do I know it's time to increase the budget?

As soon as the money already put in has earned its cost back. From there the whole return goes back in each cycle, and the budget grows as fast as the returns do — a channel returning two-for-one roughly doubles each cycle — until the last dollar added stops paying.

When should I stop scaling?

When the last dollar added stops paying. Overall return can still look healthy while the marginal return has already dropped below break-even. That gap — strong average, weak margin — is where profit quietly leaks.

One sale on the first try — should I scale immediately?

Reinvest what actually came back — no more. If a sale returned two-for-one, putting that return straight back in is the fastest way up, and it's self-correcting: if the sale was luck, the next cycle returns less and less goes back in. The mistake isn't reinvesting early — it's betting more than the channel has actually returned.

Does this work for one-off sales, not subscriptions?

Yes — one-off sales are the simplest case. The margin from a sale is realized immediately, so there's no lifetime value to estimate. What matters is only what a cycle puts in and what comes back.