Most advertising budgets get set in one of three ways. By gut, because a number feels about right. By leftovers, whatever was in the account at the month’s end. Or by copying a figure someone heard at a conference or saw a competitor spending. Then the results come in uneven, busy one month and quiet the next, and nobody can quite say why.
We run paid advertising campaigns for businesses that live and die by their lead flow, and over about eight years and north of $30 million in ad spend, we have watched this play out from the inside. We have also asked a couple hundred owners the same question: how did you land on your current number? The honest answer, most of the time, is some version of “it seemed reasonable.”
What separates a budget that works from one that just burns cash is this: the good ones are built from the business’s own economics, and they come with a way to tell when enough is enough. Here is how we reason toward a starting range, and how to know when you have spent into the right zone.
Start from your economics, not a number you saw online
The most useful budget conversation does not start with the budget. It starts with what a customer is worth to you.
Say a typical customer is worth $1,000 over the life of working with you, and out of every ten leads, about one becomes a paying customer. (A lead is simply someone who raises their hand, by calling, submitting a form, or asking for a quote.) That is a 10 percent close rate, which means each lead is worth roughly $100 to you.
That $100 is your break-even cost per lead. Pay more than that to generate a lead and, on average, you lose money. Pay less and there is room for profit.
Run the same logic at whatever step your ads are pointed toward: phone calls, form fills, a booked appointment, an add-to-cart. Work out the most you can pay for that action and still break even. That is your break-even cost per action, or CPA, the most important number in the exercise. Just remember it is a ceiling, not a goal: come in under it so there is margin left over.
Turn that into a monthly budget the platforms can actually use
Knowing what one action is worth is half of it. Turning that into a monthly number is where most do-it-yourself budgets quietly go wrong.
The ad platforms run on machine learning, software that improves its guesses as it sees more results. Google and Meta decide who to show your ads to, and they get better at it the more conversions they can watch. (A conversion is just a completed action you told the platform to count, like a call or a form fill.) Starve the system of that data and it never really learns, so results stay erratic and your cost per action stays high.
So there is a minimum volume worth budgeting toward. On Google, the rule of thumb most of us work with is around 30 conversions in a rolling 30-day window before the automated bidding, which Google calls Smart Bidding, optimizes with real confidence. The quick way to a starting number: multiply your break-even CPA by that threshold. At a $100 CPA and Google’s 30-a-month bar, that is about $3,000 a month to give the system enough to work with.
Meta runs on a different clock. Each ad set, the group where you set your audience and budget, needs roughly 50 conversions per week, not per month, to exit what Meta calls the learning phase, the early stretch when the system is still working out who responds. Run the same math: 50 a week at a $100 CPA is about $5,000 a week, or north of $20,000 a month, to fully feed one ad set. That gap is why, on Meta, we point the campaign at the cheapest reliable action and keep the money in a few ad sets rather than spreading it across too many.
None of these are magic numbers. They are floors that clear the “enough data” bar, and they are projections you will refine the moment real results arrive. But a floor is a useful thing to know before you begin.
Of course, you can still make money without spending this number. You just won’t be fully utilizing the algorithms.
More money is not always better
It is tempting to assume that if some budget is good, more must be better. It is not so. Picture the efficiency of each dollar as a curve that rises, peaks, and then falls: a budget can run too cold, just right, or too hot.
Too cold is the problem above. Below the data threshold, campaigns stay stuck learning, results swing week to week, and your cost per action sits higher than it should. Add budget past that point and things improve, because the system gathers more data and optimizes better. In the accounts we run, this middle zone is where the math finally starts working in your favor.
But it does not last. Eventually you have reached most of the people actively looking for what you sell. To spend more, you have to show your ads to colder audiences, or show the same people more often, and you start bidding against yourself. Cost per action climbs, and past a point more money genuinely means worse performance.
So how do you find that ceiling before you waste money discovering it the hard way? A few signals:
- On Google, check the metric called Search Lost IS (budget), short for search impression share lost to budget. It is the share of available impressions you missed because your budget was too low. At 20 percent, more budget can capture demand that already exists. At 0 percent, you are already getting every budget-limited impression there is, and adding money will not buy more.
- Watch your marginal cost per action as you scale. When a budget increase makes your CPA jump without a real lift in volume, you have passed the peak.
- Lean on the platforms’ forecasting tools. Google offers bid and budget simulators that estimate how changes might play out (not available for every campaign, but worth a look). On Meta, watch your frequency, the average number of times each person sees your ad, alongside your CPM, the cost per thousand impressions. When both climb together, you are saturating your audience.
For the rigorous version, there is incrementality testing: hold back part of your audience as a control group and measure whether the extra spend produces sales you would not have gotten otherwise. It is the only sure way to know that added budget is truly additive.
So your sensible range lives between the economics floor from the last section and this saturation ceiling. That band is your answer.
Where we would actually tell you to start
A range is not a number. Here is how we would narrow it down, based on what tends to hold up.
Concentrate, do not sprinkle. A pattern we keep seeing is a modest budget split across two platforms and four campaigns, none of which ever gathers enough data to get good. Back one platform and one campaign long enough to clear the data threshold, then branch out once something works.
Give it a real runway. A budget needs time to exit the learning phase and then produce enough results to judge fairly, so we ask owners for at least a couple of months before drawing conclusions. Kill a campaign in week two and all you have bought is the learning, with none of the payback.
Match the budget to the job and the sales cycle. Awareness, lead generation, and direct sales need different volumes and different patience. Where months pass between first contact and a signed deal, build that lag into the timeline, or the early numbers will look worse than they are.
Budget the whole machine, not just the media. The real number is not only what you hand to Google or Meta. It is the media spend plus creative (the ads themselves), tools, and the cost of someone managing it. Counting the media alone is how people underfund the parts that make the media work.
Pick something you can sustain. A steady budget you can hold for six months beats a big burst you abandon after one, because consistency is what lets the system keep learning instead of restarting.
So our advice is usually the same: start in the lower-to-middle of your range, prove it out, and scale toward the top as the data earns it. Begin where you have room to be patient, not where you are counting on an immediate return.
It is a projection. Now go make it real.
Everything above is a projection. It rests on assumptions: what a customer is worth, how often a lead closes, how the platforms behave with your offer in your market. The first few weeks of real data are what actually matter, and they will revise some of those assumptions. That is not a flaw in the method. That is the method.
And that is the point. You now have a defensible starting number and a clear way to tell whether you are underspending, with demand left on the table, or overspending, past the point of diminishing returns. That beats a guess every time.
So treat it as a loop, not a one-time call. Project, launch, measure, adjust, then go again. A budget is a dial you tune as the numbers come in, not a figure you carve into the wall once and hope for the best.
Want us to build these projections for your business? Let’s talk.
