The question we hear most about paid advertising is "how much should we spend?" It is the wrong first question. The right one is "what is a customer worth to us, and what can we afford to pay to get one?" Answer that and the budget sets itself. Here is the framework we use to turn ad spend from a cost that gets cut into an engine that funds its own growth.
Step 1: Work out what a customer is worth
Take your average order or contract value, multiply by gross margin, and multiply by the number of times a typical customer buys over a year or the length of a contract. That is the gross profit a customer generates. Decide how much of it you are willing to spend to acquire them: many businesses settle around a third, which leaves room for delivery costs and profit.
Example: an average project worth $6,000 at 50% margin, with clients typically returning twice a year, generates $6,000 in gross profit a year. Spending a third of that, $2,000, to win a customer is comfortably worth it. That number is your maximum cost per acquisition, or target CPA.
Step 2: Turn the target CPA into a lead target
Ads rarely produce customers directly; they produce leads that sales or a checkout converts. If one in five qualified leads becomes a customer, a $2,000 target CPA means each qualified lead can cost up to $400. Track this number obsessively, because it is the one your campaigns are actually optimised against.
If you do not know your lead-to-customer rate, start with a conservative guess, then replace it with real data after the first month. Do not skip the step.
Step 3: Split the budget three ways
Whatever the total, divide it into three roles:
- Protect (about 60%). Campaigns that already hit the target CPA: branded search, retargeting, proven audiences. This is the money that reliably comes back.
- Scale (about 25%). The best-performing campaign from last month gets extra budget, increased in steps of 20 to 30% so the platforms and your CPA have time to settle.
- Test (about 15%). New channels, new creative, new audiences, each with a fixed cap and a decision date. Most tests fail; that is the point of capping them.
The percentages are a starting point. What matters is that every dollar has a role and a rule for moving to the next bucket.
Step 4: Give tests enough data before judging them
The most common mistake is switching a campaign off after three days and forty clicks. A test needs enough conversions to say something, typically thirty to fifty, before you decide. If the target is $400 per lead, budget roughly $12,000 to $20,000 for a meaningful test, or lower the bar by testing against a cheaper micro-conversion such as a pricing page view. Set the decision rule in advance: "if cost per lead is above $500 after 40 leads, stop."
Step 5: Watch the metrics that lead to revenue
Click-through rate and cost per click are diagnostics, not goals. The chain that matters is impressions, clicks, qualified leads, customers, revenue. Look for the weakest link:
- Low click-through rate: the creative or the audience is wrong.
- Good clicks, few leads: the landing page is not doing its job.
- Plenty of leads, few customers: targeting is too broad or qualification is missing.
- Customers but poor margin: you are winning the wrong kind of customer.
Fix the weakest link before adding budget. More money into a leaking funnel just leaks faster.
Step 6: Reinvest, do not just spend
When the Protect bucket comes in under target, move the surplus into Scale the following month rather than banking it. This is what makes ads "pay for themselves": a campaign that produces $2,000 customers for $1,400 is a machine you should feed. Agree the reinvestment rule with finance up front, so growth is not held hostage to a fixed annual budget written before anyone knew what worked.
A worked example
A services business starts with $10,000 a month. Target CPA is $2,000, target cost per qualified lead $400. Month one: $6,000 to branded search and retargeting (14 leads, $430 each), $2,500 scaling a search campaign that performed last quarter (5 leads, $500 each), $1,500 testing a paid social audience (2 leads, $750 each). Month two: the social test is cut, search creative is refreshed and its budget lifted by 25%, and retargeting is widened to a longer window. By month three, blended cost per lead is under $380 and the business has the data to justify $14,000 the following month, with confidence about where it will go.
Where to start
Write down your customer value, your target CPA and your lead-to-customer rate before you touch a campaign. If those three numbers exist, every other decision in paid media gets easier. If you would like help finding them, our paid and programmatic team runs a short audit that produces exactly that starting sheet.
