How Much Should You Actually Spend on Ads

Two bad answers dominate this question. A percentage of revenue, which is a rule of thumb borrowed from businesses unlike yours. And whatever is left after everything else, which is not a decision.

The right answer comes from arithmetic you can do in twenty minutes.

Start from what a client is worth

Not the first invoice. The whole relationship, and the margin rather than the revenue.

Average value of a first engagement, multiplied by how many times a typical client comes back, multiplied by your gross margin. If a client spends $8,000, returns twice more, and you make 40%, that relationship is worth about $9,600 in gross profit.

Then decide what share of that you are willing to spend to acquire one. A third is a common and healthy answer for a service business. That gives you roughly $3,200 per client.

Now work backwards. If one enquiry in four becomes a client, you can afford $800 per enquiry. That is your number, and it is entirely specific to you.

Then check the floor

The arithmetic above tells you what you can afford per enquiry. A second calculation tells you whether the budget is viable at all, and it is the one people skip.

You need somewhere around thirty to fifty conversions before a result stops being noise. At $800 an enquiry, that is $24,000 to $40,000 before you know whether the campaign works.

If that figure is more than you are prepared to commit, you do not have a budget problem. You have a channel problem, and paid advertising may be the wrong instrument for your business at its current size. That is a legitimate finding and it is much cheaper than discovering it over eighteen months.

The delay that ruins the calculation

6sense published its B2B Buyer Experience Report on 12 November 2025, based on nearly 4,000 buyer responses. Buyers now reach the point of first contact at roughly 61% of their journey, and they initiated 79% of engagements. Four out of five deals are won by the pre-contact favourite.

Their median purchase sits between $200,000 and $300,000, mostly technology and services, so the exact figures will not transfer to a trades business. The mechanism does.

It means most of the effect of an advert happens long before anybody fills in a form, and often before they would recognise your name. Somebody sees you in March, thinks nothing, and enquires in September having decided you were the obvious call. No attribution system will connect those, so the March spend looks wasted and the September enquiry looks free.

Budget set purely on immediate conversions systematically underfunds the part of the spend that actually created the shortlist position.

How to split it

• Most of it on people already looking. Search, and anyone who has visited your site. The cheapest work you will ever buy, and the first place to be fully funded before anything else gets money.

• A meaningful slice on people who will look later. This is the part that gets cut first and creates the shortlist position the research describes. Fund it deliberately or accept that you are only ever competing on the last click.

• A small, permanent slice on testing. Ten percent, always running, so you are never in the position of having one channel and no alternative when it stops working.

Where AI changes what you can measure

The reason budgets get set badly is that the honest numbers were unavailable. Cost per lead was easy to produce and nearly useless. Cost per closed client, by campaign, months later, required somebody to manually reconcile advertising data against the CRM, and almost nobody did it more than once.

AI made that reconciliation routine. Matching a closed deal back to the campaign that started it, months after the click, is exactly the kind of unglamorous data work that a system does continuously and a person does never.

Two things follow. You can budget on revenue rather than on leads, which is the only number that was ever worth optimising. And you get an early read on quality: a system reading the actual enquiries can tell you in week one that these are the right size of business asking the right questions, months before any revenue figure exists.

That AI read on quality is what lets you fund a campaign long enough to work instead of cutting it in week two out of anxiety.

The three numbers you probably do not have

Everything above depends on figures most businesses have never calculated, and the calculation is where the real value of this exercise sits.

How many times a client actually comes back. Owners consistently overestimate this. Pull five years of invoices and count properly, because the whole model scales off it.

Your genuine enquiry to client rate. Not the number people quote in meetings. Count every enquiry, including the ones nobody logged, then count the ones that became work.

Gross margin by type of work. If some services carry half the margin of others, an average conceals which clients you can afford to buy and which you cannot.

Do these three and you will frequently discover the budget question answers itself, because one service line can support four times the acquisition cost of another and nobody had noticed.

Review it twice a year, not monthly

The budget is derived from unit economics, and unit economics move slowly. Recalculating monthly produces noise and encourages the exact cutting behaviour that stops campaigns working.

Set it, fund it for two quarters, then recalculate with real numbers from those quarters. If the client value has risen, the affordable cost per enquiry rises with it, which is normally where the next increase in budget should come from.

What to do if the number is small

Plenty of businesses run this calculation and find they can afford $6,000 a month, which will not sustain three channels.

Then run one. One channel, funded to the point where it can produce an answer, beats three funded to the point where none of them can. The most common mistake at small budgets is spreading them, and it guarantees that nothing ever reaches the volume required to learn anything.

And spend the rest of the effort where money is not the constraint: responding faster, following up properly, and having something worth reading when somebody looks you up. Those cost time rather than budget, and at small scale they move the number more than the advertising does.

Sources

• 6sense, The B2B Buyer Experience Report 2025, announced 12 November 2025.

AI Optimize ties closed revenue back to the campaign that produced it, so the budget is set from what actually pays rather than from what is easy to count. That work sits under Paid Ads Management.