"How much should we invest?" is the first question in every conversation and the last one to get an honest answer. The comfortable version is a round number chosen to resemble what competitors spend. The useful version is a calculation that starts at the end — the customers you need — and works back to monthly spend.
The backwards calculation, in four numbers
Four variables, three of which you already know without opening Google Ads:
- How many new customers you need per month. Not how many you would like: how many the target requires.
- Your close rate on leads. Out of 100 enquiries, how many become customers? In B2B lead generation it rarely clears 20-30%.
- Your landing page conversion rate. How many visitors leave details or buy. A well-built paid-traffic landing page often sits between 3% and 6%.
- The cost per click in your category. The only number you have to estimate from outside.
A worked example. You need 10 customers a month. You close 25% of leads → you need 40 leads. The landing page converts at 4% → you need 1,000 clicks. Your category CPC is £4 → £4,000 a month. If that figure is out of reach, the problem is not the budget: it is that one of the three upstream rates has to improve, or that this channel is wrong for that objective.
The strength of this calculation is not precision — it is approximate by design — but that it shows exactly where the economics break. A high CPC can be offset by a better conversion rate. A low close rate cannot be offset by anything: it gets fixed before the click.
What costs look like in 2026
Benchmarks are for framing an order of magnitude, not for predicting your account. Per WordStream's 2026 data across Google and Microsoft Ads, the all-industry average CPC was around $5.42 and cost per lead around $66.69. The spread is the interesting part:
| Industry | Indicative average CPC |
|---|---|
| Attorneys and legal services | ~$9.87 |
| Home and home improvement | ~$8.33 |
| Dentists and dental services | ~$8.00 |
| Travel | ~$2.14 |
| Restaurants and food | ~$2.05 |
| Arts and entertainment | ~$1.63 |
Two practical readings. First: an $8 click is not "expensive", it is the price of a category where one customer is worth thousands. Second: in the same study cost per lead fell for the first time in five years after a long run of increases — a sign of settling, not an invitation to plan against last year's costs.
The daily budget is not a daily cap
This is the most common misunderstanding, and it occasionally produces alarmed phone calls. The budget you set is an average, not a limit. Google can spend up to twice the average daily budget on a single day when the auction justifies it, balancing back over following days. The guarantee is monthly: spend does not exceed 30.4 times the average daily budget within a billing cycle.
Operationally: to get from a monthly figure to a daily one, divide by 30.4, not 30. And an isolated 180% Tuesday is not a system fault worth reporting.
Average daily budget or total budget
For time-boxed campaigns — a launch, an event, a promotion with an end date — campaign total budgets exist: you state the overall figure and the window, and the system distributes it. For everything else the average daily budget stays the correct setting, because it leaves Smart Bidding free to buy the good days.
The threshold below which budget is wasted
There is an amount of budget below which a campaign does not produce small results: it produces noise. Smart Bidding needs a steady flow of conversions to estimate probabilities; below that flow, every weekly swing is indistinguishable from chance.
The rule of thumb we use: if a campaign cannot generate at least ten or so conversions a week, it should not exist as a separate campaign. Merge it, or postpone it.
| Allocation mistake | What happens | Fix |
|---|---|---|
| Five campaigns at £10/day | None of them leaves learning | One £50 campaign, segmented by ad group |
| Identical budget across all markets | The worst market is over-funded | Allocate on customer value, not on parity |
| Generous budget on the brand campaign | You buy demand you already had | Keep it at defensive minimum, move the rest |
| Budget cut the moment CPA rises | Learning is interrupted halfway | Change the target, not the budget, absent an emergency |
The first month buys data, not customers
This belongs in the conversation before launch, not in the post-mortem. In the first weeks the system is estimating: which queries convert, which hours, which devices, which audiences. First-month cost per acquisition is almost always the worst of the year, and it belongs in the plan as a learning cost.
Two concrete consequences. First: a trial budget too small to complete learning is not a cautious test, it is a test that cannot answer the question. Second: since 17 August 2026 bid targets weigh more heavily on delivery in budget-limited campaigns — the change we analysed in Smart Bidding Exploration and promotion mode. A badly written target now caps spend far more literally than it did a year ago.
What to do before raising the budget
When a campaign works, raising the budget is the obvious lever and often the least profitable one. In order, check first:
- Impression share lost to budget. If it is low, more budget does not buy volume — it buys worse auctions.
- Waste in the search terms. Recovering 15% of misspent budget is worth exactly as much as a 15% budget increase, and costs nothing. That is the work described in governing negative keywords and search terms.
- Landing page conversion rate. Going from 3% to 4% cuts cost per lead by 25% without touching a penny of budget: see landing page experience.
- The quality of your conversion value. If the system optimises toward lead count rather than lead value, more budget means more mediocre leads.
Only once those four are in order does a budget increase convert into proportional growth. Before that, it does not.
If you want this calculation run on your numbers rather than on industry benchmarks, the free audit rebuilds exactly this chain: clicks required, real rates, sensible minimum budget. You can also see how we work and which services cover this ground.