Budget controls how much you spend. Bidding controls how that budget competes in the auctions from the earlier lesson. They are separate settings and people routinely confuse them.
| Daily budget | Lifetime budget | |
|---|---|---|
| You set | An amount per day | A total for the whole campaign |
| The platform | May exceed on a good day, balancing over time | Paces spending across the period |
| Suits | Ongoing campaigns with no end date | Fixed-period campaigns — an event, a promotion |
One point that surprises people: a daily budget is an average, not a ceiling. Platforms commonly spend more on days where opportunity is better and less on others, balancing out over the period. A single day above the number is normal behaviour rather than a fault.
Manual bidding means you set the maximum you will pay. It gives direct control and requires attention, and it cannot react to conditions between your adjustments.
Automated bidding hands the decision to the platform, which adjusts bids for each individual auction based on how likely that person is to do what you want. It uses far more signals than you have access to, and it is generally the better choice once there is enough conversion data to learn from.
| Approach | Optimises for | Needs |
|---|---|---|
| Maximise clicks | The most traffic for the budget | Very little; a reasonable starting point |
| Maximise conversions | The most conversions for the budget | Working conversion tracking |
| Target cost per action | Conversions at roughly a cost you name | Enough conversion history to be stable |
| Target return on ad spend | Revenue relative to spend | Conversion values, not just conversion counts |
Automation amplifies whatever you measure
Every automated strategy aimed at conversions depends on conversion tracking being correct. If tracking is broken, or counts the wrong thing, automation will optimise confidently toward a goal that does not reflect your business — and it will do it faster and more thoroughly than a manual campaign would. Verify tracking before trusting automation.
When a campaign starts, or when you change it substantially, the platform needs data before it can optimise well. Results during this period are unrepresentative — often worse, sometimes misleadingly better.
The common mistake follows directly: seeing poor early numbers, changing the settings, which restarts learning, seeing poor numbers again, and changing again. A campaign kept permanently in that state never gets the chance to work.
Set a campaign up carefully, then leave it alone long enough to produce meaningful data before judging it. Resisting the urge to adjust daily is one of the more valuable disciplines in paid advertising, and one of the harder ones.
Work backwards from what a customer is worth rather than forwards from what you feel comfortable spending. If a customer is worth a certain amount to you, and roughly one in ten enquiries becomes a customer, and roughly one in twenty clicks becomes an enquiry, you can estimate what a click can be worth paying for.
Those ratios are unknown before you start — which is the argument for a small test budget whose purpose is to produce those numbers, not to generate profit. Once you know them, scaling is arithmetic instead of guesswork.