Budget, bidding
and the weekly cycle

Use automated bidding once there is enough conversion data to feed it, which is the case for most accounts. Manual bidding still has a place for very new or very small accounts, but it rarely beats the platforms’ systems at scale.

The real work then moves to what the automation is optimising towards. If it is being fed a form fill that produces no revenue, it will get very efficient at generating worthless leads. So we define the conversion carefully: a qualified lead rather than any form, a purchase with its value rather than a page view. Where possible we import outcomes from your CRM, so the bidding learns which clicks became customers. Setting the right conversion is the job.

Our management fee is a fixed monthly amount based on the scope of the account, not a percentage of your media spend. The fee is agreed in writing before work starts and covers the weekly cycle described on this page.

We avoid the percentage model on purpose. It rewards the agency for increasing your budget, which is the wrong incentive on both sides of the table: you want spend to rise only when it pays, and so should we. A fixed fee also makes it easy to cut a channel that is not working without the agency losing income. If the scope grows, for example a new market or another platform, we discuss a revised fee with you first rather than adjusting it quietly.

Split your budget according to where demand comes from: Google for people already searching, Meta for creating demand among people who are not. There is no universal ratio, and a fixed split agreed once a year is usually wrong by spring.

As a starting point, a business with strong existing search demand often puts most of the budget into Google, while a new product nobody searches for yet leans towards Meta. From there we move money on evidence. Each week we compare cost per qualified lead or return on spend across channels, adjusted for the fact that Meta often assists sales that Google gets credit for. Budget shifts are gradual, so each channel has time to adjust.

Weekly optimisation is a fixed checklist we work through for every account, so nothing depends on someone remembering to look. It takes the same shape each week and is recorded in a change log you can read.

• Search terms reviewed and new negative keywords added.
• Placements and audiences checked for waste.
• Spend pacing compared with the monthly plan.
• Creative performance and frequency checked for early signs of fatigue.
• Bids or targets adjusted where results have moved.
• Budget moved between campaigns according to return.

Big structural changes are not made every week, because platforms need time to learn after each one. The weekly pass is about steady correction, while larger decisions are taken monthly with you.

A realistic target cost per lead is one your margins can carry, and it has to be worked out from your own numbers rather than an industry average. The starting question is what a new customer is worth to you over a reasonable period.

From there it is arithmetic: if one in five leads becomes a customer and a customer brings a known profit, you can calculate the most you can afford per lead and still make money. We set the bidding target somewhat below that ceiling, then check it against what the platform is actually delivering in your category. If the market price is higher than your ceiling, the answer is usually a better offer or landing page, not a lower bid.

Budgets run out early when daily caps are set too high for the month or a campaign suddenly gets cheap, broad traffic; they go unspent when targets are too strict, audiences are too narrow or ads are stuck in review. Both are pacing problems, and both are fixable.

Platforms can also spend above the daily budget on busy days and below it on quiet ones, balancing over the month. We track spend against a monthly plan each week, so a month does not end with a third of the budget unused or gone by the twentieth. Where a campaign cannot spend at a sensible cost, we would rather move that money to another campaign than loosen targets and buy poor traffic just to use it.