How much should we budget for paid media?
Build a paid media budget from the business outcome you need, realistic acquisition assumptions, and the amount you can afford to test. Separate advertising spend from creative, management, landing-page, tracking, and tool costs. There is no universal minimum budget or percentage of revenue that fits every agency or brand.
Key facts
- A testing budget is an investment with uncertain outcomes, not a guaranteed sales forecast.
- Qualified leads, sales, revenue, and contribution margin need separate assumptions.
- Spreading budget across too many tests can leave each with insufficient evidence.
Work back from customer economics
For lead generation, model the share of leads that become qualified opportunities and customers, then apply the contribution from a customer. For ecommerce, account for product costs, discounts, fulfillment, refunds, and other relevant variable costs. State any repeat-purchase assumptions explicitly.
Fund a defined learning question
Choose what the test needs to establish, the outcome you can observe, and a spending limit. Forecast a range of results rather than a single confident number. Use observed results to update the model rather than increasing spend just to meet an unsupported platform rule.
Include the supporting work
Budget for usable creative, a working destination, campaign setup, measurement, and ongoing review. A media-only plan can understate the cost of launching or maintaining a channel, particularly when assets or integrations do not exist yet.
Frequently asked questions
Is ad spend included in a management fee?
Only if the agreement explicitly includes it. Ask for a clear breakdown of media, services, tools, and any markups or additional charges.
Should we use a fixed percentage of revenue?
It can be a planning reference, but it does not establish what a campaign needs or what the business can acquire profitably.
Can we start with a small budget?
Yes, if the scope is narrow and the expected evidence is realistic. A small test may answer a limited question rather than prove long-term profitability.