Small budgets force good discipline. Know your unit economics, concentrate on one channel at a time, and split spend 70/20/10 between proven campaigns, tests and retargeting.
Small budgets force good discipline. The businesses we have seen waste the most on paid advertising are almost never the ones with small budgets. They are the ones that scaled too fast, before they understood their unit economics.
The way to run paid campaigns that return money when you cannot afford expensive mistakes is to know your numbers first, concentrate spend on one channel at a time, structure campaigns so the algorithms have enough data, and measure something harder to game than ROAS. Each of those steps is below.
Which numbers do you need before spending a dollar?
The most common budget-killing mistake in performance marketing is running campaigns before you know your unit economics. Four figures matter:
- Customer acquisition cost (CAC): what can you spend to acquire one customer and still be profitable?
- Average order value (AOV): how much does a typical customer spend in a single transaction?
- Lifetime value (LTV): how much revenue will that customer generate over their relationship with you?
- Gross margin: after product cost, what percentage of revenue is left to cover marketing and operations?
If your AOV is $50 and your gross margin is 40%, you have $20 to work with after product costs. Running Google Shopping at a 10x ROAS target implies a $5 acquisition cost. That is achievable in some niches and impossible in others. Know your numbers first.
Which channel comes first on a small budget?
When budget is limited, channel concentration beats diversification. The sequence that works for most e-commerce businesses runs in this order.
Start with Google Shopping. The intent is commercial and the demand already exists: you are capturing buyers who are looking for what you sell. For most product categories this delivers the lowest CAC when it is executed well.
Add retargeting second. Your website visitors are warm audiences. Retargeting across Google Display and Meta typically returns 3 to 5 times the ROAS of prospecting campaigns, because you are closing people who already showed interest.
Layer in Meta prospecting third, and only once you have proven creative from organic social and enough data to build Lookalike Audiences. Meta prospecting without those inputs burns budget on poor targeting.
How do you structure campaigns so spend stays under control?
Algorithmic bidding in Google and Meta is powerful, and it needs guardrails. Without them the platforms optimise for their metric, not yours.
For Google Shopping, use Performance Max with asset groups that separate brand from non-brand, and set target ROAS bidding on realistic targets taken from your historical data rather than wishful thinking. Give new campaigns two to four weeks of learning before you judge them.
For Meta, use Campaign Budget Optimisation with three to five ad sets at most, each targeting a meaningfully different audience segment. Twenty ad sets at $5 a day each is less efficient than three at $33 a day. The algorithm cannot optimise with that little data per set.
Why is creative the variable you control most?
Ad platform algorithms have become sophisticated enough that audience targeting differences matter less than they used to. What still varies enormously, and what you control directly, is creative quality.
For e-commerce, product-focused creative almost always beats lifestyle imagery with cold audiences. Clear product, clear price or offer, clear reason to click. On Meta specifically, video with the product visible in the first three seconds consistently outperforms static images in prospecting campaigns.
The allocation rule of thumb is 70% of paid budget to proven campaigns (known-good audiences, proven creative), 20% reserved for testing new creative and audiences, and 10% for retargeting. That structure lets you scale what works and keep learning without gambling the core budget on experiments.
What should you measure instead of ROAS alone?
ROAS is useful and incomplete. It does not account for organic sales that were going to happen anyway, it rewards campaigns that cherry-pick easy converters, and it can be gamed by over-weighting branded search.
For a more complete picture, track the new-customer acquisition rate rather than total conversions, month-over-month revenue growth from paid channels, and blended CAC across all channels. These are harder to game and sit closer to real business growth.