Optimizing to basic performance metrics can leave campaigns reeling. Profitability and incremental returns are key to finding your true North Star. The team at Mediasmith has cracked the code for optimization.
What happens when you cut your cost-per-acquisition by 30% but revenue remains flat? You may be hitting your target metric (CAC), but seem to have fallen short in actually generating additional revenue.
This happens more than you’d think.
When efficiency metrics are prioritized over economics, we can lose sight of the true goal: sustainable growth. At Mediasmith, we practice ‘economic optimization’ to ensure that our campaigns are not only efficient but also profitable.
Many successful teams can get numbers to move in the right direction. Lower CPCs. Higher click-through rates. Better quality scores. These are all essential building blocks for a strong campaign strategy, but it’s essential to ensure these are paired with profitability.
Take the brand that reduced CAC by 30% but revenue remained flat. Its team lowered CPA by shifting budget toward cheaper, high-intent branded keywords (usually a good move), and the efficiency metrics improved. Reporting looks stellar. But how do you determine the impact this move has on true incremental profitability?
One solution from our playbook is to run a holdout test. Typically, a test would go as follows: we stop bidding on your core brand terms in a subset of markets for 30-60 days and measure what happens to conversions in those markets compared to your control group.
Branded search conversions drop 30-40%, not 100%. That 60-70% gap represents people who were going to find you and buy from you regardless – through organic search, direct traffic, or because they already had you bookmarked. Now, you have a new baseline.
This is the incrementality question.
Would this sale have happened without our marketing?
We start with the math.
While creating a media plan, we make an effort to understand the economics of your business.
What’s your contribution margin by product or service line?
What does lifetime value (LTV) actually look like when you factor in realistic retention curves?
What’s your payback period? (If you’re paying back acquisition costs in four months v 14 months, your tolerance for upfront CAC is completely different)
Where are you capacity-constrained? (We don’t want to drive demand you can’t fulfill)
Once we understand the economics, we can determine what ‘good’ actually looks like. Good profit per incremental customer acquired. Good payback-adjusted LTV. Good contribution margin after CAC.
We then lead with experimentation to ensure that we identify and execute against the best opportunities for profitable growth.
Here‘s an example: a client spending about $2m annually on paid social saw campaigns performing well by platform standards: 3.2 times ROAS, $85 CPA, solid scale.
We run a geographic holdout to determine true incrementality. We take 15 designated market areas, match them to 15 control designated market areas (DMAs) with similar historical performance, and go dark in the test markets for 60 days.
The results show that conversions in the holdout markets dropped 41%, which means 59% of attributed conversions were happening without paid social influence. People who would have purchased through direct traffic, organic search, or other touchpoints.
The real incrementality looked like this:
Platform-reported ROAS: 3.2 times increase.
Actual incremental ROAS: 1.3 times increase.
This is still profitable given their unit economics, but not what the original dashboard suggested.
Here’s what can be done with that information: cut paid social spend by 40% and reallocate budget to mid-funnel YouTube campaigns and paid search non-brand terms where holdout tests showed 80%-plus incrementality.
Six months later: new customer acquisition up 31%, and incremental revenue per marketing dollar up 47%.
That’s economic optimization: making the business perform better.
The tools to practice ‘economic optimization’ are available to most of us without drastically increasing investment in tools and technologies:
Holdout testing - Turn off marketing to a control group and measure the difference. Requires statistical discipline and patience (you need to run tests long enough to account for lagged effects), but it’s the gold standard for measuring incrementality.
Geo experiments - Test different strategies or spend levels across matched markets. Particularly useful for brand campaigns and channels where individual-level holdouts aren’t practical.
Media mix modeling - Econometric models that quantify the contribution of each channel while controlling for external factors. Most valuable when you have two-plus years of data and meaningful variation in spend levels.
Synthetic control methods - More sophisticated version of geo testing. Useful when you don’t have perfect market matches or need to control for multiple confounding variables.
‘We hit our target CPA’ isn't cutting it. Private equity owners want to see clean, profitable growth. Public company boards want proof that marketing spend is driving shareholder value.
The brands that thrive over the next five years will be the ones that can model out unit economics, design clean experiments, and illustrate marginal returns to a finance committee.
