How does the channel portfolio optimizer work?
Borrowed from Finance: Markowitz mean-variance
Harry Markowitz showed that investors can reduce risk without giving up much return by mixing assets that do not move together. Marketing channels behave the same way when seasonality or algorithm changes hit them differently.
Enter the expected ROAS and variation for three channels, the correlation between them and your risk aversion. The tool searches splits in 5% steps.
Worked example
A US brand expects ROAS of 3.5 on search (sd 0.8), 4.2 on social (sd 1.6) and 2.8 on email (sd 0.5), with correlation 0.3 and risk aversion 2. The best split is 65% search, 25% social and 10% email for an expected ROAS of 3.6.
How should I read my result?
The split leans toward the highest-return channel and accepts swings.
Balanced. Good for most brands with steady budgets.
The split favors stable channels. Expect lower but steadier returns.
Embed this calculator
Questions about channel portfolio optimizer
Where do expected ROAS and variation come from?
Use the average and standard deviation of weekly ROAS by channel over the last six months.
Does it account for saturation?
No. Combine it with the saturation curve for each channel.
Why 5% steps?
Finer steps give little practical gain and imply false precision.