It’s quittin’ time, so the marketing team walks into a bar. The bartender asks, “What’ll ya have?”
The search team lead says: “A round of your top-shelf bourbon. We drove $1MM in sales last month!”
The social team lead says: “Champagne all around! We also drove $1MM in sales.”
The CTV team lead says: “Your finest tequila shots. We drove $1MM in sales, too!”
Then the CFO arrives. “Give me a shot of your cheapest rot gut whiskey,” he says. “We had only $1MM in sales last month.”
If you laughed, it’s because you know it’s true. If you guffawed, you probably work in advertising analytics.
In my 18 years of experience in this field, I’ve found that there are three fatal flaws that make this joke relatable. Here’s how you can avoid them and get everyone in on the celebration.
Flaw #1: Cross-pollinating ad platform analytics
The problem: Ad platforms are greedy for credit.
A mentor of mine always says, “Show me the incentive structure, and I’ll accurately predict the behavior.” Platforms are incentivized to take credit for business results, so they take all the credit they possibly can — plus a whole lot more. These platforms reasonably take credit for sales up to 30 days (or more) after an exposure and halo sales. They unreasonably ignore the impact of everything else that contributed to a sale and take credit for sales that would have happened anyway.
The solution: Don’t mix and match ad platform data.
There’s plenty of room for improving the execution within any given platform by optimizing audiences, placements, creative, and timing. Just know that the performance is relative to the platform. It’s not absolute or comparable to the performance of other platforms and shouldn’t be used as a source of truth when determining how much credit should be given for driving actual business results.
Flaw #2: Not understanding return on ad spend
The problem: Confusing “a” (attributed) with “i” (incremental) ROAS.
aROAS and iROAS are not the same thing. They aren’t even correlated. aROAS is simply a counting exercise that gives credit for every sale that happened after exposure to an ad within a certain timeframe (typically up to 30 days, but sometimes even longer). iROAS, or incremental return on ad spend, gives credit only for sales that would not have happened without an exposure to an ad as measured through a test and control or modeling methodology.
The solution: Clarity is kindness.
Don’t fall into the trap of thinking that attributed sales are actually incremental sales or that strong aROAS is a leading indicator of iROAS. It isn’t. If you are unsure of which ROAS is being used, ask. You need to be focused on incrementality.
Flaw #3: Not casting a wide enough net
The problem: Marketing isn’t the only sales driver.
There was only $1MM in sales last month, and no analytical wizardry can change that fact. Work backwards from there to figure out what each business driver (inclusive of, but not limited to, marketing) actually contributed.
The solution: The mix is the fix.
Marketing Mix Modeling (MMM) is the only viable option (bonus points if you can inform MMM with test vs. control experiments). There are several methodologies, each of which has its strengths and weaknesses. But the two most important things are to ensure that there’s a comprehensive data set of independent variables (such as seasonality, price, and promotions) and that the model is able to distinguish between base and incremental sales. As a reminder, base sales are sales that would have happened within a given time frame regardless of marketing activities executed and are driven by things such as overall brand health and the long-term effects of product quality, distribution, and advertising. Incremental sales are sales that would not have happened within a given time frame without the execution of things like simultaneous advertising and promotions. These are the sales that should be factored into the iROAS of any given campaign or channel execution.
Side note: aROAS will factor in total sales (i.e. base sales + incremental sales), which is how you get nonsensical equations like 1+1=3.
You can avoid these three fatal flaws in your analytics strategy by effectively communicating these problems to your stakeholders and deploying the solutions above. You’ll ensure accountability and empower your organization to make better investment decisions to drive business results. And most importantly, your entire team – including your CFO – will happily raise a glass at your next happy hour.
