The goal of marketing is to change customer behavior to generate incremental revenue. Learn more about what that means here.
Maybe you have encountered this situation before: digital marketing is firing on all cylinders, producing win after win. Despite this, sales are still not meeting expectations.
The key to understanding why this happens–why some marketing “works” and some does not–is incrementality.
What Is Marketing Incrementality?
In marketing, an incremental sale is a sale that would not have happened if the marketing campaign was taken away. To put it simply, an incremental sale is one that was definitely* caused by a marketing campaign. (*definitely as in probably, as defined by statistics)
A real-life example to illustrate this concept:
I’m a bridesmaid in a wedding. The bride gives me a link to the dress she wants me to buy. It’s only available from one online retailer. I need to purchase the dress or drop out of the wedding.
If I happen to view or click a retargeting ad as I make the purchase, those ads didn’t convince me to do anything. If I had never seen the ads, I still would have purchased the dress.
Now, let’s say this retailer also happens to sell dresses for everyday wear. There are product recommendations in my shipping confirmation email. I see a dress that I like and purchase it.
I was vaguely considering updating my wardrobe for summer, but I had no solid plans. If I hadn’t received the email with the recommendations, I wouldn’t have purchased anything else from the dress retailer. This is an incremental sale.
Why Is Incrementality Important?
I recently posted the following poll on Twitter:
You invested $100 in a digital marketing campaign and it returned $500 (last click attribution).— Alex (@heyitsalexP) August 2, 2021
90% of the conversions you drove via the campaign would have taken place within 7 days, even if you never ran the campaign.
Was this campaign successful?
It’s a bit ambiguous, but the central premise is this: your return on ad spend doesn’t matter if the purchases you won were not incremental. Paying $100 to earn $500 that you could have earned for free is not a “win”.
You can see from this example that it’s possible to achieve outstanding campaign-level performance AND drive no incremental lift in sales.
It’s easy to make a false association between marketing return on ad spend (ROAS) and business growth. If my marketing activities are profitable, and sales are going up, then surely, sales are going up BECAUSE of my marketing? Right? Not always.
Oftentimes the reason a given marketing activity appears to drive a high return is because you’re inserting one more marketing touch in front of a purchase that probably would have happened anyway.
Branded paid search–ads that are triggered when a person literally types your brand name into a search engine like Google–are a perfect example. This is often the highest ROAS channel in any digital marketing mix.
The prospect is aware of your brand and motivated enough to seek you out. You’re not convincing the prospect of anything by placing one more paid step between them and your website.
How To Measure Marketing Incrementality
The most accessible way to determine if a marketing campaign is driving incremental sales is to perform a holdout test. This is when you take your target audience, split it in half, and only serve the advertising experience to one half of the audience (the test group).
Then you measure how many members of each audience convert, and how much they spend on average. If the test audience spends more, and the difference between the two groups is statistically significant, the campaign is driving incremental revenue.
There are a few challenges to this approach. If you’re attempting to run a holdout test on an entire marketing channel, business performance may decline. To get the most accurate results you should only run one test at a time, and each test could last several weeks.
There are some principles you can use to build your paid media strategy with incrementality in mind, without holdout testing everything. There are also more advanced statistical techniques for determining incrementality, which may require machine learning. I will cover both of those in a future piece.