What is incrementality in marketing – and why it’s vital for marketers to obsess about incrementality?

“I click, therefore the ad is working” -philosophy isn’t cutting it anymore in 2023. Understanding incrementality plays a key role in improving marketing effectiveness.

Measuring the success of a marketing campaign used to be much more difficult in the past.

Companies’ media mix consisted of offline media channels which don’t leave traceable footprints marketers could then measure to understand which channels were driving consumers to stores.

Let’s look at an imaginary Outdoor gear & clothing retailer ForteSell.

During the early 2000s, ForteSell ran marketing campaigns on a frequent basis. After each campaign its marketing manager collected the sales and marketing data and analyzed:

• What happened to the advertised products’ demand during and after the campaign
• What part of the demand uplift was caused by different marketing activities
• What part of the demand uplift was caused by a discount (if there was any)
• Were there any effects that were caused by external variables, such as competitor activities or holidays

As we can see the analysis wasn’t a light one, and you couldn’t say with full certainty that this ad caused this sale.

Yet it told the marketing manager much more than what the current attribution tools tell.

Because it measured incrementality.

How today’s marketing campaigns are typically measured

The age of digital marketing has twisted our perception of how marketing effectiveness can and should be measured.

Campaigns and channels are measured by their clicks and conversions, which at first makes total sense. After all, customers can’t buy without first clicking an ad.

Moreover, as the purchase wouldn’t have happened without that last click which then converted into sales, applying last-click attribution (which is the most common attribution model in digital marketing) should reveal the most effective channels in each campaign.

Right?

Well, it’s not as straightforward as we would wish it to be.

Let’s fast-forward ForteSell to the early 2020s.

The company has moved from offline to online media due to its measurability and better reach amongst younger audiences.

Google Analytics analyzes the campaign performance for them, providing a neat report of conversions allocated by source.

The past few years have been great in terms of the results: Consumers have spent more time online, and the shift to online-only seems to have worked out nicely. Especially search ads have been working spectacularly.

There’s just one problem. Since the pandemic started to simmer down, there seems to have been a decline in digital media ROAS, even with channels that used to drive amazing results.

ForteSell’s digital marketing team has tried to combat the decline by increasing its digital media budget aggressively, but this hasn’t produced desired results.

Frustrated, the Head of Digital Media at ForteSell shuts down the Search ads to make savings, as they’re forming a substantial part of the media budget.

"Their accidental “lift test” did, however, yield an interesting insight into the incrementality of the search ads: a large part of the conversions driven by search ads has not been additional demand generated on top of the baseline demand, but demand stolen from organic channels."

The team logs into GA next week to see what damage this move has caused.

To their surprise, the overall conversions have remained almost the same as in the past. The dip in Search ads conversions has been complemented by a rise in organic traffic.

What in the world has just happened?

The difference between attribution and incrementality

The above example describes the misconception about digital attribution and especially last-click attribution.

Incrementality reflects how much additional demand marketing has generated on top of the demand that would’ve been there without any short-term marketing.

Attribution results reflect how the demand is divided by different sources. But it doesn’t distinguish whether a conversion would've happened regardless of that channel or not.

This might sound like a minor difference, but in reality it can distort our understanding of marketing effectiveness.

Let’s return to our legendary Outdoor gear & clothing e-commerce ForteSell!

By solely relying on the last-click attribution results tools like Google Analytics provide, ForteSell’s marketing team would’ve been lost after halting the search ads.

Their accidental “lift test” did, however, yield an interesting insight into the incrementality of the search ads: a large part of the conversions driven by search ads has not been additional demand generated on top of the baseline demand, but demand stolen from organic channels.

The team decides to apply Marketing Mix Modeling (MMM) to get to the bottom of the problem.

Unlike attribution modeling, MMM utilizes time series analysis to learn how different factors affect the final output. An extremely simplified example would look like this:

The outcome looks something like this:

By analyzing numerous samples the model tries to find the signal between the sales uplifts and internal (mostly what media channels have been active) and external (does for example seasonality, weather or competitor activities explain part of the results) factors.

The difference between attribution & MMM results?

Well, this:

Reason for the difference: One measures incrementality and one assigns credit based on certain guidelines (for example last-click attribution assigns 100% credit to the last touchpoint prior to the conversion). 

Attribution models, such as last-click attribution, are inherently deterministic : They will assign credit to touchpoints that have led to the conversion without further thought on whether the conversion or contribution has been incremental or not.

Models that measure incrementality, such as Marketing Mix Modeling, start by estimating what is the baseline, so what would’ve happened without any marketing activities. Demand on top of the baseline is then modeled based on the time-series analysis similar to the graphs above (except that the base sales is never as flat as in the example).

This approach typically reveals a “less effective” but much more realistic picture of marketing effectiveness when comparing the results to attribution modeling results.

The upside?

The results are now showing the true effect of your advertising. How your marketing dollars are converting into new sales.

Which leads to the question of whether decisions about budget allocation should be based on attribution at all.

If the intention is to drive additional demand on top of the current results with the media investments, attribution might fall short on this task.

Incrementality vs. attribution: Which is better?

Let’s get back to our ForteSell example once more.

After implementing MMM the team found out that a large part of the growth during Covid has been due to an increase in the baseline. As indoor activities were shut down people opted for outdoor activities, and as shopping malls were closed consumers had to resort to e-commerce.

Without accurate baseline estimates the growth in demand could be easily attributed to the channels that triggered the conversions. But as the team now knows (thanks to MMM), a large part of these conversions would’ve happened one way or another – and not mainly due to the ad.

Opting for incrementality instead of attribution in this case helped the ForteSell team to discover the reason behind the declining ROAS (in fact the ROAS wasn’t declining, it was just set to the correct level).

Although the example is fictitious, it reflects the situation in real life. A study by Tom Blake, Chris Nosko, and Steven Tadeliset in 2015 revealed a similar effect in pausing branded search ads:

Understanding incrementality plays a key role in optimizing marketing effectiveness, whether it’s offline, online, or both.

If you’re investing +20 000€ a month in +3 media channels and haven’t measured the incremental uplifts for these channels, you should start today. One option is to run lift tests , which reveal the incrementality within the tested channel.

Alternatively, you can discover the incremental uplift for all of your media channels with Sellforte’s Marketing Mix Modeling Software-as-a-Service.

Curious to learn more? Book a demo.