Attribution and incrementality: solving marketing’s holy grail

Avinash Kaushik

Avinash Kaushik

Chief Strategy Officer

7th March 2023

~ 5 min read


Marketing budgets are always under a measure of scrutiny for value being delivered, but this is never more true than in tough times. When the broader economy is in a bit of a pinch, and there's uncertainty related to a bounce back, it never hurts for a Chief Marketing Officer (CMO) to have a complete and provable assessment of marketing's impact.

One challenge CMO's face is that marketing is so many things. Brand and performance, obviously. Online and offline. Owned, earned, and paid. Mass audiences and influencers. And, that's just scratching the surface.

If you narrow the focus to often the largest piece of marketing budgets - paid media - the discussion rapidly leads to two concepts: attribution and incrementality.

In this blog which follows on from our presentation at PMW Unlocked 2023, we break down these two concepts and reflect on how an optimal focus on what's essential will ensure your CMO becomes your Chief Financial Officer's (CFO) BFF.

The last click trap

We'll use this scenario throughout today's article: You spend £3m on advertising. £1m each on Bing, Facebook, and email marketing.

When you pop open your favourite analytics tool and look at the conversions resulting from your £3m in spend, it uses the last click attribution model to assess success. It reports 300k conversions from Facebook.

So, was the £1m you spent on Bing, and the £1m you spent on email entirely useless?

No. Of course not. 

The converters simply happened to make the last click before conversion on Facebook.

It’s not all that difficult to see that using last click attribution to assess the success of your marketing initiatives is a career limiting move. 

The upside to attribution modelling

Attribution modelling is the science of ensuring credit is distributed across all marketing activity that has a positive impact.

In the context of our simple use case, attribution modelling ensures that we understand the value of £1m spend on Bing and £1m spend on email marketing.
We’re incredibly fortunate that tools like Google Analytics already include attribution modelling solutions for free. When you login, you’ll see several models you can apply, such as first click, even attribution, linear decay, to name a few.

Our recommendation is to skip all those models and jump directly to the sexy and cool thing of the moment - machine learning.

Google Analytics includes a data-driven attribution model. It uses machine learning to analyse millions of rows of media strategy, consumer behaviour, and outcomes to identify patterns that more accurately identify the role of every marketing channel in driving success. The model ingests, analyses and identifies what a team of humans can’t.

When you finish this analysis you'll be able to identify that while Facebook had a key role in driving conversions, your £1m investment in Bing drove 50k of assisted conversions and your £1m investment in email marketing drove 150k. 

In an instant, your assessment of what's driving success changes. You proceed to make smarter decisions. There's a lot of business profit in your world. And best of all, your CMO loves you.

The upside of incrementality

It’s crucial to understand that attribution is not the same thing as incrementality.

Incrementality is the art and science of identifying results that would not have occurred otherwise.

Translation: If we hadn’t spent the £3m on Bing, Facebook and email, how many of the 300k conversions would we lose?

Would we lose 20% of the conversions? 30%? That's the incrementality question.

Attribution is based on marketing's claimed conversions. Incrementality identifies marketing's actual conversions.

Incrementality is hard to measure (if you need help, ping our world-class analytics team!). We recommend breaking your initiatives into three parallel paths - channel-silo incrementality, xStack incrementality and portfolio incrementality.

1. Channel-silo incrementality

Channel-silo incrementality is the answer to the question ‘how many of the 50k conversions from Bing would I lose if I hadn’t run paid search advertising on Bing?’

Channel-silo incrementality measurement is done using conversion lift studies (CLS), which is offered by all major advertising platforms. If you ask your Facebook representative to set up CLS for you, you'll be able to identify in three weeks (or a little more, depending on your Facebook spend) that of the 100k claimed conversions, the actual conversions are 40k (or 5k or whatever).

Across your major ad platforms, use this data to make immediate tactical improvements to how you spend your budget. Your CFO will like this.

2. xStack incrementality

xStack incrementality is the answer to the question ‘Do I need to spend £1m on Facebook and £1m on Instagram separately to get all the benefits of advertising on the Facebook Stack?’

Measuring xStack incrementality will identify that you don't. If you look holistically across your spend on Google, YouTube and Display, you can spend less than one + one + one to get the same number of conversions.

You'll use matched market tests (geo experiments) to measure xStack incrementality. You'll make changes to your quarterly budget allocations and maximise value. And your CFO will love this.

3. Portfolio incrementality

Portfolio incrementality is the answer to the question ‘if I look across my entire marketing spend, brand, performance, owned, earned, retail, promotions, everything… what's marketing's true incrementality?’

This is the hardest question to answer. You'll need to use media-mix models (MMM). This will require an extraordinary amount of data and spend, to separate signal from noise. Ideally, you'll use an MMM approach that’s smart enough to look at your spend as a portfolio (most MMMs look at the world in silos).

At the end of this exercise, you’ll get the two golden KPIs - incremental sales driven by marketing. Cost Per Incremental Sale. You'll use this to set annual budgets, and create higher standards for every single marketer to meet. Your CFO will be delirious.

The bottom line

Do you need to work on both attribution and incrementality (did we already say you should call our team)?

If you’re a small business (less than £5m in annual marketing spend), we recommend only focusing on attribution. At that level of spend, it’s very hard to get a solid enough signal from incrementality measurement. Attribution will change your entire marketing execution approach.

If you’re a large business (more than £20m in annual marketing spend), we recommend solving incrementality first, and attribution second -  and to do this based on incremental conversions. It will change your life.

Good luck!

Sign up to Croud’s Digital Digest

Get Croud's monthly newsletter, which is packed with the latest news from Croud from across the globe, along with updates and commentary around the latest developments in the digital marketing space.

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.