The wrong answer to the right question
Attribution seems to be the word on every marketer’s lips right now. Listen in on any topical marketing discussions and you will hear it flagged as the hot topic of the year. For several years in a row now.
Dive with me into the depths of attribution, and I will answer the questions that are keeping performance marketers awake at night.
What is attribution and what does the messy middle mean?
As a marketer, you will likely have multiple marketing channels or at least multiple campaigns. The more marketing you run, the more interactions a customer will have with your business. We call all of these interactions for individual customers their customer (or buyer) journey.
For example, let's say they first find out about your business in a facebook ad. After weeks of exposure to different ads, organic content, and offline interactions, they convert. Below is an example of their customer journey.
This may seem simple enough, but the reality is that it is rarely so linear or clear. So much can happen in a customer journey, both online and offline, and we can't easily understand what goes on from when a customer first discovers our brand to the time that they convert. This is the messy middle.
Marketing attribution is the logic we try to apply to make sense of the messy middle. It's the discussion about how we allocate conversion credit to each touchpoint in a users' journey and therefore attribute conversions to specific marketing channels.
Attribution data helps you to know how many conversions you have from each channel, which in turn allows you to better allocate media spend.
Why is marketing attribution not the answer?
Marketing attribution is the holy grail. The solution is ever elusive and of great significance, promising to provide eternal conversions in infinite abundance. But marketers are focusing too intently on the latter part. We are so blinded by the promise of “eternal conversions in infinite abundance” that we disregard the elusiveness of achieving it.
This is the problem. There is no marketing attribution solution that accurately and thoroughly resolves the challenges below.
What are the challenges with marketing attribution?
Some marketing touchpoints are not trackable
Tracking clicks in digital channels is often easy, but it is reductive to disregard offline channels (even if your business is 100 percent online). By this, I mean conversations amongst consumers, offline sales efforts, billboards, etc. There are so many different marketing channels we can use, and this makes it significantly harder to track each and every one of them.
And with privacy coming to the forefront, tracking digital marketing activities is increasingly challenging.
Attribution modeling is designed for touchpoints we can track, and if we lose visibility on clicks and conversions, using attribution models will get more difficult and less accurate.
Touchpoints are codependent
The purpose of multi-channel marketing is to run ads that work together to create an environment conducive to encouraging customers to convert. A YouTube ad alone might achieve nothing, because users are not ready to convert at that moment. But if they have grown familiar with your brand after seeing a display ad, they might be more likely to convert when they see you on social media, or your ad might spark curiosity and they search for you on Google.
Marketing attribution models allocate credit without acknowledging the bigger picture of the impact of combinations of marketing efforts.
With attribution it's like saying your striker scored all the goals in your football match last week, and therefore you should put more strikers on the field. But in reality, you of course need all positions on your team if you want to score.
You need to look more holistically to see this bigger picture. You need to look for particular teammates working better together or the correct balance of strikers and defenders you need to win the league.
Similarly, your marketing efforts will affect each other, and you need to understand that to increase your marketing ROI (ROI or ROAS shouldn’t be your only goal though). For example, activity that is typically higher in the marketing funnel does add value, but it can be harder to prove with classic attribution in ROAS calculations.
Different models will give different solutions
There are multiple attribution models, and you could argue for the use of any of them. If you and I both go away and apply our own choice of attribution model to understand which marketing channel is contributing most to conversions, we will come back with different answers.
And when there are multiple answers that will give different results, how do marketers know which to use? Especially if we consider bias, people and ad platforms will tend to use models that put their work in a better light.
Which attribution model should I use?
For the aforementioned reasons I am not a fan of marketing attribution, but we should not disregard it entirely.
These are also all click-based and therefore disregard view interactions which will influence conversions, too. But impressions (and similar metrics) are harder to track, which makes these the core models for marketers.
When setting up conversions in the advertising platforms, it is important to select attribution models that work for you. This will then be used to optimize marketing campaigns.
Check out some of the most common models below.
Last touch attribution model
This is often the default model in most advertising platforms and, as with most models, it is a position based attribution model. It provides all the credit to the last interaction before a user converts.
First touch attribution model
The opposite of last touch attribution, it provides all the credit to the first touchpoint.
Linear attribution model
First and last touch are common models, but the issue is they only consider one of the many touchpoints. The step up from this is linear attribution which provides equal credit to all.
Time decay attribution model
If you want something more similar to last touch, the time decay model is for you. It provides a significant portion of the credit to the last touch and very little credit to the first touch.
Other multi-touch attribution models
The above are both examples of a multi-touch attribution model. This umbrella term includes linear attribution, time-decay attribution, U-shaped model, W-shaped model, and data driven attribution.
Data driven attribution model
This seems to be the trendiest of the multi touch attribution options right now. It uses machine learning to compare different conversion paths, identify patterns and create a model specific to each advertiser. This is the model that we use, and Google Ads recommends, for measuring and optimising conversions in ad platforms.
What attribution tools exist?
Most ad platforms provide some level of attribution analysis. Analytics platforms are a good place to start, for example Google Analytics. You can also use Google Ads to compare attribution models, which will provide some insight on what might be the right model for you to optimize.
And when it comes to reporting, there are multiple tools on the market claiming to solve attribution. But above, I mentioned multiple challenges with attribution, all of which are fundamental issues with attribution that sadly cannot be solved even if the tool is omniscient (which in itself is extremely hard to achieve amongst increasingly privacy driven consumers).
Take the average consumer, they will likely browse from a phone and a computer, often misleading us to believe they are different users. On top of this, if they have multiple emails, it’s hard to also connect these. Now add privacy into the mix: they might use an ad blocker or reject cookies, which means we lose out on even more signals. We cannot guarantee that we connect all the dots and create a clear enough picture of each user's journey.
The wrong answer to the right question
I've talked enough about the attribution problem. It's about time to provide you with a constructive solution. Ultimately, many marketers are just looking to know "how to identify what is driving conversions."
Well, it’s about contribution not attribution.
With contribution, you stop trying to fit the messy middle into a rigid model, and you instead try to understand what the contributing factors that led to conversions are.
What is contribution?
Contribution is the science of understanding how your multi-channel marketing strategy contributes to your end result of conversions.
If you increase spend in Facebook ads, what happens to conversions or search volume in Google ads?
Because everything is intrinsically linked, we need to better understand the relationship between all of our channels.
How can I understand the contribution of marketing channels?
There are different levels of difficulty to really understand the contribution, from basic analysis to advanced modeling. Analytical science sounds intimidating but it doesn't have to be (especially if you have a marketing data hub taking care of most of the steps for you, we of course use funnel.io).
Below, I will walk through how you can get started depending on the data skills you have available to you.
AB testing with ad-platform functionality will help you understand the best structure for your campaigns. But if you want to test a multi-channel approach then geo testing is a great option. Try increasing social media or display spend in certain regions and look holistically at your ROAS for each region.
A simple place to get started is plotting your metrics side by side. Can you see any patterns or potential relationships?
Be careful when it comes to correlation and causation. If your Google conversions increased in August is it because your Facebook impressions increased at that time or because summer is just a great time for your business? Use your marketing experience to help add colour to your data.
When you are ready to take your analysis to the next level, look into regression analysis. It is a statistical method used specifically to find a relationship between two variables. Plot your data on a graph and draw a line through this data (your regression line). The best fit line will help you understand how strong the relationship is. This can be used, for example, to understand the impact of display media spend on direct conversions.
Marketing mix modeling (MMM)
If you are a marketing professional with access to more advanced data teams, I would recommend trying MMM. It is not the one solution to solve all your problems, but it can provide an interesting take on which channels are best contributing to conversions while considering other external factors, historical data, and data from your marketing data sources.
If you are going to get started, make sure you have a data expert to work with the model and a marketer to provide the context behind the numbers. And bear in mind that there are multiple ways to implement MMM which can again provide different results and answers that are not entirely objective.
To get up and running, check out Robyn from Facebook. It's an experimental, ML-powered and semi-automated MMM open source R package.
Attribution modeling provides some value to marketing teams, particularly when setting up conversions in ad platforms and selecting a model to use for optimization purposes.
For marketing reporting purposes, don't find yourself striving for the marketers' holy grail. It simply doesn't exist; marketers are trying to force the complex and dynamic human behavior into neat little boxes to make our life easier and sometimes justify more budget.
Instead make use of all the data at your disposal, explore every avenue of analysis, and use this knowledge to provide guidance to your gut feelings and create the perfect data-driven marketing strategy.
Understanding how your marketing channels contribute together is the perfect start.
Next time you get an invite to the latest webinar on attribution, or your manager asks you to investigate the best attribution tools on the market, go further with skepticism and ask yourself "is this really the answer to our marketing woes?" Maybe even send them here so we can finally move on to bigger and better topics within the marketing industry.