Multi-touch attribution is one of the attribution models that is widely discussed in the marketing world, often cited as “a game changer” and is also the subject of research on the benefits of applying Markov Chains to your machine learning process.
However, only a third of marketers report using any type of multi-touch attribution initiative as part of their analytics framework.
On the other hand, Google launched its attribution capabilities as part of Google Analytics in 2013 and the free version in 2017. So, most businesses could, in theory, be using some form of attribution in their analytics process. (In that case, if you’re not familiar with the other attribution models, click here )
Most companies move from “doing nothing” to a project where they can start on the Attribution journey once they understand its benefits and the need to maintain a data-aligned strategy.
Data organization
A survey by the Multi-Touch Attribution Think Tank has shed light on the difficulties in adopting MTA within enterprises.
For companies that work or plan to work on MTA projects, the main critical points cited by them are related to problems with access, organization and data connection.
Multi-touch attribution quality chart
MMA – Multitouch attribution research
No surprise here, data is the cornerstone of an MTA project, without it there is little that can be done. Implementing operations is first and foremost about organizing and linking data from multiple different sources in order to identify individual users across multiple platforms.
But how can you create a multi-touch attribution project?
In order to help you better understand this complex attribution model, we have identified the basic milestones and steps that a company needs to take to achieve a multi-touch project. Check them out below:
1 – Understand what you are trying to achieve.
Digital maturity chart
Analytics Continuum by Math Marketing
Multi-touch attribution operates in the realm of advanced analytics. In our Analytics Continuum framework, it falls under the insights category. This means that to achieve it, machine learning and other data science skills need to be in place.
MTA is not a tool that can be acquired, but a capability that must be developed. Having the code number of philippines right skills available is the key to project success.
2 – Make sure the basics work
You may have heard the saying, “garbage in, garbage out.” This means that a statistical model is only as good as the date you input it.
This is also important for the basics of any digital marketing project. Do you have a proper framework for creating and managing Campaign UTMs (Urchin Tracking Module Campaigns)? Are they created strategically, with consistent naming, with organized rules that are available for reference? If the answer is no, then this is where you should start.
This also applies to your media investments: Is the information available, organized correctly?
Additionally, make sure your web analytics solution is implemented correctly. User IDs, events, and custom categories should be up-to-date and shared by you and your team before you begin your MTA journey.
3 - Use your Google Model
If you advertise with Google Display and Search, and your Search Console is set up correctly, you can start using Google Analytics for attribution models. It works very well within the Google ecosystem, with some important limitations when trying to connect with other sources like Facebook or Twitter. However, it’s a good first step.
4 – Finding a common denominator with your data
This is by far the most complicated part of the process . A Salesforce study shows that customers use 10 channels on average to communicate with brands. To build an MTA model, you need to find a common denominator across all these different sources in order to find a single user, the same user, across all these platforms.
For customers who have already visited your website once, this should be less of an issue. Whether you develop a proprietary pixel or use a combination of Google user ID and CRM customer ID, you should be able to track all user interactions and visits to your website, even for an unknown customer.
This gets trickier when it comes to impressions. Given that some (or most) of your customers won’t be coming to your site all the time, being able to find a common denominator across multiple audiences is key.
Here, we’ll mix data with statistical inferences to connect all the different audiences into a single user. It won’t be perfect, and there will be some blind spots. But it’s a process, and with each iteration your model will get better.
5 – Build your attribution model
Once your data is connected and attributed to a user, it’s time to work on your attribution model. In order to accelerate the return on your capacity investments, starting with a basic attribution model (U, Time Decay, or Linear) is a good place to start.
However, we always advocate a personalized attribution model, in which the value of each impact is assessed based on the reality of your business. This requires training the model, constantly adjusting and refining it.
Analytics in general, and Multitouch in particular, cannot be treated as a project with a beginning and an end. It is primarily an operation, which requires constant improvement and which evolves.
6 – Removal and Plateau Effect
Once a custom MTA model is built, two new KPIs (key performance indicators) will become the norm in your daily optimization meeting – Removal Effect and Plateau. Removal Effect shows how many conversions will be lost by removing a particular medium from the marketing mix.
Plateau, on the other hand, will predict the maximum conversions expected from a given medium, given the current conditions. These two KPIs will enable better media planning and optimization, improving your marketing teams’ ROAS (return on advertising spend) .
Show me the money
Building a multi-touch attribution model is an investment of money, time, and staff. The expectation is that these investments will translate into more sales, lower acquisition costs, or ideally both. For this reason, many large companies hire specialists to take care of the project and thus achieve assertiveness and return on investment.
Conclusion – Multi-touch attribution is the solution to that old adage attributed to John Wanamaker: “ Half the money I spend on advertising is wasted; the problem is, I don’t know which half is .” Being able to understand every interaction with a customer and, most importantly, the value of that impact on the customer journey is what all marketing analytics and data science teams should be striving for.