Marketers have struggled for years to understand the value of their activities. Purchase journeys remain ambiguous for most marketers despite the availability of innovative solutions in analytics and data science across the industry. It’s still often unclear what influence marketing activities have on a sale, for example.
Purchase journeys can be analysed through customer interactions on a firm's marketing channels. One such solution is to deploy out-of-the-box marketing attribution models in analytics. However, customer journeys are complex. Marketers often quickly find that off-the-shelf models do not accurately reflect the intricacies of customer journeys.
The way forward is for marketers to abandon the default, out-of-the-box methods of marketing attribution that have been used for years and embrace Data Driven Attribution (DDA). In this article, we'll explore the inadequacy of out-of-the-box marketing attribution models and dive into why DDA provides a focused lens to understand multi-touch journeys.
It’s time to trust in the objective results of the machine.
What is Data Driven Attribution (DDA)?
Data Driven Attribution is a machine learning method to ascertain the influence of marketing channels in customer decisions. Instead of relying on outdated default marketing attribution models (with assumed rules on how channels behave), machine learning models like DDA remove subjectivity, providing a detailed view of the real impact channels have made on final purchasing decisions.
Why does using the right marketing attribution model matter?
Not all marketing activities are as influential as each other. A social media post is likely to have a different level of influence to a PPC ad, even though they may be connected to the same overarching campaign. As a result, it would be inaccurate to assume that all activities are equally as effective in driving (for example) conversions.
Understandably, it is not easy for marketers to uncover the impact of their campaigns when multiple activities are running across channels - sometimes at the same time. Apart from determining which channels work best for messaging or products, marketers will also want to plan and budget effectively. This is especially critical, given the fact that marketing budgets are often the first to be squeezed.
Further complicating the picture is that different channels play different roles in the consumer's purchasing decisions across industries. For example, digital marketing activities that generate impact for a retail B2C business will translate differently in the private healthcare sector. The same is true for B2B applications. A one-size-fits-all solution simply doesn’t exist, which is why default marketing attribution models will never be adequate.
Marketing attribution models also help marketers identify the best channels for:
- Introducing a customer to the business
- Closing and decision making moments
- Planning and scaling marketing activities
- Reacting quickly to changes.
There are two types of marketing attribution models:
Basic: Out-of-the-box models, such as first click, last click, positioning based, time decay. All are rules-based, typically assuming that decisions are made by consumers in a linear fashion.
DDA (Data Driven Attribution): Instead of using predetermined rules and assumptions created by humans, machines are left to learn from the data and determine the optimal credit weighting for each channel.
Basic, rules-based attribution models are often the go-to solution for marketing departments as they are simple to understand. But they don’t tell the full story.
What’s wrong with basic marketing attribution models?
We’ve listed the main types of basic marketing attribution models and why they’re not ideal:
- Last click: All credit goes to the last click before the conversion or sign-up
- First click: The reverse of the above
The above methods fail to take into account any other touchpoints that may have influenced a customer journey. The methods below do account for various stages of the conversion process to varying degrees, but they still rely on the assumption that customer journeys are linear. As a result, each channel earns credit on recency or their position in an overly simplified decision process:
- Last non-direct click: The final indirect channel used before the conversion of the customer receives all the credit
- Time decay: Credit increases depending on the recency of the conversion or sign-up, with the last activity earning the most credit
- Linear: Every stage receives an equal amount of split credit
- Position based: First and last interactions gain equal split credit.
DDA is the only attribution method that establishes a representative view of customer journeys, which are (after all) unique to each individual company. In Data Driven Attribution, a machine learns from large swatches of customer journeys over a given period, analysing all the messy and complex interactions between channels as they are without assuming linearity.
DDA assigns credit based on the relationships between channels that have been computationally analysed. Therefore, DDA models can ensure credit is more objectively attributed. That’s significantly more representative of the reality of actual customer journeys, which are messy and complex.
The ‘best’ marketing attribution model is…? Subjective
DDA provides the most complete view of channel performance. Yet, algorithmically, there is more than one way to run DDA. Marketing platforms such as Google Analytics 360 apply a game theory approach using Shapley Values.
Other platforms may use probabilistic methods such as Markov Chain. More digitally advanced companies will be using custom-built DDA models, which are optimised regularly. This means that whilst DDA is the superior choice over basic linear models, there is no single "best" way to apply DDA. Every business is different, and marketing activities will vary for unique sets of customers.
For marketers intending to test the waters with DDA before making a big commitment to build their own custom models, perhaps the friendliest and most accessible way is through Google Analytics 4. Google Ads now also uses DDA as the default attribution model for Pay Per Click marketing activity.
Give credit where credit’s due with data-based marketing attribution models
Customer journeys are much messier, more complex and harder to understand than out-of-the-box attribution models. That doesn’t mean that marketers should rely on default attribution models: quite the opposite.
With more powerful tools than ever at marketers’ disposal, there’s never been a better time to create Data Driven Attribution models. A mature data team can consult, design and engineer the analytics infrastructure needed to ingest and model large amounts of customer journey data. Across platforms, marketers can establish a clear view of where to assign credit and finally discover the impact of their activities.
Design next level marketing attribution models
At 26, we work with brands across industries to advance their data maturity. We guide ambitious marketers to set up their teams for success, collecting, organising and using their data to power excellent decision-making.
For more on how CMOs and senior marketers can use data to increase their influence, download our free white paper now.
Tap into our latest thinking to discover the newest trends, innovations, and opinions direct from our team.