It depends on who you ask, but the consensus seems to be that a combination of both attribution methods is best

Probabilistic attribution and deterministic attribution are two methods used to assign credit for conversions in marketing and advertising.

Both methods have their own use cases. A deterministic approach can be useful in cases where last touch is the key to measure or cases where the data is not enough to make proper probabilistic attribution. A probabilistic approach can be the most appropriate when a more accurate measurement is needed, especially when the customer journey has multiple touchpoints.

Probabilistic Attribution

Probabilistic attribution uses probability theory. It models the complex customer journeys that often involve multiple touchpoints and interactions by assigning a probability of conversion to each touchpoint. The probabilities are based on the data available on the customer journey and the model assumptions; then, the probabilities are used to assign credit to the touchpoints by assuming the probabilistic dependencies between them.

Probabilistic attribution models can take many forms, but they all share the same principle of using probabilistic assumptions to assign conversion credit to different touchpoints. Some common models include Markov Chain models, multi-touch attribution models, and Shapley value models, each with their own specific assumptions and strengths.

This approach allows for a more accurate understanding of the relative value of different marketing channels and tactics and can help to identify the most effective touchpoints in a customer journey and how they interact with each other. However, it can be computationally more complex and difficult to understand than deterministic attribution, especially for those without a strong background in statistics or probability theory.

Deterministic Attribution

Deterministic attribution assumes a definite cause-and-effect relationship between touchpoints and conversions.

It typically assigns 100% of the credit for a conversion to a specific touchpoint, such as the last click or the first click, regardless of any other interactions that may have taken place.

Some common models include last click, first click, linear and time decay attribution models; each of them have the characteristic of assigning the full credit of a conversion to one single touchpoint.

Deterministic attribution models are simple to understand and implement, but they do not account for the complexity of customer journeys that often involve multiple touchpoints and interactions. Because it only takes into account a single touchpoint, it does not help to understand the relative value of different marketing channels and tactics and can lead to an oversimplification of the customer journey.

This method can be useful in cases where last touch is the key measure to evaluate performance or when data is not sufficient to make proper probabilistic attribution, but it may not provide an accurate representation of the customer journey, and it may not fully capture the value of touchpoints that happen earlier in the customer journey.

The Pros and Cons of Probabilistic Attribution

Pros of probabilistic attribution include:

• The ability to account for the complexity of customer journeys, which often involve multiple touchpoints and interactions.
• It can help to more accurately understand the relative value of different marketing channels and tactics.

Cons of probabilistic attribution include:

• It can be more difficult to understand and interpret the results than deterministic attribution, especially for those without a strong background in statistics or probability theory.
• It is computationally more complex and might require more resources to implement.

The Pros and Cons of Deterministic Attribution

Pros of deterministic attribution include:

• It is simple and easy to understand, especially for those without a strong background in statistics or probability theory.
• It is computationally less complex and might require fewer resources to implement.

Cons of deterministic attribution include:

• It does not account for the complexity of customer journeys, which often involve multiple touchpoints and interactions.
• It does not help to understand the relative value of different marketing channels and tactics.

What Do Marketing Experts Think?

There are many marketing industry experts who have written and spoken about the pros and cons of probabilistic vs. deterministic attribution.

One of the known experts in this field is Avinash Kaushik, a digital marketing evangelist at Google and author of “Web Analytics 2.0” and “Web Analytics: An Hour A Day.” He said that the “last-click” attribution model is fundamentally flawed, as it ignores all of the other interactions that happen across channels and devices before the final click. He suggested using a combination of different attribution models, including both probabilistic and deterministic models, to get a more complete picture of the customer journey and the relative value of different marketing channels and tactics.

Another expert in this field is David Raab, founder of the Customer Data Platform Institute. He has written extensively on the subject and argues that deterministic models are simpler to understand and implement, but they tend to overlook the complexity of customer journeys that typically involve multiple touchpoints and interactions.

While probabilistic models are better suited to take into account the complexity of customer journeys, they can be more difficult to understand and interpret. He suggests that in most cases, a hybrid approach combining elements of both deterministic and probabilistic models is the best way to accurately understand the customer journey and the relative value of different marketing channels and tactics.

Another anonymous expert states that one of the most significant challenges with deterministic models is that they do not take into account the complex customer journeys that are typical in today’s cross-channel and cross-device world. They suggest using a probabilistic approach that can be a more effective way to understand the true impact of different marketing tactics on conversions and revenue.

These experts and many others in the industry have pointed out that the choice of which method to use depends on the specific use case, and that a combination of both methods may be most appropriate in many cases.

Which Attribution Method Fares Better as Third-Party Data Comes Under Attack?

The demise of third-party data may impact the use of both probabilistic and deterministic attribution methods, but in different ways.

The restriction or lack of availability of third-party data can make it more challenging to use probabilistic attribution models, as they typically rely on large amounts of data to make accurate calculations and to make the assumptions behind the model. Without sufficient data, it may be more difficult to assign conversion probabilities to different touchpoints, which can result in less accurate or less reliable attribution results.

So, the limitation of third-party data might favor deterministic attribution methods. These methods are typically simpler to understand and implement, and they don’t require as much data as probabilistic methods. Furthermore, if the data that is available is incomplete or unreliable, it can be difficult to create accurate probabilistic models, but deterministic models are less affected by this.

That being said, it’s worth noting that the limitation of third-party data can also force companies to focus more on first-party data; that is, data they collect themselves. This can be a good opportunity to use that data to make better decisions about customer journeys and improve the accuracy of both deterministic and probabilistic methods, which can be used to better understand the customer journey, their behavior and interactions.

Learn more about LeadsRx Attribution™ or click and schedule a brief demonstration today:

Request a Demo