In a previous post, I wrote about some research we explored in our newest guide, Multi-touch Attribution for Companies with Sales Teams. In this post, let’s look at a few common and advanced attribution models and concepts so you can make the most of the research and implement a b2b marketing attribution model in your organization. But first, why should you consider multi-touch revenue attribution? Simply, it can:
✔ Reduce your cost per lead and ultimately cost per customer by valuing all parts of the marketing funnel, making optimizing marketing faster and easier.
✔ Remove subjective variables from optimization to help you make the right decision.
✔ Decrease sales cycle and sales costs by bringing in more of the “right” leads and less of the “wrong” ones.
Common attribution models
If you’re new to multi-touch attribution, these basic models may be the place to start.
Even: The most basic model where credit is applied evenly across all marketing touches. This is typically used when your goal for ongoing marketing contact and no one touch is known to convert them to a customers. For example, media companies when each page view has equal monetary value, because it is ad driven.
Time Decay: Most of the credit is applied to the touches that most recently drove the desired action. This is typically used when you want ongoing contact, but are looking for a final action, such as a sale. For example, companies with a VERY short sales cycle, such as e-commerce where there is an immediate revenue-generating call to action.
U-Shaped / Position: 40% of the credit is applied to both the first touch and the last touch with the remaining 20% being applied evenly to all other touches. This is best used when your goal is to drive awareness and also action, such as companies with longer sales and marketing cycles.
Custom: Custom weighting for the different touches as set by your organization. This could be based on cost or effort, time, or a host of other factors, such as position (weight the first and last touch higher). Later in this guide we’ll explore a few advanced custom models.
For those who are a bit more adventure or who are already using basic models, here are some advanced concepts to apply in your organization.
Adaptive, or algorithmic, attribution models have the ability to work for your company, so you optimize your media based on your customers actual performance through the marketing funnel, rather than a fixed model.
The common models mentioned earlier in this post are fixed models, where changes are made manually and are generally built using your current media plan as a starting point. It’s a great way to get started, but means optimizations are slower and time consuming as they are done manually and not real-time.
With adaptive attribution models, the model directly and dynamically informs media based on what’s performing. For example, bids could be automatically raised on AdWords if they are bringing high-quality customers.
One other way to apply attribution modeling is based on marketing type. If a generic ad points potential customers to the homepage, but a whitepaper starts people deeper in the funnel, the whitepaper should get more of the revenue credit even though they may both be first touch.
With lead generation you often receive a person’s job title and with this you can apply role-based modeling. For example, if one conversion path or marketing touch is known to bring more senior decision makers that convert faster or more cheaply, you can adjust your model to optimize for these leads.
Negative attribution modeling is taking the concept of attribution modeling is applying negative credit in the rare case where a marketing touch actually hurts conversion. Hopefully this is one you never have to implement.
Finally, cross-device attribution modeling can be helpful for when a touch happens on multiple devices, such as mobile then laptop. This is one area ad networks are taking a more serious look at as the world shifts from mobile being a second screen to a first screen. It comes with a unique set of challenges as cookies are set at a per machine basis. Facebook's ad exec, Carolyn Everson, said at a Fortune-sponsored event at CES 2014 that they are “barely scratching the surface of what it could accomplish” and that what’s missing is the ability for companies to get credit for advertising that runs on its website that, in turn, affects buyer decisions alongside other factors, like TV.