Updates and Improvements in OWOX BI Attribution

There’s no perfect attribution model, but we’re working hard on it. That’s why we’ve improved OWOX BI Attribution , and added even more new features. In case you haven’t tried using our model yet, you can read how it works and see what makes it different from the others.

In this post, we answer the most popular questions about the updated model, such as:

  • How to give credit to the event helping, but not necessarily required for a conversion?
  • How to determine the order of steps within the funnel if customers take different paths to a conversion?
  • How to evaluate user actions if users skip steps, go back to upper stages of the funnel, or take the same step multiple times?

Being able to answer these questions is especially important to SaaS, B2B, and other non-Ecommerce projects with nonlinear conversion paths. Below, we outline the changes in our attribution model that were made to provide such projects with better results.

1. Evaluate each step in complex, nonlinear funnels

Many businesses strive to evaluate events that are not necessary for placing an order but are still important for the business. Such events could be webinar registrations, PDF downloads, newsletter subscriptions, «Contact us» page views, etc. Where should these events be placed in the funnel if there are multiple scenarios for the customer journey along the funnel?

To save our customers from cracking their skulls over these questions, we’ve calculated the probability of a visitor moving on to any other step in the funnel. This means that there’s no need in setting a certain sequence of key events in the funnel when evaluating channel performance.

For example, your funnel has the following steps:

  1. Visiting the website.
  2. Opening a product page.
  3. Reading a blog post.
  4. Watching a webinar.
  5. Subscribing to the Trial.

A user comes to your website, and that’s Step 1. Next, the user reads a post in your blog (Step 3), follows the link to your product page (Step 2), watches the webinar (Step 4) and, eventually, subscribes for the Trial (Step 5). As you can see, the user has skipped some of the steps, but then moved back to what you consider as upper stages of the funnel. Moreover, all of these events have happened during different sessions.

Here’s the diagram to visualize the user journey:

To determine the value of each step, OWOX BI Attribution will calculate the probability of a user moving through these exact steps:

  • Step 1 to Step 3.
  • Step 3 to Step 2.
  • Step 2 to Step 4.
  • Step 4 to Step 5.

However, there are cases when users don’t take all of the predefined steps to achieve a goal. Imagine that your user skips a few steps and goes straight to the fourth one, and then makes a conversion. Back in the time we used to evaluate all of the funnel steps, then give the value of the skipped second and third steps to the fourth one.

Thus, all the conversion value should only be given to the steps that participated in the customer journey. We’ve changed the logic of calculations in OWOX BI Attribution with respect to the above.

The conversion value is now distributed only among the actions that contributed to conversion directly.

2. Evaluate repeated funnel steps

Sometimes users take the same action in different sessions. For example, they may view the same product page, read the same blogpost, remove and add a product to the cart in different sessions.

We used to give the value only to the sessions in which users took a certain funnel step. However, we came to realize that every repeated action influences the user getting closer to a conversion. The closer in time the repeated action, the bigger the influence of the action. That’s why we now evaluate all the repeated user actions according to the time decay logic.

First, we determine the value of the step in general. Next, we check how many times the step was repeated, and how long before the conversion it was taken. After that we distribute the step value among all of the events.

For example, in the diagram above we see that the user repeated Step 4 three times. Each session in this case will receive a weighted percentage of value for this step. Session 1 will get the lowest value, Session 3 will get the highest one as it was the closest to the conversion.

3. Integration with Google AdWords

Another frequently asked question is what to do next, after the value of each marketing channel is determined. Before, we’ve published a post in our blog explaining how marketing specialists and chief marketing officers can use custom attribution models , and how to evaluate the cost efficiency of the OWOX BI Attribution. Now we’re gladly sharing the news that the attribution results can be used to automatically manage bids and optimize budget allocation.

We’ve integrated OWOX BI Attribution with Google AdWords. Thanks to that, you can automatically send the results of the attribution calculations to the above-mentioned services to manage your bids based on the accurate evaluation of channel performance.

OWOX BI Attribution helps evaluate the input of each ad channel into the customer journey, and take account of the mutual campaign influence. You’ll improve the overall ad channel performance by optimizing your bids based on the attribution results. Learn how to set up the integration in our Help Center.

4. Evaluate DoubleClick ads

We’re constantly working on OWOX BI Attribution to let our customers add any events they consider important, to their funnels. That’s why we’ve built the integration for OWOX BI Attribution with DoubleClick.

This means that you can now evaluate how DoubleClick ad impressions contribute to sales. Imagine that a user visits your website, then sees a banner ad on a third party website. A few days later, the user gets back to your website via another channel and makes a purchase. By adding the display ad impressions to the funnel, you’ll discover the influence of display ads on the customer journey to a conversion, even if there were no clicks on display ads. To set up the integration, follow this link.

To sum up

  1. Users can often skip some of the funnel steps, move back to previous steps, and repeat certain steps in the conversion paths.
  2. To evaluate the steps that participated in the conversion journey, and to find out if the user moved back to previous steps, we’re now calculating the probability of all the possible transitions within the funnel.
  3. We’re evaluating repeated steps based on the time decay logic. The closer in time the event to the conversion, the more value it receives.
  4. Now OWOX BI Attribution provides more accurate results for projects with complex, nonlinear customer journeys along the funnel.
  5. Our clients can send the attribution results to Google AdWords.
  6. You can now evaluate how DoubleClick ads contribute to the customer journey.

Don’t forget that you can help us improve OWOX BI Attribution. Simply write your suggestions and ideas in the comment section.

Used tools