all insights

Marketing attribution: a fictional Tableau case study

Using attribution modeling to uncover opportunities in the customer journey

by Lefteris Grammatas

To stay competitive, companies need to continuously deliver better and better experiences for their customers. But the range of marketing touchpoints can be overwhelming to manage, and the customer journey can rapidly become a series of disconnected and suboptimal interactions.

So, how do we design cross-channel customer experiences that are complimentary to every customer and simultaneously unique to individual customers? By starting with an analytically complete picture of the customer journey and using attribution modeling to examine how our different channels contribute to the buying cycle.

Analyzing the customer journey

To analyze the customer journey, we need visibility into customer interaction data across the entire customer journey at a customer record level—from the first interaction, to purchase, to product usage. But with the exception of e-commerce businesses, most of us don’t have robust data on every interaction at an individual customer level—and that’s probably okay. You can still get a great deal of value out of attribution by applying analytical proxies.

  • For known customers: Try to merge the customer journey as much as possible at the most granular level. Use cookie information, login information, customer data, and purchase information.
  • For anonymous customers: Assign a monetized value to digital actions, to evaluate performance. For example, in the automotive industry, we could apply a monetized action around test drive requests for prospects and customers who physically come in and test drive vehicles. This could be an aggregate number on a daily, weekly, or monthly basis.
  • If you can’t merge the data: Most of your owned media is in a digital format that makes it easy to use a web analytics technology to track the vast majority of the customer journey. Web analytics technologies can also be used to append offline information at a customer level.

Using attribution modeling

Once you have a complete and normalized customer record, you can apply attribution modeling and customer journey analysis to identify:

  • The time between interactions, the number of interactions, and the sources of interactions through your owned media in the customer’s buying cycle.
  • Which properties are the least and most valuable in converting the sale and where the missing opportunities are.

From an attribution modeling perspective, the easiest models to explore online or offline data are conditional models. The most common types of conditional models are first touch, last touch, linear, and time decay.

Marketing touchpoints

To demonstrate how touchpoint attribution can be applied to your business, we took an anonymized sample dataset from an automotive client to analyze and provide optimization recommendations. The data includes both online and offline interactions pre- and post-sale. We used Tableau to explore the data and extract insights, but virtually any business intelligence tool is capable of this level of modeling. Our fictional case study is for an automotive client that we will name Tetra Automotive, a luxury car manufacturer.

The challenge

The CMO of Tetra Automotive wants to know what channels to invest in next quarter. Tetra’s key consumer touchpoints are: offline dealer visits, website, mobile web, and email.

Let’s say Tetra created a master customer profile through a web analytics technology and integrated offline and online customer records to address the question.

The solution

1. Monetizing customer interactions

Tetra’s customer journey expands across online and offline channels. As a result, customers are constantly using digital touchpoints that influence the sale, but the business could not connect online and offline interactions—until now. The same challenge occurs when a consumer visits one dealer for the first time, but purchases at another dealership. We know that book a test drive and dealer visitshave strong correlations with sales, and for that reason we assign a monetized value that reflects their influence in the sales cycle. The purchase and service check-in actions are revenue-based events, and from there we were able to evaluate their profit margins.

We created the bellow monetization scale based on each action’s estimated profit margin:

Investment pyramid

The next step is to amend your customer data set with these monetization values, which we were able to aggregate and report on at a channel level.

2. Evaluating customer interactions on a lifecycle basis

For Tetra automotive, the customer lifecycle has the following stages:

  1. Consider and explore: when the customer is browsing
  2. Vehicle selection: when the customer is searching for a car
  3. Purchase: when the customer is purchasing a car
  4. Get the vehicle: when the customer is picking up the car
  5. Service: when the customer is bringing the car for service

To report on these five lifecycle stages, we created a URL and action classification map, and then assigned the appropriate classification to the customer interactions.

The result

We used Tableau to generate insights that will be actionable and easy to understand based on the data set:

1. Analyst interactive dashboard for data exploration

Analyst interactive dashboard

2. Executive storyboard to communicate recommendations

Executive storyboard to communicate

We then aggregated the findings in a Tableau storyboard to showcase key findings. One of the most revealing findings was the fact that digital channels were underutilized in the early stages of the buyers’ journey—where most buyers visit dealerships first. But this also suggests brand awareness may be driving the vast majority of prospects to the dealer and digital channels could and should be used to drive more leads at the consider and explore stage. The findings also highlight that the automotive client needs to do more to engage customers on mobile and digital channels after the sale.


You need to be where your customers are, but you don’t necessarily need to optimize at the one-to-one personalization level to improve the customer experience and boost sales. By applying some simple conditional modeling and aggregating individual cross-channel interactions, you can uncover overlooked opportunities to improve customer engagement.