Marketing Attribution Statistical Models, Explained

In this post, I will explain (as well as I can) what is marketing attribution and why you should know about it.

In today’s fast-paced, multi-channel world, understanding where conversions come from is key to a digital marketer’s success. Advertisers have spread to all corners of the internet, commandeering social media, blogs, influencers, emails, cookies….

But, in such a fragmented digital landscape, how do you pinpoint what led a customer to buy from you?

The answer? Marketing attribution and modeling. 

While the process isn’t foolproof, marketing attribution models can help you determine where you’re going. This leads to increased precision in your campaigns – hello, conversions!

But before we dive into the nitty-gritty of modeling, it’s essential to know what the heck market attribution is. 

What is Marketing Attribution? 

In simple terms, marketing attribution helps you identify which strategies bring in the dough. These models show you the path that client took to reach that CTA button. 

Marketing attribution is a key component of modern digital advertising. In today’s world, it’s rare for a customer to buy a product based on a single ad. In fact, many customers don’t even buy from the original device of contact. Making sense of purchasing data can get very complex, very quickly. 

To give an example, let’s take a look at Sally Shopper. 

Sally Shopper notices your ad for hand-carved spoons on Facebook’s mobile app. She clicks on the ad but goes back to scrolling without buying anything. Still, your cookies follow her around the internet. 

Then, Sally switches to her laptop. She sees one of your banner ads on her favorite blog and decides to sign up for your newsletter. But it’s not until a week later that she redeems a 15% off welcome coupon on your site. 

This is one example of one customer – and already, we have two devices and four points of contact. (And we didn’t even discuss the TV ads you paid for). Multiply this by the number of your potential customers, and it’s easy to see where you can get lost between the initial click and final conversion.

There is a joke in marketing circles “We are sure that 50% of our marketing budget is bringing profit, we just do not know which 50% it is.” 

This is where marketing attribution modeling comes in. By using technology and models to track your customers, you can analyze what works best for your business and double down on the campaigns that are contributing to the bottom line. 

What is Marketing Attribution Modeling?

Marketing attribution modeling is a way to use preset models to weight customer contact with your brand. The data from these models can help illustrate the effectiveness of your marketing campaigns. 

Marketing attribution models work by assigning values to your campaigns at the user level. (Think of our example above, but for every customer who interacts with your brand).

Then, statistical analysis extracts the data from each model. This person-centric approach is valuable for digital marketers seeking deeper insights into customer behavior. 

But we should also note that there is no perfect model with all the answers. To correct this problem, marketing mix modeling uses several models based on aggregate data. This method, too, is imperfect, but it can provide a more comprehensive view. 

Because each model serves a different function, it’s best to define a list of goals you hope to accomplish. An effective suite of models should tell you:

  • How brand perception affects your conversions.
  • Which messages bring customers in.
  • The advertisements that most likely to convert to sales.
  • The role of message sequencing.
  • If there are any external factors that consistently impact your sales. 

To achieve these goals, marketers use one or both categories of marketing attribution models: single- and multi-touch. We’re going to dissect these in further detail below. 

Single-Touch Marketing Attribution Models

Single-touch marketing attribution models are incomplete representations of a customer’s interaction with your brand. These models assume a single message is responsible for converting clicks to sales. 

Unfortunately, this means that single-touch models don’t account for the importance of cumulative exposure, or repeated contact over time. 

But these models are still helpful in analyzing your data at face value.

First-Touch Attribution

First-touch attribution models give full credit to the ad or keyword that introduces a consumer to your brand. In our example with Sally Shopper, the Facebook ad would receive 100% of the credit for bringing in a new customer – even though she didn’t buy anything until a week later. 

First-touch models assume that the initial contact is the most important because they introduce consumers to your brand. Thus, they apply full credit even if your customers don’t buy product or sign up for your newsletter. 

Unfortunately, this means that they can over-prioritize unimportant methods of marketing. Not to mention, first-touch models often exclude profitable bottom-of-funnel tactics such as remarketing. 

As such, first-touch models are most useful for allocating funds for new traffic campaigns. This includes top-of-the-funnel activities such as lead generation and forms. 

Last-Touch Attribution

On the other hand, we have the last-touch marketing attribution model. This model assumes that the final ad before purchase is the most important. As such, there is no credit assigned to any engagements aside from the click that led to a sale. (In Sally Shopper’s case, this would be the promotional email with a 15% coupon). 

Last-touch attribution marketing models are excellent at driving conversions. As such, they are often the default model in marketing software.

Unfortunately, they tend to credit lower-funnel campaigns such as retargeting and branded searches. This comes at the detriment of your non-converting – but awareness-raising – campaigns. 

Multi-Touch Attribution Models

Unlike the single-touch approach, multi-touch marketing attribution models assign credit to all touchpoints leading to a conversion. But they also presume that not every interaction should receive equal weight. 

While these models are more accurate, they do come with drawbacks, too. 

For example, the way that models are weighted may assume that some methods of marketing are more valuable, regardless of when consumers contact those ads. Others may weigh all interactions equally. 

As such, the best approach with multi-touch models is to employ more than one at a time. By extrapolating data from several analyses, you can paint a clearer picture of your marketing efficiency. 

Linear Attribution

Linear attribution models are the simplest form of multi-touch attribution. These record and weight equally each touchpoint from exposure to sale. 

In our example, that means that Sally Shopper’s model would give 25% credit to each:

  • Facebook
  • Banner ads (as a result of cookies)
  • Newsletter sign-ups
  • Promotional emails

While this method doesn’t exclude any marketing tactics, it also doesn’t consider which tactics may have been more effective in conversion. 

Time Decay Attribution

Time decay models are similar to the last-touch model in that they presume ads closer to conversion are more valuable. But, unlike single-touch models, they account for all points of contact. 

To do so, ads viewed closer to conversion are weighted heavier than initial points of contact. For example, a time decay model may rate Sally Shopper’s conversion as such:

  • Facebook ad: 10%
  • Banner ad: 20%
  • Newsletter signup: 30%
  • Promotional email: 40% 

Because each brand interaction is given credit on a downhill slope, this model may not accurately weight earlier ads in the funnel. But still, it does impress the importance of continued marketing efforts to convert customers in the first place. 

U-Shaped Attribution

U-shaped, or position-based, models score each point of contact separately. But unlike other models, it does not assume that each interaction is equal. 

Instead, the first and last point of contact before a sale receive the greatest weight at 40% credit. The other touchpoints then share the remaining 20%, like so:

  • Facebook ad: 40%
  • Banner ads: 10%
  • Newsletter signup: 10%
  • Promotional emails: 40%

This model emphasizes two important touchpoints: bringing the customer in for the first time, and the ad that led to the conversion. The downside is that it ignores any in-between campaigns that may have affected consumer decisions. 

W-Shaped Attribution

W-shaped attribution models use the same principle as the U-shaped model. But they also include a core touchpoint called the “opportunity stage.” 

In a W-shaped model, the first touch, opportunity creation, and lead conversion each receive 30% of the credit. The other 10% is then divided amongst any other points of contact, like so:

  • Facebook ad: 30%
  • Banner ads: 10%
  • Newsletter signup: 30%
  • Promotional emails: 30%

Full Path Attribution

Full path marketing attribution models build on the W-shape to assign the most credit to major milestones.

The difference between full path and W-shaped models is that the “major milestones” are not limited to the opportunity stage. For instance, other points of contact, such as signing up for a newsletter, may get a hefty percentage of the credit. 

Additionally, full path models also heavily credit post-opportunity follow-up interactions. 

Lead-Conversion Touch Attribution

As the name implies, lead conversion models give the most weight to the final conversion point. However, it also emphases some of the smaller parts of the customer journey, too. These may include first contact, newsletter signups, and email marketing interactions. 

As such, lead conversion models are some of the most popular ones. But because they focus on turning clicks to sales, they may not properly weight important early contact. 

Custom or Algorithmic Attribution

Custom marketing attribution models use machine learning to assign credit to various touchpoints. While this is the best model in theory, it requires a baseline of prior customer data. Additionally, this model requires you to adjust metrics as your data grows over time. 

Common Marketing Attribution Mistakes

With any of these models, it’s important to note there are plenty of ways to misinterpret the data. This may lead to a misattribution of credit. The danger here is that you may defund a critical component of your marketing campaign based on poor data. 

Thus, before you make any decisions on these models, you should check for these mistakes first.

Correlation-Based Bias

Occurs when one event looks like another. This can lead to the wrong marketing tactics taking credit for conversions they didn’t drive.

Brand-Behavior Bias

Happens when models misinterpret the relationship between consumer behavior and brand perception. Often times, this means that brand-building initiatives receive too little credit.

Missing Message Signals

These are more common in aggregate marketing. This happens when a model assumes an ad is ineffective when in reality, it’s promoting to the wrong market. 

Cheap Inventory Bias

Occurs when an inaccurate representation makes low-cost media appear more effective. Because the costs are lower, marketers may be unwilling to consider a flaw in the data. 

In-Market Bias

describes what happens when a customer clicked an ad to purchase a product they already intended to buy. Like brand-behavior bias, this gives a false representation of a campaign’s effectiveness. 

Choosing the Right Marketing Attribution Model

When comes the time to choose a marketing attribution model, it’s important to consider factors such as your business and consumer base. Furthermore, you’ll want to take into account your sales cycles, marketing methods, and offline ads. 

Even armed with all this data, you’ll still need to use several models to get the most accurate depiction. 

To that end, it may be wise to consider:

  1. Naming objectives for each marketing campaign.
  2. Making a customer journey map of touchpoints within the buying cycle.
  3. Ensuring your models are actionable – and editable.
  4. Promoting ads that raise brand awareness as well as sales.
  5. Testing both your ads and models – and changing what doesn’t work.

How to Get Started with Attribution Marketing

There are a couple ways to approach marketing attribution modeling. 

For instance, advanced programs can link with your CRM and marketing platforms. This lets them collect and analyze data in droves. 

Marketing attribution programs make the process easier because, as we mentioned, there are many ways a customer may find your brand. In this multi-device, multi-platform world rife with tightening privacy laws, every scrap of data counts. 

Not having to sift through it yourself is just a bonus. 

Aside from dedicated platforms, there are other ways to gain insight into the customer decision-making process. While it’s impossible to get 100% accurate data, you can improve your numbers by:

  • Setting up pixel and conversion tracking with Facebook or Google
  • Creating a brand-wide UTM tracking and tagging system
  • Determining the goals of your attribution marketing scheme, such as:
    • Finding out where customers first come to contact with your brand
    • Discovering which campaigns lead to clicks or newsletter sign-ups
    • Learning which marketing schemes generate conversions

Most important, however, you want to make sure you understand how your marketing attribution models interpret consumer behavior.

Executive Summary

If you are starting out in digital marketing, you may have just a few traffic channels which are quite easy to monitor. As you grow, you will be using multiple traffic sources and most importantly a wide variety of marketing campaigns.

It is difficult to know where is your bottleneck is or where you conversion happens when you have paid campaigns, organic traffic, re-targeting, email campaigns all at once.

Marketing attribution will help you. With use of various marketing attribution models you will be able to differentiate between the contribution of various advertising channels towards the conversion.


  • Vlad Falin

    Vlad Falin is a founder and blogger at, an affiliate marketer, and a digital marketing expert. Prior to starting this blog, Vlad accumulated over 7 years of experience in digital marketing, working on his own affiliate marketing projects, and providing consulting services. In this blog, Vlad writes to thousands of readers about starting a business online and reviews SaaS tools.

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.