Marketing Attribution: Definition and Models

The user goes to the website, sees the ads and immediately buys the product. That’s what every business wants. Unfortunately, this rarely happens — 98% of people do not visit a company’s website for the first time to make a purchase. On average, a purchase requires seven touches, and among these touches, each channel will more or less influence the user's decision. How do you know which channels are the most effective? Select and use a marketing attribution model.

What is Attribution in Marketing

Attribution modeling is the distribution of value (weight or points) across the channels that led to the conversion. Attribution shows the customer journey from the first touch to the desired action.

What does it mean?

Let’s take a look at John. He saw an ad for sneakers on Facebook, became interested and clicked on the ad. On second thought he decided not to buy it yet. Later, John was shown a banner ad for the same sneakers, but this time on Google. He went to the website again and even added the product to the cart, but forgot about it. After 5 days, he saw an advertisement again. This time on Instagram. He even took a screenshot so as not to forget and look later, but he did not click on the advertisement. Later that evening, John found a model of the sneaker on Google and finally bought it.

Which channel brought the client?

If you think it was organic search, then this is one type of attribution. But what if it was the first touch on Facebook ads? Then it is another type. Evaluating all channels equally is the third one.

Attribution is a tool that helps to track touches with a brand and understand where to invest money, and on which channel the budget is not worth spending.

Also, with attribution models, a conversion window is set — the period of time between the first touch and the purchase. There may be options: from 1 day to 28 or more. It depends on the length of the sales cycle.

Let's say we are launching a campaign to sell a certain brand of cars. In this case, the attribution window should be at least 28 days, because hardly anyone decides to buy a car spontaneously in 3-7 days or less.

How to track customer touches with a brand? Different platforms collect only their own data: Facebook, VK, Google services and others. If you want to track what result each channel brings, you can use a tracking pixel. It is a piece of code that collects information about user actions. All you need to do is to add this code to the pages of your company website.

Why You Need to Use Attribution in Marketing

The customer journey to purchase, where he saw and bought, is rare, especially if the product is complex and expensive. After the first acquaintance, your customer goes through several stages: clicking on a banner ad, subscribing to a newsletter, searching for a product and information about it on Google. If the company does not know how the customer actually comes to the purchase, then it risks investing the budget in the wrong channels.

How marketing attribution helps businesses:

• The ability to properly distribute the marketing budget between channels and reduce the risk of losing money, as well as increase ROAS and ROI.
• Better personalization. The more you know about your customers, the more likely you offer them what they need and do this using the right channel — the customer-friendly one.
• Building a relationship with a client. Relationship marketing is the key to long and profitable relationships with users or partners. This type of marketing has replaced the paradigm of fierce competition and conflict. The built relationship with the company retains the customer and helps to personalize purchases and services.

Types of Attribution Models

When it comes to marketing attribution, one size doesn’t fit all. The same model will not work for every business. For example, for the real estate industry, counting the last click is a bad idea. It can take several months from the first touch to the purchase of an apartment. Let's look at different types of attribution models and discuss their pros and cons.

First Click Attribution

The first interaction with the user, such as a click on a banner ad, gets the most value per conversion. It is estimated that the platform, where the first touch took place, brought the desired action. The rest of the channels are not taken into account.

Pros:

• Easy to understand and implement.
• Opportunity to evaluate brand awareness campaigns and demand generation.

Cons:

• It's not an advanced model, there are optimization limitations. In other words, we always evaluate only the first channel. Is it effective or not? Should we increase the budget for it or limit it?
• “Accurate” and “impartial” — it is not about this model, because First Click Attribution does not take into account the merits of other touches.

Last Click Attribution

The main value goes to the source that the user visits before making the desired action. Last click attribution has long been the most popular and even now it is the default setting in Google Analytics.

Pros:

• A good option for tracking purchases that happen quickly and spontaneously.
• The model is accurate and easy to evaluate. For example, if users clear cookies, all data will be lost. But with the last touch attribution model, it’s not a deciding factor, because the period between the last click and the moment of conversion is short.

Cons:

• Other types of interactions are not counted, so you might draw the wrong conclusions about your campaign.

Last Indirect Click Attribution Model

This model is similar to the previous one, but it gives more value not to the click, but to the last marketing action before lead conversion.

Pros:

• The model filters direct traffic that may have been the result of campaigns providing more informative and accurate data.

Cons:

• As with last-click attribution, we can’t track the customer journey.

Linear Attribution

The value is distributed across all touchpoints. The linear model seems more impartial than last-click and first-click attribution, but it remains controversial because all channels cannot be equally effective.

Pros:

• Simple but more advanced than single-channel models.
• A balanced view of your entire marketing strategy.
• It will show which channels have value — it will display each channel that participated in the path to conversion.

Cons:

• Although we can see all the channels that participated in the conversion, we will not be able to find out which one worked better. In other words, the model is useless if you need to distribute the budget for each channel in terms of efficiency.

Time Decay Attribution

More weight in this model belongs to interactions that were closest to the conversion. The first touches are less valuable.

Pros:

• Scientific and logical model.
• Suitable for transactions with a long sales cycle, such as purchases in B2B or in the real estate industry.
• Helps evaluate customer relationship building campaigns.

Cons:

• It is almost impossible to define which channels brought the user first.
• The risk of overestimating the merit of a channel. The last visit before buying is not always decisive.

Position-Based, or U-Shaped Attribution

The first and last touchpoints receive 40% of the value each, while the remaining 20% is divided between any touches that occurred in the middle of the funnel.

Pros:

• A powerful tool for the businesses that initially have a plan for multiple touchpoints before buying, rather than fast conversions.
• The most important channels receive value: the user's first acquaintance with the product and the last touch that motivates him to buy.

Cons:

• It doesn't consider marketing efforts beyond lead conversion. The first touch is not always successful. For example, the user doesn’t always understand what you are selling after the first touch. In this case, the following touches will introduce him to your product. Sometimes the actions in the middle of the funnel are more significant: notifications about a new price, newsletters about new products, perhaps, motivate them better than direct advertising.

Data Driven Model

It is an algorithmic attribution model that uses machine learning to evaluate the customer's actions before he makes a purchase. Algorithms choose the most effective ads, keywords, and campaigns. In this attribution model, the patterns that led to the purchase or, conversely, prevented it, are found and compared.

Pros:

• The model is advanced and provides accurate data.
• Opportunity to assess all channels that you use to communicate with the customer.

Cons:

• It’s difficult to set up and work with it. You’ll have to find powerful software and competent analysts.

W-Shaped Attribution

The model looks like a U-Shaped attribution, but this time it’s not only extreme points that get weight. We have an additional touchpoint. It’s a touch in the middle of the path, where there is a possibility that the client will become a buyer. For example, a company launched a mailing list, held a private sale, and so on. These points receive 30% each, and the remaining 10% is divided among other touches in the funnel.

Pros:

• Evaluating potential touchpoints that could have induced a customer to purchase in the middle of the journey. It gives you an opportunity to optimize a campaign, add or exclude broken channels.
• Suitable for lead generation campaigns.

Cons:

• It is not always clear how the customer met the brand in the middle of the journey, 10% of the value for all other touches is negligible.
• Complexity of customization.

Full-Path, or Z-Shaped Attribution

The Z-shaped model takes the concepts of the W-shaped model. So we have the first touch, the last one and the stage of the conversion opportunity. But this time you get a touchpoint where the user becomes a lead: he subscribes to the newsletter or sends his email in exchange for something (usually content). 22.5% of the value is divided between four touchpoints, and the remaining 10% — between the rest.

Pros:

• Accurate and complete tracking of the customer's path to conversion.
• More opportunities to define strategies that work.

Cons:

• Difficulty in implementation. The model is only suitable for companies where marketing and sales are fully synchronized, and marketing deals directly with existing sales opportunities.

Custom Attribution

A model works for companies with well-built marketing and sufficient resources: technology and competent analysts. The value is assigned manually for each interaction, but only after deep analytics of a large amount of data.

Pros:

• A complete picture of the customer journey with all the nuances.

Cons:

• This model is for advanced teams because it’s difficult to set up.
• It requires a lot of data to analyze, which takes time. Therefore, the model will pay off for a large sales cycle.

Conclusion

Working with attribution models requires resources: time, team knowledge, advanced software. Used correctly, the attribution model will become a reference point for the company. A successful business knows not only the portrait of its client but also his path to purchase, which means it gets more chances to offer its product or service to the customer at the right time and in the right place.