Customers can go through several touchpoints before they buy a product. In marketing, a touchpoint refers to a specific marketing channel a customer interacts with on their customer journey. For example, a customer sees a sponsored post on Instagram about a product (awareness stage). Then they see a Pinterest post that gets them thinking (consideration stage). Finally, they come across another post on Facebook that prompts them to visit the website. So, they make a buy (conversion stage). Here, the sponsored Instagram post, the Pinterest, and the Facebook post are touchpoints. So which touchpoint was the deciding factor in your customer’s decision and conversion? This is where attribution modeling comes in.
What is an attribution model?
Each touchpoint affects a client’s decision to buy a product from you. Hence, all touchpoints a customer comes across influence their decisions. But, there is no way to say for certain to what degree a touchpoint influences the customer. Hence, we use attribution modeling to delegate points to each touchpoint. Thus, you can define the attribution model as a set of rules. These rules help you determine the credibility of each touchpoint in the conversion path of a customer. Attribution modeling is an excellent analysis tool. Especially for companies that use multi-channel marketing strategies. With it, you can evaluate the impact of each marketing channel.
What are the types or models of attribution modeling?
There are two main types of attribution modeling. We classify attribution modeling based on the number of touchpoints it analyses.
Single-touch attribution modeling
The single-touch model is the simplest of the two. It assigns value or credit to a single touchpoint in the entire conversion journey of a customer. There are many types of single-touch attribution models. However, there are two that stand out.
The first-touch and the last-touch model are the most popular for obvious reasons. The first touch is popular because it initiates the awareness stage. You can not move a customer along the funnel if there aren’t any customers in the first stage. People usually consider the last touchpoint as the tipping point, as it is the point of conversion.
There are several limits to the single-touch model. The model is not comprehensive enough since it only uses one touchpoint. Customer journeys are complex, so you can not define them by a single touchpoint. Arguably, any touchpoint between the first and last has a higher impact than the others. Therefore, such types of attribution modeling have many limitations.
Multiple-touch attribution modeling
The multi-touch model factors each touchpoint a customer interacts with throughout their journey. It is a complex model, much like most things in marketing. However, multiple-touch models are far more accurate than single-touch models. The crediting systems correlate each touchpoint’s value to its importance in the buying journey. Therefore, the model is comprehensive. Plus, it offers richer insight into marketing strategies and customer journeys.
Types of single touch models in attribution modeling
First interaction model
It is also called the first click or the first touch attribution model. In this model, we assign the entire credit to the first point of contact a customer makes. Companies use this model to determine which channels catch the customer’s attention. The first touch does not lead to a conversion directly. But they are crucial in marketing since it introduces the customer to the product/brand. The first interaction models are excellent for new companies trying to get a feel for customers in the awareness stage. Additionally, brands with short buying cycles can invest in first-touch attribution models. Since such types of attribution modeling are simple and effective in such cases.
Last interaction model
It is the complete opposite of the first touch attribution model. In the last interaction model, you give the whole credit to the last touchpoint a customer has interacted with before they make a purchase. In this model, people consider the last touch point to be the tipping point. Thus, it is the only touchpoint responsible for a conversion of a customer in this model. Much like the first touch model, it is simple. Yet effective in some cases. Especially so, if you want to know which campaigns have the highest conversion rates. For example, companies that have short impression-to-purchase durations.
Lead Creation model
The distinction between a prospect and a lead is significant in marketing. What’s the difference? A prospect may or may not buy from you. However, a lead is most definitely on the lookout for your product. The difference is thin, but it can help you tailor your strategy to fit better. We give this tipping point 100% of the credit. Additionally, it will help you identify the touchpoints that tipped a potential client into a qualifiable lead. Such touchpoint includes signing up for webinars, opting in for a newsletter, or giving contact information for further marketing communication.
Last non-direct touch model
You may also know it as the last non-direct click attribution. The first and last interaction models do not help you understand the overall effectiveness of your marketing channels and strategy. However, it prioritizes a non-direct touch. A non-direct touch refers to the touchpoint when a customer clicks a bookmarked webpage or manually enters the URL. Such non-direct touchpoints are also imperative in the conversion path. It is also the default attribution model in Google Analytics. We do not consider direct traffic in this model since this is a single-touch attribution model. Plus, with this model, you can monitor elements you can control or influence, such as paid advertisement clicks and other earned channels.
Types of multi-touch models in attribution modeling
Linear attribution model
A linear attribution model factors for all touchpoints in a customer’s journey and assigns equal importance to all of them. This attribution model is one of the most popular multi-channel models. Let’s take an example to better explain how this model works. For instance, if a customer goes through 20 touchpoints before they make a purchase, then each touchpoint gets 5%. Or a customer goes through 10 touchpoints, hence you allot each touchpoint at 10%. It is a relatively easy multi-channel attribution model. By considering all touchpoints, you can start optimizing your strategy for each of those touchpoints. This model has a higher degree of accuracy compared to single-touch models. However, with this method of attribution, you lose the data that is required to optimize for specific channels. Additionally, you are allotting the same value to low-value and high-value points in this method.
Time decay attribution model
In the time decay attribution model, we assign the most credit to the interaction that resulted in the customer’s conversion. All touchpoints leading up to the purchase receive a certain amount of credits. The further back a touchpoint is the lesser value it receives. This model accounts for all touchpoints but still prioritizes the touchpoint that converts the customer. While this attribution model is excellent for optimizing conversion, it has one limitation. It does not give enough credit to touchpoints that occur in the awareness stage. Arguably, touchpoints that introduce a customer to products are as important as the converting touchpoint.
Position-based attribution model
This model is a mix of linear and time decay attribution models, taking the best parts of each. In this model, we give the first and the last interaction 40% of the credit each. The rest of the 20% we disperse between the rest of the touchpoints. Position-based attribution accounts for the introductory and converting interactions heavily since we considered them to be the most crucial points in a customer’s journey. However, blindly assigning a big chunk of the credit to any (or in this case, two) points can be detrimental to value calculation. The first interaction can be a low-value interaction, like emails, or a high-value point, like a paid search ad. Assigning both the same value makes no sense from a marketing point of view.
How do you select the right model for you?
From studying the models, we know that each has a different method of evaluating touchpoints. We use each model for a specific reason, such as analyzing only conversion points or first touchpoints points, etc. The conversion value of the marketing channel will differ depending on the attribution model employed. The goal of using attribution models is to optimize a part of your marketing channel or the entire marketing strategy. Therefore, formulate an objective before you dive into selecting an attribution model. The objective will act as a guideline when you have to choose a model. Since you are looking for specific areas to optimize, select touchpoints that bring the most value to your business. Hence, select an attribution model that accounts for your most crucial touchpoints and that works well for your business. You should also keep in mind that attribution models are not perfect. Each of them has as many limitations as there are advantages. So, do not expect that a single model type you cover all your bases.
What tools can you use in attribution modeling?
Many vendors supply attribution modeling software. Each software/tool will be different. You will only know what works best for you after you try them out. Here are the ten most popular attribution tools:
- Google Analytics
- Adobe Analytics
- Active Campaign
Advantages of attribution modeling
Optimize your budget for marketing channels
Attribution models help you assign credit to marketing channels. You do this by analyzing the performance of the marketing channel over time and the conversion value it has. Hence, you have data from an analysis model that will help you determine your budgeting options. Hence, you are equipped to decide the marketing budget for individual marketing channels depending on their inherent value. Therefore, you can invest more in channels that drive conversion and reevaluate marketing channels you can optimize. Plus, it will give you hindsight into your marketing channels’ ROI that you can improve with data generated from attribution modeling.
Provides insight into customer behavior
Customer behavior information is by far the most crucial data a marketing team can gather. With attribution models, you can gather and analyze data on individual channels, campaigns and ads. This gives you the data you need to understand customer behavior. Such insight includes how your customers interact with your ads and campaigns and which channels are popular among which demographics. Additionally, it helps you improve your product and strategy to better attract a wider base of customers. Plus, you can use this data to re-target your customers with marketing materials later.
Limitations of attribution modeling
There is no true ROI
The attribution model explicitly factors for online marketing methods since it is much easier to monitor. Additionally, you can not gauge the impact of office marketing channels. Hence, we do not include them in the attribution model. Such offline marketing channels include billboards, posters, pamphlets, and word-of-mouth recommendations. These can also influence a customer on their journey. When you are assigning credit and ROI for media channels, you are not accounting for sales from offline channels. Thus creating a distorted view of the ROI, the ROI of online marketing channels is not the true ROI.
Does not account for external factors
Similar to offline marketing channels, the attribution model also does not account for certain factors. These factors include discounts, the seasonality of purchases and products, promotions, and the economy. These factors are secondary, yet they have a significant impact on buying decisions. This is because a dip in pricing has a greater impact on sales than advertising. More often than not, we link discounts to certain marketing channels. Hence, the effect of promotions dilutes marketing analysis. Consequently, this results in unreliable data and impact factors on marketing channels.