We’re in a challenging moment for almost every organization, and understanding the impact of every dollar spent has never been more important. Attribution modeling allows marketers to analyze how much credit each marketing channel and customer touchpoint should receive for a given conversion. Over time, this approach enables marketers to optimize their programs toward the channels and touchpoints that drive the most value. However, all attribution is not created equal, and different attribution models vary in terms of how they distribute that credit.
In order to realize value through attribution, you need to understand which of these models will work best for your business. To help, we’ve laid out the six most common attribution models and when to use them. There’s no universal attribution model, and the one that works for you will be determined by your customers’ unique buying habits and your goals.
What are the different types of attribution models?
Sometimes referred to as “first-click” or “first-interaction,” these models assign 100% of the credit for each conversion to a single source. In keeping with the name, it’s the first recorded interaction a customer has with your business that’s credited with driving the ultimate conversion — no matter how many additional touchpoints fall in between. For example, if a customer finds your site through organic search, is subsequently retargeted with display advertising on a publisher site or a social platform, and then makes a purchase, it’s still organic search that gets all the credit.
First-touch models have the advantage of being simple. Rather than distribute value across multiple channels, they give all credit to the first point of contact. They can be an easy way for an emerging marketing program to get a sense of how its consumers behave and where they are best reached to fill the top of the funnel. It’s also an effective model for products with a short buying cycle where potential customers don’t typically have more than one interaction with the brand prior to purchase
Like first-touch attribution, last-touch attribution models assign full credit for conversion to a single interaction, in this case the most recent interaction an individual had with your business. Whether it was clicking an ad, receiving an email, or interacting with a social media post, the last thing a customer does before they convert is assumed to be the only thing responsible for driving that conversion.
Last-Touch attribution has the advantage of being simple to implement and evaluate on an ongoing basis, so it’s a good choice for businesses that need to gain some basic insight into consumer behavior to better understand their funnel. It’s also an effective model for businesses with a short buying cycle where the considerable lag between exposure and purchase is fairly small. However, last-touch doesn’t account for a lot of the complexity inherent in modern multichannel digital marketing, which exposes consumers to many messages that contribute to their ultimate conversions.
Last Non-Direct Touch Attribution
If your business has anything but the shortest and simplest of buying cycles, then last-touch and first touch may not be enough to really help you understand the effectiveness of your various marketing channels. Used as the default attribution model for Google Analytics reports, last non-direct touch assigns no credit for conversions to direct traffic, such as when someone clicks a bookmarked link or manually enters a URL to reach your site.
Like simple last-touch and first-touch attribution, this assigns 100% of the credit to a single interaction, but it doesn’t consider direct traffic to be an attributable channel. However, like simple last-touch attribution, it also doesn’t assign any credit to other interactions that may have led up to that final interaction, making it difficult to understand the impact of your whole multichannel marketing program. So this model is best used for products with a short buying cycle. By excluding direct traffic, it focuses on marketing elements you can control or influence, such as clicks from paid and earned channels.
Single- vs. Multi-touch Attribution Models
The first- and last-touch models can be good places to start, particularly if you are able to track and report on deduplicated conversions in a single, integrated data source. However, If your enterprise’s buying cycle is more complex — meaning that customers normally interact with your brand several different times along their conversion journey — then a multi-touch attribution model is required. The below examples distribute fractional credit for a given conversion across the brand interactions that preceded it, within a predetermined lookback window.
If your enterprise’s buying cycle is more complex — meaning that customers normally interact with your brand several different times along their conversion journey — then a linear attribution model gives insight beyond the first or last interaction. Unlike the previous models we’ve discussed, linear attribution can account for multiple touchpoints on the road to conversion, but it weights all touchpoints equally. If a customer interacts with a Facebook ad, a display ad, and a paid search placement before they buy, then the credit for that conversion is split three ways.
Linear attribution can help you to understand the impact of your entire marketing campaign across multiple channels. However, it’s limited in that it can only split credit equally between all touchpoints. In reality, it’s rarely the case that every interaction with your brand is equally important in driving a conversion. Some channels and messages are usually more valuable than others, and this model can lead marketers to overvalue some channels and undervalue others based on the egalitarian way that it splits credit. Still, it’s a helpful way to begin to demonstrate the value of a multichannel or multitouch marketing strategy.
Time Decay Attribution
To get closer to understanding the actual value of the various interactions leading up to a conversion, you need to consider not just the order in which those interactions occurred but also their timing in relation to the actual moment of conversion. A time decay attribution model does this by distributing credit across multiple events, but assigning more credit to those closer to the time of conversion, based on the hypothesis that those events had a greater impact on the decision to purchase.
This model is particularly effective for businesses with high-consideration products where building a relationship over time is a critical part of the sales cycle. However, time decay attribution gives less weight to top-of-funnel marketing activities, something that should factor into your thinking when you assess the efficacy of your full marketing funnel using this model.
Position-based attribution models assume that the two most important interactions a customer has with your business will be the very first and the very last before conversion. Following that hypothesis, position-based attribution assigns fixed credit for every conversion to the customer’s first and last points of contact with the brand prior to conversion. The remaining credit is shared equally among any interactions that happen between those two points. The default* position-based model in Google Analytics assigns 40% to the first and last interactions, and splits 20% evenly among all those in between.
This model works well for businesses that expect their prospects to have multiple interactions with their brand prior to purchase. It captures the impact of top- and bottom-of-funnel activities, both of which are critical for businesses with longer sales cycles, and it also assigns some value to other marketing activities that keep prospects warm, renew interest, or further existing engagement.
* The 5 models described above are the baseline models in Google Analytics, that can be customized in a multitude of ways, including the # of days in the lookback window, user engagement type (length of session or # of pageviews), time decay half life, and custom % allocations for the Position-Based model.
Algorithmic or Data Driven Attribution
Algorithmic attribution models, called Data Driven or DDA in Google Analytics, use machine learning to dynamically assign fractional credit across multiple interactions leading up to a conversion event. They work best with a high volume of interactions and conversions. For example, the minimum requirement in Google Analytics is at least 600 conversions per month. While algorithmic models help take the guesswork out of deciding which model fits the needs of your particular business, they are also less transparent than their rules-based counterparts.
In theory this modeling approach provides the most reliable predictive value, but this is of course dependent on data volume and quality. Assuming this is the case, an algorithmic model will apply to the widest variety of business models, media strategies, etc. Whether your products have a long sales funnel or are more of an impulse buy, the model will adjust to reflect this landscape. This capability often comes with a price tag. In Google Analytics multi-channel funnel (MCF) reporting, data-driven modeling is only available with a 360 (paid) license. However, the latest beta Attribution reporting feature does offer free data-driven attribution, excluding direct traffic.
Finding the right model
If you already have a lot of insight into your sales cycle, then you may already know that one of the models outlined above will work for you. However, many businesses have unique funnels with particular needs that may not be fully met by a standard attribution model. Often the path to a custom attribution model that fits your business goals is a gradual one. Our marketing attribution experts can help you to find the right model to drive ROI for your business. Reach out today for a consultation.
how we can help you.