Note: For a new version of this blog post, check out The Pros and Cons of Attribution Models: A 2021 Update.
Understanding and demonstrating marketing ROI isn’t easy. It’s a challenge that’s long plagued marketing organizations and one that remains challenging given how much easier it is today to collect marketing analytics than it’s ever been before.
A big part of that challenge is simply not knowing where credit should be due. If a customer has 13 touch points with a company before deciding to make a purchase, how do you determine the role each one played in earning that sale?
Revenue attribution is the relatively recent practice stepping in to answer that question. A number of different revenue attribution models have been developed with the goal of determining the best way to assign weight to each touch point customers make.
Revenue attribution may not be actively deployed in every marketing organization yet, but it’s increasingly in use by the ones that really care about tracking ROI.
Ross Graber from Sirius Decisions recently shared the results of a small survey he did to figure out how many businesses are currently implementing the practice. He found that while nearly half aren’t using revenue attribution at all, 31% already are, and 21% are in the beginning stages of doing so.
Marketers are clearly interested in seeking out a better way to calculate the ROI of their activities. Revenue attribution is one of the strongest solutions available to do so, but it’s not one-size-fits all.
The 8 Main Revenue Attribution Models
There are eight main revenue attribution models in use by different businesses, each with their proponents. Here’s a look at the benefits and weaknesses of each.
First-touch attribution is one of the more basic revenue attribution models businesses use. It’s based on the assumption that the most important and valuable marketing activity is the one that makes your prospect aware of the brand to begin with.
There’s a certain logic to this: no sale ever gets made if a business doesn’t know you exist. But it’s not widely used, nor is it even the most popular single-touch attribution model out there. The first touch point for B2B customers is usually several steps away from an actual sale, which makes it seem less powerful in getting customers to that point than the touch points that come later in the process.
First-touch attribution is like crediting a first date with a marriage. Most marriages require a long period of time together beyond that first meeting, not to mention things like compatibility and shared goals. In addition, the vast majority of first dates don’t actually lead to marriages to begin with. Which brings us to our pros and cons.
Pros of First-Touch Attribution
- The first touch point is, undeniably, very important. A lead has to hear about you before they can possibly consider buying from you.
- If raising awareness is the primary challenge and goal of the campaign you’re evaluating, then that first touch is a key metric to capture.
- First-touch attribution is fairly simple to track. There are no complicated calculations or arguments about proper weight distribution.
Cons of First-Touch Attribution
- It only tells a small portion of the story, and a part that many people don’t even consider most important.
- Almost every customer that makes that first touch then has a series of further, nurturing touch points before they get to the point of sale that play an important role. First-touch attribution fails to give them credit.
- Being easy makes it overly simplistic, does few favors if you want to achieve an accurate picture of ROI.
Last-touch attribution is the other single-touch attribution model commonly employed and the most popular method on the list. It has the simplicity of the first-touch model, but shifts the credit entirely to the last action a prospect takes before reaching the point of sale.
Like the first-touch attribution model, this also makes intuitive sense. The step that gets people to buy is clearly important to earning revenue for the company. The numbers for first touch points are probably huge, but most of those leads fall off somewhere along the way. The most important ones are those that make it to that last step, which is why that last touch point is considered the most important by many.
Using the same marriage analogy, both people may have gone on tons of first dates and probably spent a fair amount of time in the getting-to-know you phase with other people – but the one that makes it to the wedding day is clearly the most important (unless there’s a divorce later, but that’s a whole other story). That last step is the one that really counts.
Pros of Last-touch Attribution:
- It highlights the step that most directly leads to revenue: the sale itself (or at least what immediately precedes it).
- If conversions are the primary goal of the campaign you’re evaluating, then the last touch point is the most logical and useful one to measure.
- It’s easy. As with first-touch, there are no complicated calculations or arguments about proper weight distribution. And it’s the model Google Analytics defaults to showing, so it’s the one most people who haven’t made a conscious decision about what revenue attribution model to use employ by default.
Cons of Last-touch Attribution
- It ignores everything that came before. The first touch, the lead nurturing campaign that slowly helped convince the lead to buy – all of that played a role in the sale, but gets ignored when it comes time to allot credit.
- Not all leads that reach the point of that last touch take the same path to get there, so there are lots of versions of the buyer’s journey you’re ignoring if you only look at the last step.
- As with first-touch, because it’s so easy, it’s overly simplistic. You can’t depend on last-touch attribution if you really want to get an accurate picture of ROI.
Evenly Weighted Attribution
Sometimes called the linear attribution model, this is where we take a step up from single-touch attribution to multi-touch methods. This is a key distinction because it means we finally stop ignoring all those touch points in the middle. If your leads are often taking 7, 10, 15 or more steps before getting to the point when they’re ready to make a decision, all of those steps play some role in getting them to that point.
The challenge then is to figure out how much credit each one of those 15 touch points deserve. The easiest answer is what gets us the linear attribution model: just give them all the same amount of credit!
The math is easy, you know you’re seeing ROI with more accuracy than with either of the single-touch models, and you can move on with your day.
Now for the downside – most touch points aren’t created equal. We’ve already addressed how intuitive the idea is that the first and last touch points are especially important in making sure people become customers. Do they really deserve the same amount of credit that all the others do?
Pros of Evenly Weighted Attribution:
- It gives everything credit. You’re no longer leaving important parts of the story out.
- It’s more complicated than single-point attribution models, but still fairly easy to figure out.
Cons of Evenly Weighted Attribution:
- It’s still not ideal for accuracy. You know some touch points play a bigger role than others. You just do. So why act like they’re all the same?
Time Decay Attribution
This is another multi-touch attribution model, but one that aims to get at the issue of acknowledging that different touch points have different value. The idea behind time decay attribution is that each step in the process gets your prospect closer to that sale, making each one slightly more valuable than the last.
This model, like some of the others, makes a certain amount of intuitive sense. Lots of people make that first touch point, a good amount make it to a second or third, but the numbers dwindle as you get up to touch points five, six, and seven. Each one brings the prospect that much closer to being one of the special ones that becomes a customer.
It’s still not perfect though, as the sequence itself doesn’t always tell you how important the touch point is to making a decision. If a lead sits through a demo the week before they buy, but click on a link in an email to read a blog post the next day, is that blog post really doing more heavy lifting in earning the sale than the demo?
Pros of Time Decay Attribution:
- Like the linear model, all touch points get credit. That’s good!
- The time decay model recognizes the important point that certain touch points make more of a difference than others.
- Those later touch points are doing the work of getting leads ever closer to the ultimate goal: earning you more revenue. As such, it makes sense for them to get more credit.
Cons of Time Decay Attribution
- It’s still easy to argue about accuracy. Does the sequence really tell us which touch points are most important to convincing a lead to become a customer?
- We made an earlier case in this post for the importance of the first touch point, this model gives it the least credit, even though many would argue that initial awareness is one of the hardest and most important goals to achieve.
If you combine the two single-touch attribution models with evenly weighted attribution, what you get is position-based attribution. This model awards the first and last touch points the most credit (the specific percentage is up to you), and then divides the rest of it evenly between all the touch points in the middle.
If you believe, as many marketers do, that the first contact a customer makes with a brand and the last action they encounter before purchase are the most important, then this model makes the most sense.
While clever, we’re still not in attribution utopia. All those touch points in the middle may not actually be equal in the influence they have on a customer. And you can definitely find marketers happy to argue that the first and last touch points shouldn’t be given such a special place in the marketing hierarchy, but the model addresses a lot of the weaknesses of other models on the list.
Pros of Position-based Attribution
- Gives all touch points credit, without assuming all are equal.
- Provides awareness and decision marketing activities more credit than consideration-level ones, which matches how many marketers allot value to the different stages of the buyer’s journey.
- Combines the best of the first-touch, last-touch, and evenly weighted distribution models.
Cons of Position-based attribution:
- The math is more complicated. Many people go with 40% for the first and last touch points, with the 20% left over divided amongst the rest, but your organization may have team members that argue the distribution should be different.
- It treats all of the touch points in the middle as equal, although in practice some likely play a bigger role than others in helping you win the sale.
If your organization uses a demand waterfall model, then you may be thinking that the action that shifts your lead over from a marketing qualified lead (MQL) to a sales qualified lead (SQL) really needs to get more credit than these other models are giving it. W-shaped attribution has your back.
With the W-shaped attribution model the first touch point and last touch point are still given the most credit, while the touch point where a contact converts to a lead (which usually falls somewhere around the center, hence the W) is given extra credit as well. The rest is divided evenly between the remaining touch points.
This answers the issue of acknowledging that not all those middle touch points should be given equal weight. Yay! But it still oversimplifies what the other middle touch points – those that don’t make the middle of the W – contribute.
Pros of W-Shaped Attribution:
- It gives all touch points credit, while recognizing that some should bear more weight than others.
- It makes sense for businesses that use the demand waterfall methodology.
Cons of W-Shaped Attribution:
- It’s more sophisticated than some other models, but still oversimplifies the value of different activities beyond the big three that are given extra credit.
So we’ve pretty well established by this point that the biggest challenge of revenue attribution is figuring out how to decide which touch points are worth more than others. For many marketers, it makes sense that the actions people take that demonstrate the most engagement are those worth the most.
Handing over contact information in order to participate in an hour long webinar is a more active move, as well as one that shows a higher level of commitment, than watching a video or clicking on a link. Interaction-based attribution thus provides more credit to the touch points that require more action on the prospect’s part.
Interaction-based attribution requires some input from marketing team members to determine which touch points should be considered higher-engagement ones, and which should be treated as more passive ones. The model then assigns different weight to the actions based on that determination.
Pros of Interaction-based attribution:
- It gives all actions some credit, without assuming they all have the same value.
- It gives greater weight to the actions that show a greater level of engagement, which likely correlate to a higher likelihood to buy.
- It gives marketers the chance to weigh in on which touch points they feel are higher value than others.
Cons of Interaction-based attribution:
- It requires subjective analysis from marketers, which opens the door to debates and differences. Figuring out which touch points should get more weight than others is complicated and could cause dissension in your ranks.
- It’s based more on how people feel than what they know about how influential specific touch points are.
Statistically Inferred Revenue Attribution
Statistically inferred revenue attribution is a response to many of the weaknesses of earlier models. It ensures every touch point is counted, gives different weight to each, and does so based on the evidence provided by data rather than by assumptions or feelings (not that those don’t have some value to add as well).
It requires having access to the historical data of your marketing activities and campaigns and a tool to analyze the patterns in how well different activities perform. Despite these hurdles, statistically inferred revenue attribution gets us the closest to accurately determining which marketing activities should be worth the most when calculating marketing ROI.
Pros of Statistically Inferred Revenue Attribution:
- It gives every touch point credit, without assuming they all have the same value.
- It takes a lot of the guesswork and assumptions out of the process by using data on performance that’s collected over time.
- It’s the revenue attribution model most likely to provide (reasonably) accurate numbers for how much each marketing activity is worth.
Cons of Statistically Inferred Revenue Attribution:
- It requires historical context and the deployment of technology to parse the data effectively.
The exact approach to measuring success will undoubtedly vary from organization to organization. Nonetheless, selecting the most appropriate and exacting attribution approach will help you better measure what touches do the most to move the needle for measuring more accurate ROI.Icon made by Freepik from www.flaticon.com is licensed under CC BY 3.0