Table of Contents
Attribution is one of the most complex issues in eCommerce. Essentially, it’s trying to understand what drives a human to make a decision—which is nearly impossible, as people generally aren’t rational in their decision-making (APS).
So, what makes a consumer go from never hearing about your brand to making a purchase? You may never really know why, but you can understand how by tracking what marketing campaign they see and through which channel they see it. Sound simple? It’s not. Here’s why.
Why using a marketing attribution model is important
What is marketing attribution? Imagine you’re sitting behind the counter at your store on Main Street. You can see all of your customers as they walk in your door. Where are they coming from? You could let them tell you. But ideally, you’d step outside onto the street to see for yourself.
Knowing where your customers have been before they purchase is critical to figuring out what marketing channels are most successful for you. Relying too heavily on vendor-reported data (e.g., sitting behind the counter), instead of also doing your own tracking using a third-party analytics platform like Google Analytics or Daasity (e.g., going out to the street and verifying), can result in a skewed picture of where and how customers are interacting with your marketing before they make a purchase.
Using an attribution model provides you with a standard way to look at the performance of your different marketing channels, which you can then use to help you think through a decision process about where to spend. It also helps you take vendor data (e.g., Facebook attribution) with a grain of salt. At Daasity, we recommend approaching vendor attribution with “pragmatic optimism”— you hope everything is performing well, but it’s probably not performing quite as well as the platform says it is.
Without data from multiple sources and an attribution model that helps you compare the effectiveness of different channels, you won’t know what an order really cost you. It’s also harder to clearly justify where you’re spending your marketing dollars.
There’s no one “right” answer for eCommerce attribution
There are multiple ways to approach marketing attribution. By tracking attribution, what you’re doing is assigning value to a marketing channel that results in a purchase. However, no single interaction is usually 100% responsible for a conversion. This means that when you do marketing through multiple channels, it can be really hard to determine which interaction had more
A customer purchased on March 15. However, they first came to your site on March 1 through a Facebook ad, then back again on March 4 through a PPC ad. On March 10, they typed your URL directly into their browser but didn’t purchase. On March 15, they came back to your site directly and finally purchased.
Common attribution models
To get a clearer picture of which marketing channels/interactions are resulting in conversions—for example, when you have a situation such as the above scenario—you use an attribution model to analyze your data. In general, there are a handful of different methods that are most commonly used for attribution, including first-click attribution, last-click attribution (also called first-touch and last-touch attribution), time-decay attribution, and linear attribution—check out this Daasity blog post for descriptions of each.
First-click attribution and last-click attribution are the most straightforward attribution models and can be a good, simple way to start figuring out attribution. For example, querying which channels customers interacted with first (“first click”) could help you analyze the
However, as mentioned above, a purchase typically results from multiple interactions through various channels. Therefore, if you only look at last-click attribution, for example, you won’t know if there’s another channel that frequently contributes to conversions, might be influencing the journey to purchase, and could be a valuable channel to shift more money to.
This is where a multi-touch attribution model comes into play. A common and relatively straightforward approach is a linear model, in which you allocate the same amount of credit to each channel. For example, if there were four “touches” before purchase, you would assign a 25% “credit” to each of those channels for bringing in 25% of the revenue:
In general, it’s informative to look for patterns in your overall marketing mix. For example, you may attribute a last click to Facebook, but notice that it’s frequently connected with email and that they result in more conversions when paired together.
There’s no one right way of doing attribution. And it will never be exact. But knowing about the different models allows you to apply what you think will help guide you to make smarter decisions about your marketing.
Determining value for money
While an attribution model can help you understand which channels are bringing in sales, it’s a good idea to also compare performance across channels – use the metrics in your six-pack, including ROAS, CPO, and CPA. For example, if you attribute 25% credit ($25 revenue) to a social media ad, but the ad cost you $20, you might determine you’re not getting enough value from that channel and shift more budget to a better performing channel.
Accuracy in attribution: 6 best practices
We recommend six best practices that can help to make your attribution insights more accurate.
- Get your UTM codes in order
Use standardized naming conventions for UTM codes (snippet of code added to the end of a URL to track the performance of campaigns) in all your marketing as a first step – even before you worry about your attribution model and analytics. As with most things data-related, the output is only as good as the input. In other words, if you aren’t adhering to consistent parameters for UTM codes, it makes it really difficult to track where visitors to your site came from and which campaign they saw. We find that although most brands use UTM tracking, they often don’t use it in a consistent manner.
- Use a third-party analytics platform
It’s helpful to use an analytics platform like Daasity as a third-party referee when comparing marketing channel performance. Vendors tend to try and make their data look as good as possible. Therefore, it’s important to use your own data to do a reality check. Know that when you do this, you’ll likely see different ROAS or CPO numbers.
- Use discount codes
For brands that do direct response marketing via various channels, discount codes can be an effective way to track attribution especially for channels that don’t have a direct click such as radio ads and podcasts.
- Use surveys
When you want to know, ask. It’s a good practice to help validate your online data by administering a post-purchase survey to ask customers how they found you. Enquire Labs is a great tool that allows you to install a post purchase survey on your Order confirmation page, asking “How did you hear about us?” You can blend this data with your click attribution data to get a more accurate view of where customers are learning about your brand.
- Set the right attribution window
The attribution window you choose depends on what makes the most sense for your brand’s sales cycle. For example, an attribution window for a product or service with a two-month sales cycle would look very different from a product or service that consumers typically make a quick, two-day decision about. Having a logical attribution window enables you to track the multiple interactions over a defined period of time that led to the purchase.
- Get agreement on the model
Everyone will have an opinion about how to do attribution. However, it’s important to come to an agreement on the attribution model and have everyone use the same numbers so that you’re comparing marketing performance apples to apples. If the team doesn’t trust the numbers, it will be difficult to come to a consensus on what to do next.
Start simple – but start
If you’re just getting on the marketing attribution bus, a good rule of thumb is to keep the model simple while your business is relatively simple. You can get more sophisticated over time by adding other data points like post-purchase surveys and discount codes to verify the digital signals. Whether you choose first click, last click, linear, or some other model, remember, there’s no magic option that will give you end-all, be-all accuracy. However, by choosing a model and applying some best practices, you’ll benefit from a great quantitative tool to help guide your decisions about where to invest your marketing dollars.
