For those of us who are looking to reduce marketing spend, multichannel attribution is a big topic. When you’re looking at which of your channels work best for you, it’s important to know which have the biggest impact on your overall sales.
However, multichannel attribution models come with their own set of pros and cons that can leave you asking yourself which is the best for your ecommerce store. I’m going to break each type of model down, because multichannel attribution is crucial to getting the biggest return on your investment.
What is attribution?
Attribution is the concept of assigning credit for a sale to the channel or advertising that a customer used. For example, if a customer comes to your site via Facebook and purchases from you, we could give Facebook some credit in pushing the sale.
Attribution is a bit of a tricky subject, because there are several different models to choose from, and most data and analytics tools don’t show you enough information to make an informed decision about it.
Attribution is also a crucial aspect in affiliate marketing, where many ecommerce merchants can credit a heavy amount of traffic and sales. However, given the challenges of giving proper credit in marketing attribution, many online sellers pay out too much, or even not enough.
Additionally, even when not using affiliate marketing, online merchants often spend far too much on the channels that just aren’t working for them. Choosing a correct marketing attribution model will show you exactly which channels to focus on and where to tweak your budget to get the biggest ROI.
Different Marketing Attribution Models, and How to Choose what’s right for your Ecommerce Site
As I mentioned, marketing attribution models are varied and complex. However, there are definitely some that offer a more accurate picture of attribution.
It’s also worth mentioning how many sites track visits. Each visit to your site is counted as a session once the site tracks cookies in your visitor’s browser. Unless a customer pays for a few orders across devices, these sessions will be unique for each device.
Because each session is unique, and there’s no way to tell if a session on mobile is the same user as a session on desktop.
If the customer uses the same email address on separate devices to complete an order, it can be tracked, but not otherwise.
This also throws a complicated wrench into multichannel attribution, as we can only look at attribution on one single device unless they get linked up. Keep this in mind as we look at different attribution models.
Last Click Attribution:
Last click attribution is the idea that all of the credit for the sale goes to the last channel that a customer used to visit your site and purchase a product.
While this is the most popular marketing attribution model, it also paints a very black and white picture that isn’t always representative of which channel had the biggest impact on sales.
In addition to last click is last non-direct click attribution. This functions basically the same way that last click attribution does, however it disregards all touchpoints where the source is direct traffic. While this is useful if the last click happens to be from direct traffic, it has many of the same flaws as last click attribution in that it takes no account of any other touchpoint.
For example, check out this order above: Paid Google Ads clearly had the biggest impact on this sale. However, using last click attribution, the credit would go directly to organic Google, which doesn’t represent the real customer journey at all.
While it’s true that organic Google search had the most significant impact on the purchase itself (as seen by the time spent on the site and the number of pages visited), the customer likely searched for the company’s name because they’d already decided to purchase thanks to the Google AdWords campaign.
Understanding the full picture of your customer journey map will show you exactly how your customer is behaving throughout their purchase. Unfortunately, last click attribution models fail to show you the whole picture.
First Click Attribution:
First click attribution is the exact opposite of last click attribution: the full credit of the sale goes to the first channel a customer used when visiting the site.
The thought process here is to attribute to the first channel that caught your customer’s attention and credit it with the full sale. However, like last click attribution, this has the same problem: every touchpoint in between is completely ignored.
Using the example above, while paid Google ads would get the credit for the sale (and perhaps rightfully so), this ignores that social media had an impact on the sale as well.
Linear attribution tries to take each channel into account by equally distributing credit to each channel that served as a touchpoint for your customer.
For the above example, there were 6 touchpoints between discovery and sale (that we know of). Ideally, each channel would get 16.6% of the credit, with Google paid ads getting three times that amount as it served as three separate touchpoints (so 50% of the credit).
While this model does take the full customer journey into account, it might not always be fair to give each channel equal credit. For example, the first Google paid ad resulted in only one minute of time on the site, with 4 pages viewed. The second one was a bounce- the customer didn’t navigate any further than the home page.
Direct had a significant impact, as the session lasted for 12 minutes and 8 pages were viewed. While Facebook only had 4 consulted pages over 2 minutes. The last channel, Google Organic, was the channel used to purchase, where 10 pages were consulted over 32 minutes.
Obviously, each channel had a different level of impact. Is it really fair to attribute 16.6% of the sale equally across them? And how do we put a value on the initial discovery versus the channel that ultimately brought the conversion?
Position based attribution:
The position based attribution model counts the first and last clicks with more credit, and distributes the remaining credit to the channels in between evenly. The typical position based model is 40% to both the first and last channels with 20% distributed to any channels in between.
While this solves a few problems that first and last click attribution had, it still has the same issue as linear attribution: while none of the channels are disregarded per se, some in between the first and last click might have had a more significant impact on the sale.
Time Decay Attribution:
Time decay attribution is when you assign a significant portion of the sale credit to the last channel, and less to channels the further away they are from the sale.
While time decay might be one of the best attribution models at mitigating many of the challenges that other attribution models face, it still isn’t necessarily perfect. After all, there is a much smaller amount of credit to the channel that was responsible for the initial discovery.
Like the others, it doesn’t take into account the amount of impact the channel had on the sale, nor the significance of the interactions that a customer had on your site.
The Challenges of Choosing an Multichannel Attribution Model for Ecommerce
Each of the above attribution models has its pros and cons. The biggest running theme is that these models don’t attribute enough credit to the channels that had the biggest impact on the sale.
While we can all argue that the first and last touchpoints are the most critical to a sale, it completely ignores retargeting efforts and any touchpoint in between. I think we can definitely argue that each touchpoint plays its part in the customer journey.
So how do we mitigate these challenges and choose an attribution model that gives correct credit to each channel?
Start by Mapping your Customer Journey
Understanding exactly how your customers interact with your ecommerce site and brand while they move through your sales funnel is critical for understanding how much weight you attribute to each channel.
For example, if most of your customers are discovering you through social media, you would add more credit to that since it’s essential for your overall branding. However, if social media plays a small role or has minimal impact on your customer journey, it wouldn’t make sense to start shifting large amounts of budget to it.
When you have a clear picture of the path your customers tend to take, you’ll understand how much each channel impacts purchase. Granted, not all customers will be the same, there will definitely be exceptions.
For more information, check out the article we wrote on perfecting customer journey mapping.
Look at Key Aspects of your Customer Journey
When choosing the right multichannel attribution model, it’s important to make sure that it’s right for your ecommerce store.
While I could sing the praises of each model over an over, those attribution models won’t work for you if they aren’t right for your customer journey map.
There are a few that seem to work a bit better, like linear attribution, position based attribution, and time decay, as they don’t leave channels out of the mix. However, each of them has their flaws too.
Make sure you look at the quality of the customers that are coming to your site from each channel. Let’s use the example for illustrating attribution models earlier:
- While AdWords was responsible for the ultimate discovery of the brand and site, each visit was a minute or less in duration, often resulting in a bounce from the customer.
- While Facebook had a much more significant impact, the customer ultimately didn’t purchase and only consulted 4 pages.
- Organic Google had the biggest impact resulting in sale, and a 32-minute long session.
What do we know from this?
While AdWords was responsible for discovery, it had an low impact on the sale. The customer coming in from the ad click was not ready to buy, nor willing to browse the site. This can be attributed to a lack of keyword optimization, or perhaps just low lead quality from AdWords.
Social media did a bit better in terms of how engaged the customer was. We can perhaps attribute this to a product promotion on Facebook that got the customer directly interested in a specific item. The customer also looked at four other pages before leaving the site.
Organic Google search likely came from a customer searching for the name of the site or the product. While this channel had the highest impact, it’s likely that the customer was searching because they had already decided to buy.
How your Data Shows you the Answers to Multichannel Attribution
One of the best ways to really nail multichannel attribution is to create the model that’s best for your online store based on the key aspects of your own data.
A cookie cutter attribution model will always have flaws because it’s not built around your own data.
For example, if you look at the differences in revenue and order count between last click and Divvit weighted, there is far more attributed to last click while the weighted model disperses a bit more across the channels that have the highest impact on the actual sale.
Your best bet for a truly clear picture is to define your own multichannel attribution model based on the data you already have.
Start with the attribution model that works best for your customers, and add on factors and weights based on the data you’ve been accumulating. Creating a custom multichannel attribution model based on your own store is key to getting a real grasp on your marketing spend.
What to Consider when Choosing a Multichannel Attribution Model
There are several things to consider when you choose the right attribution model for your ecommerce store:
How many touchpoints does your brand typically need?
The Divvit average is 5.5 touchpoints between discovery and conversion. However, your customers might need more or less time to make their decisions. Think about how many channels you might need to attribute to, and where your cut off point is.
Which channels bring in the most customers for your ecommerce site?
Which are your best performing channels? Which channels bring in the most customers that purchase? It’s important to consider which channels to attribute to, and if you have a significant amount of channels, it might be better to group them up.
Which channels have the highest impact on average order value and the number of products purchased?
Not all leads are created equally. Some channels might bring in customers that are looking for deals and only purchase small-amount products. Others might be bringing in customers that are willing to spend more. Decide which of your channels have the highest impact on your overall revenue.
Understanding what key points to consider is crucial to making sure that you’re spending your budget wisely. If you know that one channel has a higher impact than another, tweaking your marketing spend will be the best way to optimize your ROI.
Divvit makes this process easy by showing your customer journey from beginning to end so you can make an informed decision about which of your channels performs the best for your customers.
By hovering over each channel in your customer journey, you can see how long they spent on your site, how many pages they visited, and whether or not the lead from that channel was qualified.
Above all, never stop tweaking your attribution model. Just because it works today doesn’t mean that it will work as your store evolves and grows. Testing everything is the way to be as data focused as possible and to keep an approach that is guided by the information you have at your disposal.
Making an informed decision about your marketing spend will increase your bottom line. There is always room for optimization, and if you can decrease your spending on channels that have a lower impact and reallocate those funds to channels that work for your ecommerce site, your margins will see a significant jump.