Understanding Attribution for E-Commerce

Last updated: November 14th, 2018.

Anyone who has spent any time in e-commerce knows that our world is bursting at the seams with fairy tales (“Ooh, look! This free webinar will teach me a killer funnel that will increase my sales by 500% overnight!”)

Those are nonsense, clearly. But there are plenty of traps on the path to a successful e-commerce store.

When it comes to paid marketing, there are key things you can rely on to truly help you.

Finding the right attribution model to apply to your advertising data, to determine where your sales are really coming from, is one of the true keys to the kingdom of profit.

With an effective model of your advertising funnel, you can jettison the tactics that aren't working and put more effort into the ones that are.

The following article outlines the main attribution models that are available to most people through Google Analytics or other attribution software and when to use them, with a little insight from our experiences developing attribution tools at Order Metrics

Why Should You Trust Us?

This was written by George Sylvain, co-founder of Social Print Studio, a successful e-commerce store that has been selling prints and profitable for 8 years.

Additional help was provided by Hank Kronick,founder of Order Metrics. In some cases we worked with researchers to make sure we covered all angles. We have a combined 16 years of e-commerce experience and have personally the techniques described in this guide, and seen the results in practice.

Choosing An Attribution Model - Why It's Important

The goal of attribution is to know the true value of each marketing touch point, by understanding how they work together.

Many marketers over-value the last click in their sales funnel because thats the one that delivered the sale.

Also, until recently the last-click attribution model was the only one available.

Consider a situation where the last click before purchasing was always from an email marketing campaign.

Last click analysis would suggest you should cancel all your other marketing, because the email marketing is the only one delivering paid conversions. But that would be wrong-headed.

You need to understand where that customer came from originally before they got on the email list, and to do that, you need to attribute a portion of each conversion value to that original marketing touch point as well as to your email marketing.

6 Main Attribution Models

You need to know a little about these 6 attribution models that are standard in Google Analytics as well as many other attribution modeling tools. Then we can go about explaining which one to use and why.

  • Last Interaction: This is the default setting that Google Analytics uses to track your campaign. It gives 100% of the credit to the interaction that immediately preceded the sale. Important? Absolutely. Short sighted? Incredibly. While utilizing the data from this model could help increase sales by adding focus to interactions that close them, ignoring any other marketing touch points that got your customer to the conversion point will limit the potential of your campaign.
    When to use: If you only want to know the last ad clicked before a sale. Not advised, old-school techniques.
  • Last Non-Direct Click: This model shifts the credit to whatever ad campaign, organic click or paid action led to a conversion, excluding any that came direct. Direct is when someone types in your URL or it can happen when someone follows a link from outside the browser. It overvalues the previous non-direct click though, discounting whatever reason they came to your site directly. This can be useful for boosting sales by over-investing in effective ad campaigns but relying on it completely could lead to an over-reliance on paid content.
    When to use: Could be good if you are only running one ad campaign.
  • Last Ad Click: In this model, all the credit goes to Ads, wherever they may be along the actual path to conversion. It’s a great way to make Google look great. It’s also a great way to analyze the effectiveness of your Ads campaign. If a certain ad is performing stronger than others, this model will make it look like a star. It pays no need to organic search or direct traffic.
    When to use: If you only want to understand the role Google Ads play in your purchase funnel.
  • First Interaction:This is the “Once upon a time” model. When tracking your customer’s sales story, why not start at the beginning? This is a great way to monitor the top of your funnel and figure out what hooks are most effective at sending your clients down the path to you. However, be careful not to discount the actual channel that created the conversion.
    When to use: Only good when studying brand awareness - how people initially hear of you.
  • Time Decay: The first Multi-Touch model! This model doles out credits based on how long it takes for your customers to convert. Channels they saw five days ago that influenced them still get a little love, but not as much as the channel that immediately led to the conversion. It’s a pretty efficient model and a good starting point into this process. However, it tends to undervalue the first touch. If a strong ad hooked them in the beginning and got them thinking about you, this model may not give that the credit it deserves.
    When to use: Good for longer sales funnels, when you want to know which ad finally pushed someone to convert but not ignore any of the others.
  • Position-Based:This is the leader of the pack for many companies - the Goldilocks model for e-commerce in our opinion. In the position-based model, 40% of the credit goes to both the first and last touch points, and then the remaining 20% is spread evenly among the rest of the journey. Does it have the potential to undervalue your mid-funnel touch points? Absolutely. However, that can be worth it to give proper credit to both the value from the first touch point and the efficiency of the closing touch point. This is the model that will be most effective for the vast majority of small to medium sized businesses. If you want a “set it and forget it” model, position-based is the way to go.
    When to use: ASAP! In the early days of ad-tracking, Last-Click was the only option. But with ad-tracking so advanced, you need to use a multi-touch model like Position Based to truly understand your business.

Other Attribution Models

In addition to the default models, there are a few other ways to structure a campaign if you want to go deeper into the digital weeds. There are other data-driven models that use algorithms to assign credit, which typically are developed by your own data science team. You can also use Google’s Custom Attribution Model Tool to set your own weight for the parts of the funnel that you believe are most effective.

Google Analytics has an Attribution Model Comparison Tool, which lets you see exactly how the credit distribution will be affected by up to three different models at once, letting you get a ton of data across models. You can get a sense from this of the different decisions each model can lead you to make.

The Power Of Position Based Model

We strongly believe that for small and medium sized e-commerce businesses who are running a few different ad campaigns, the position-based model makes the most sense.

Start with 40% of the value assigned to the first way the customer found you, whatever that is, often a Google Ad or Social Referral

Then you say, 40% of the value is assigned to the last click, the one which led to a sale.

But you still assign some value, the remaining 20% to any other marketing touch points the customer interacted with in between.

This multi-touch attribution model gives you a theoretical model of how much of your total Customer Acquisition Cost you can assign to each step on the customer purchase journey.

Google Analytics and other tools like the reporting inside Facebook Ads all default to last-click or last non-direct click. If you have never thought about changing it, you are going to be overvaluing the last click. All the data you are seeing in the last-click model is telling you that you need to double down on that final step of the flow. But that just doesn't make sense to you, the business owner(unless you are selling ad clicks!).

So if you are new to this, the biggest thing we believe you can do is to change your Google Analytics model to Position-Based and then see how many sales Google Analytics is telling you came from each channel. Chances are you'll see a whole lot of different numbers and can make far better choices about where to invest those precious ad dollars.

About This Guide

The Order Metrics Profit & Growth guide is designed to help you unlock the most profit for your e-commerce business.

This is not intended to be nonsense content marketing like you find on other company blogs.

These are real, tested techniques could help you grow your business and put some cash back in your pocket so you can invest in things that matter to you.

Check out the other articles we have written back at our profit guide index.

Want to find even more ways to improve your bottom line?

Order Metrics

OrderMetrics - E-Commerce Profit Dashboard

Check out Order Metrics - A profit management tool for growing ecommerce merchants. The best way to track and measure your e-commerce store performance across multiple storefronts , advertising providers and shipping providers. Customers report increasing their margin by 5-6 percentage points after using Order Metrics.

Try Order Metrics for Free