Attribution for eCommerce 101

    Aidan Corbett

    May 15, 2020

    Marketing, Analytics

    eCommerce attribution is perceived as one of the most complex areas of analytics, so much that many eCommerce founders, as well as marketers don't really bother to think about it.

    However, it is tough to understand the unique value of each marketing channel without it.

    In an eCommerce world where ROAS and gross margins are king, eCommerce attribution is a tool for marketers to understand what to cut and where to double down.

    But what is attribution?

    An attribution model is a backward-looking analytics system that allows you to credit your various customer touchpoints for a percentage of the sale or conversion. That includes your advertising channels (Facebook, Instagram, Google Ads), touches with your support team, and even visits to any physical stores if you have them.

    Types of attribution models

    There are many attribution systems used by marketers and analytics platforms. These are some of the most popular:

    1. Last touch, or last click: the final touchpoint gets all the credit for the sale, regardless of any other touches that may have set up the sale. It is very common, effortless to set up, and very popular because of that, but it is flawed. Assigning all the weight to the last action, disregards most of the buyer's lifecycle.
    2. First touch, or first click: the opposite from last touch - the first touch gets all the credit, regardless of any efforts you may have done to nurture the contact on their way to the sale. Great if you want to understand what type of campaigns bring awareness to your brand, but often inaccurate because of cookie expiration times.
    3. Linear: all touchpoints get equal weight. Unlikely to be accurate, but accurate enough. By definition, not all touches are made equal, so the spread between the real value and the value attributed by the linear model is likely to give you an inaccurate view of things.
    4. Time-decay: all touchpoints get credit. However, the touchpoints closer to the sale get drastically more weight than those early in the path. In our opinion, the hardest to set up (particularly if you don't have an expensive analytics solution) but the most accurate to eCommerce mechanics.

    eCommerce Attribution Models
    There are tens of other attribution models: U-shaped, W-shaped, Positional, Algorithmic, etcetera, but those four are the most popular.

    The perfect attribution model for your eCommerce brand

    Figuring out which attribution model works best for your brand depends on your vertical, your LTV/AOV multiple, and your marketing spend breakdown, among other factors.

    Most businesses that are advanced in attribution choose a custom model, set up by them or a consultant that provides proper weighting to touches based on that business' unique mechanics. eCommerce attribution modeling is a creative art, and like all, it requires interpretation.

    Which marketing channels do you use? Which ones don't you use? Which ones are directly converting? Which ones are assists, or 'soft conversions', that set up sales later in the funnel, how does the average purchasing path look like? Where are you finding the best ROAS?

    Those are all questions that have to be answered, which happen to be unique to your business.

    LTV and Assisted conversions

    The two top things marketers LOVE to forget.

    I see ROAS calculations being done on average purchase (cart) all the time, which hurts. It is five times more expensive to acquire a customer than to retain it, and that number keeps climbing every year. That means the marginal cost of every sale after the first, particularly in advertising, becomes minimal. That means that whatever margin you made on the first sale will only increase the second and third time they purchase. Most eCommerce companies understand this pretty well at the sales level. However, in their attribution model, eCommerces sometimes ignore it, resulting in a misreading of each channel's value.

    Variation in purchase frequency matters when making channel spend decisions. Do not think about first conversions as much as you think about the long term, and you will be surprised at the value of your funnel.

    On the other hand, assisted conversions are the name of the game, in a world where virtually 98% of visitors won't buy the first time they visit you, and carts get abandoned two-thirds of the time. Regardless of how targeted your campaigns are, how unique your creative is, how much social proof you have on-page, chances are you aren't doing much better than the industry.

    What differentiates top eCommerces is their understanding that the first visit (touch) is the start (not the culmination) of the relationship and their ability to create additional touchpoints around that first experience to lead to sales, with each of those small interactions (touches) participating towards an eventual sale.

    That is why the first touch and last touch are flawed: nobody is buying in the first touch, and unless you create these assisting experiences in the middle of the funnel, nobody will give you the opportunity of a last touch. Think long term, high frequency.

    So, where do I start

    Glad you asked! Analytics is a great place to start. Open your analytics site and go to:

    Go to Analytics and on the left sidebar > Conversions > Multi-channel funnels > Top conversion paths.

    That page should give you an idea of which pages people are interacting with before converting (or achieving one of the goals you've set up in your analytics).

    Analytics Conversion Paths
    Analytics Conversion Paths

    In that same section, Time lag will give you an idea of the number of days from first ad impression to conversion for whatever channel you are looking at.

    Now, onto Assisted conversions, which will give you Google's best idea of which soft conversions and touches along the way participated in your conversions. Remember, you can always narrow down by channel, source, or even landing page to get even more specific results.

    Now, this all assumes you have set up proper UTM tracking in the past. If you have not, there are plenty of excellent guides online on how to do it. I'd invite you to take time right now to do it.

    Why does all this matter?

    Data should be the one driver of your marketing, investment, and by extension, business decisions. If you don't use data to guide, you will make the mistake of cutting a high-converting ad campaign, or doubling down on the wrong channel, or worst of all, leaving sales on the table (all of those, mistakes I've made more than once earlier in my career).

    Understanding what converts and what doesn't, even when it is not self-evident at the surface level, is the marketing version of knowing what sells and what doesn't. If you are in eCommerce, you probably know the second one reasonably well. In my opinion, understanding the first one, however, is the difference between your business being a Volvo, slow and steady, and a rocketship.

    Want to grow your eCommerce business? Get funded today