Attribution Modeling for E-Commerce: 5 Things All Sellers Need to Know

January 9, 2017

I’m not going to lie to you: this can get very tricky. It’s so complex that many e-commerce platforms don’t even bother with it.

An attribution model is how you assign credit or value for sales and conversions across various customer touchpoints. It includes all your digital channels – paid search, display, email, social media, organic search, referrals – and the impact that each one has on the eventual conversion.

In the good old days, this was easy. You ran a radio ad, for example, and that brought in five new customers worth $250. That one touchpoint – the ad – got 100% of the credit for those sales. Simple.

But today? According the Content Marketing Institute’s annual report, marketers use an average of 13 tactics, seven different social media platforms, and three paid advertising channels in their efforts.

The route people take to find out about you, learn about you, and ultimately buy something from your e-commerce store can be long and convoluted. Your sales funnel can be big…very big.

How are you supposed to keep track of effectiveness?

Attribution modeling.

But just as the tactics, strategies, and channels have gotten more complicated, so too have the models to assign value to them. They can be basic and rules-based, or elaborate and algorithm-based. Single-touch or multi-touch.

In fact, there are at least five different models that are widely used, and even more depending on how you define and break them down:

  1. First Touch (aka First-Click) assigns 100% of the credit to the first touchpoint in a conversion path. This is great for insight into how people find you (and the top of your funnel), but if they hit three other touchpoints before converting, does it really deserve all the glory?
  2. Last Touch (aka Last-Click) gives the very last touchpoint full credit, no matter how many others they may have traversed. Easy to track and set-up, but almost universally considered worthless nowadays. There’s too much going on beforehand, and it gives zero notice to the top and middle of the funnel activities.
  3. Linear assigns equal value to each step in the conversion path. If a customer traveled through four touchpoints before buying, each would get 25%. This is better – every point is considered and valued – but it tends to overvalue minor ones and undervalue key touchpoints.
  4. Positional favors both the first and last touch – typically giving them each 40% of the credit – while dividing the remaining 20% amongst the middle touchpoints. Obviously, the model can drastically undervalue the middle, especially in a long path.
  5. Time-decay is a simple algorithmic model that gives most credit to the point closest to conversion, and increasingly less as you move away from it. While it still favors the last touch, it does give some kudos to every step along the way, and as such, it’s the preferred model for many marketers and business owners.

And I haven’t even mentioned your best bet: the custom option (a model based on your platform, audience, marketing, and specific business goals). Avinash Kaushik has a great walkthrough of setting up your own custom model on Google Analytics over at Occam’s Razor.

But be prepared. He begins the post with these ominous words: “There are few things more complicated in analytics (all analytics, big data and huge data!) than multi-channel attribution modeling.”

And he’s not lying. It can be – and usually is – complex, tedious, and frustrating. A lot of it is trial and error over time.

That said, it’s absolutely worth it. Implementing an attribution model helps you understand what’s influencing customers to buy, how they shop, where they’re coming from, and what channels and tactics deserve the lion’s share of the credit (and therefore increased allocation of your marketing budget).

As you start out on your attribution adventure in the e-commerce sphere, there are five things you’ll want to keep in mind before you take that all-important first step.

Assisted Conversions Are the Bulk of Your Sales

Up to 98% of visitors to your site will not buy on their first visit. 55% will leave within 15 seconds of arriving.

84% of consumers say they either “completely” or “somewhat” trust recommendations from family, friends, and colleagues about products. 88% trust reviews from strangers as much as those from people they know.

Average cart abandonment rate is 68.81%.

Econsultancy found that 88% of consumers turn to online reviews when considering a purchase.

All of this is just to point out the obvious: few people are showing up at your digital storefront and buying something on their first visit. They drop by, do some poking around, check out some online reviews, explore your social media accounts, search for upcoming sales or coupons, do a little more virtual window shopping, check out what others are saying about you on Twitter, and finally return to your site when they’re ready to slap down that cash.

Each of those leads them just a little closer to the sale. Each interaction assists in the eventual conversion. The bulk of your sales come from these assisted conversions, so you’d best be tracking them and giving at least some credit to each one so you truly understand your customer behavior and how you should be marketing to them.

Want to see for yourself? Head on over to Google Analytics and check out Conversions > Multi-Channel Funnels > Path Length to see exactly how many conversions are happening after just one interaction, two interactions, three interactions, and so on. You’ll be amazed.


Next, look at Conversions > Multi-Channel Funnels > Assisted Conversions. Take a close look at the last column labeled “Assisted/Last Click or Direct Conversions”. A value less than 1? That channel is typically the last touch point before a conversion (i.e. sale). Greater than 1? It’s a step along the way. Using either first or last touch attribution? Those valuable cogs in the machine are being ignored.


I’ll say it again: assisted conversions are the bulk of your sales. Your model needs to recognize that.

There is No Ideal Attribution Model

Every model has its limitations and shortcomings. In a perfect world, we could just point at the one model to rule them all and be done with it. But it doesn’t work that way.

A custom model is best, but it does take awhile to collect the necessary data and customer understanding to pull it off. And if your data is wonky, it’s going to give you wonky insight.

Attribution modeling is as much art as science. And as much data-supported fact as intuition and educated guesswork.

Your custom model has to reflect your customers and your business. You need to consider the behavior that’s important to your goals, and the “soft” conversions (signing up for the newsletter, requesting a customer rep contact them, and so on) that ultimately carry value because they result in sales…no matter how far down the line.

What are your top conversion paths and top assisted conversions? That tells you what your customers are doing. Are you using those to their full potential? Are you throwing money away on channels that don’t make an appearance in those reports?

Customers search, consider, and buy in different ways. Make a model that works for them. And you.

You Need to Rely on Data

There’s no room in attribution modeling for blind guesses. Your efforts should be guided by hard data at all times and in all things.

Rely on Google Analytics and the enhanced ecommerce plugin to collect the data you need. Set up goals and funnels. Turn to the reports to inform your decisions.

You should be tracking all your efforts as accurately as possible to guarantee solid data and insight.

  • Use auto-tagging options in Adwords, Bing Ads, and DoubleClick.
  • Use UTM parameters for social campaigns.

If you’re spending time, energy, or money on something, you should be collecting data on it. Far too many e-commerce platforms don’t actively check and use their analytics. Don’t fall into that group. You’re a savvy business owner, not a statistic, right?

Analytics should not scare you. Anyone can get started with it, regardless of their comfort level with technology (and you more than likely won’t even need to mess around with the code thanks to easy integration with platforms like Magento, Shopify and WordPress plugins).

The more data you have, crunch, and consider, the more accurate your decisions, and the better your understanding on the conversion paths your customers are actually taking to the (virtual) checkout counter.

Want to grow your business and revenue? Get the data. Use the data.

You’re spoiled for choice in the 21st century, and practically every third-party service you employ offers its own set of data collection and analysis (and/or streamlined integration with the most popular analytics platforms) for their particular area. Take advantage of the windfall.

LTV Should be Factored In

Do you want to sell once and be done with it? Or would you prefer customers return again and again? Let’s assume the answer is obvious (because it is).

Consequently, you need to consider the lifetime value (LTV) of each customer. Your top customers – the creme de la creme in the top 1% – are worth up to 18x more than an average customer.

Far too often, though, we give little to no thought to those returning customers and the value they bring with them.

They get ignored, and the conversion path for them might slip through the cracks. Not good. They’re not coming back the same way new customers are discovering you. Their path is unique and needs to be recognized…because it deserves credit for the revenue it’s bringing in.

So look at the traffic in your entire funnel, especially at the top. Chances are, you’ll find a good chunk of those returning customers are coming via direct (entering the address or using a bookmark), social media (when you tweet or post about sales and…

Get more information about this news here: Read More