Attribution modeling for eCommerce - Are you doing it right?

Posted on 18 July, 2016

What if I told you that you have attributed your marketing budget all wrong?

I know you won’t believe it. You have followed the best practices of attributing your marketing budget to the channel which has the highest conversion rate and stopped advertising on the channel which had a low conversion rate. You are technically right, but that is far from reality.

By default, Google analytics attributes the credits to the last click which led to conversion. It does not tell you the full story in terms of channels which assisted a conversion. The default story from analytics is somewhat like :-

Once upon a time, the end.

A lot of business owners still take decisions based on this; they do not take into consideration the channels which helped in getting that conversion. This is a particular case with eCommerce stores who are just starting out and learning bits and pieces of analytics. The journey of a customer on digital platforms is pretty complex; it is quite different from what Google Analytics shows by default. The real story of a customer in digital time is something like this:

Once upon a time, Timmy (our hero customer) found an awesome product on Facebook shared in one of the random groups; he clicked the post and landed on our website to explore product features and price. This time, he was there just to explore the product and not buy it.

(2 days later)

Timmy found his colleague having that awesome product and how it helped him make his life less difficult. Seeing this, Timmy wanted to check out the product features again.  He did not know the name of our awesome brand so he just typed in the name of the product and voila, there was a search ad for our awesome product. He explored the product again, checked out other products in the category as well. But the need was not so strong that made him buy the product. He left the website without purchasing (sad)

(Few hours later)

While browsing the web, Timmy saw an eye-catching ad which was from our retargeting campaign. It was the same product Timmy was interested in and the ad offered an extra 5% discount just for Timmy, he quickly clicked on the added and gave his email to receive 5% discount code. Now he is having the 5% discount code but he won’t make a purchase directly because Timmy is a rational customer. He took the power of Google in his hands and searched for same products with a cheaper price. At the end, he ended on an article which states that why our product is cheapest and of the best quality. He was impressed and ready to buy. He quickly found our awesome products in the bookmarks he saved, copied the coupon code from email and made a purchase. (Finally)

digital customer purchase journey

As Timmy directly came on the website, google analytics by default credits the direct channel with conversion, but is that right? No! There were various channels involved in conversion without which conversion would not have taken place. This was a pretty complex conversion path and involved many channels. Not all customers have this long conversion path, but there are people like Timmy as well. If you rely on default Google Analytics conversion metrics to divide your marketing budget, you are not taking the right decision.

It is also possible that some eCommerce stores who might not have attribution issues where most of the people make a purchase in their 1st visit and there are no assisted conversions. Those are exceptions, if you are one of them, stop reading this now!

How do I know if I have Attribution issue?

Google Analytics has the answer.



You can check that out in Multi-Channel Funnels section of Google Analytics. For those of you who have never used it before, here is a video from Google Analytics to give you a brief introduction.


You are having an attribution problem if,

-  You have assisted conversions

-  Your conversion path is long and path length is 2 or more

In short, you have attribution problem if your multi-channel reports look like the image below.

How to solve attribution issues in Google Analytics?

Google Analytics has an awesome tool called ‘Model Comparison Tool’, you can access it under Conversion > Attribution. It is a free tool from Google Analytics which allows you to attribute credits to all your marketing channels. You can select up to 3 models for comparison.

There are 7 different attribution model provided by default which you can start to explore.

  • Last Interaction: This is the default model in the model comparison tool. It gives all the credit to the last click i.e. 100% credit to the last channel.
  • Last Non-Direct Click: This model gives 100% credit to the last campaign. A campaign is anything but direct traffic. If a direct visit led to conversion, the click before direct will get all the credit in this model. It highly under-values direct channel.
  • Last Adwords Click: All the credits under this model is credited to Adwords irrespective of its position in the conversion path. This model is used to analyze the effectiveness of AdWords campaign.
  • First Interaction: First interaction model is a really simple model; it gives 100% credit to the first channel.
  • Linear: This model distributes credits among all the channels. As simple as that. If there are 4 channels involved, each one will get 25% credit.
  • Time Decay: As the name suggests, this model gives credits based on time. A higher percentage of the credit will be assigned to the channels that are at the end of conversion path.
  • Position Based: Position based model gives more credit to the first and last channels while equally distributing rest of the credit to the middle channels.

Which is the best attribution model for eCommerce?


Every model has their own advantages which are used for the purpose of analysis but none of them is a perfect model which you can rely upon. You can try experimenting with all of them and see surprising results in conversion data.

One of the most interesting things about eCommerce stores is that every eCommerce store has different types of traffic sources; there is not one model which can be used for all eCommerce stores.

Custom Attribution Model is to the rescue in this case. Custom attribution allows you to customize all the above models as a starting point so that you can customize credits given to each channel based on its importance to your business.

To use Custom Attribution Model, click ‘Select Model’ on Custom Attribution tool. You will see a drop down of all the models, click ‘Create new custom model’.

I start with Position based model because it allows to put weights on the position of interaction, you can select others if they suit your business objectives. The first step here is to assign percentage of credits to each position.

Assigning the amount of percentage is not a random guess here, it depends on your traffic sources and their importance. It is different for everyone. You have to keep traffic sources in mind before making this change. Above attribution is just an example, if your product discovery is important for you, you can assign more percentage to the first interaction. The last interaction is most important because it led to final conversion.

The second step is to set a Lookback window. It used to set the number of days for which a conversion should be credited to a particular channel. Number of days, again is not just a random number. You can land on that number by looking ‘Time lag’ report. Best practice is to use the upper limit in the number of days in the time lag report plus a bit more. For eg. you find that the maximum number of days a customer takes to make a purchase is 50, you can keep 60. (P.S. This is just a thought which has worked for me, you can try anything which suits your business type)

The third step is to assign credit on the based on engagement. We have 2 options 1) Time on site 2) Page depth

Time on page is dependent on a lot of things and require custom codes to get it right.

Page depth is a simple and useful metric. If a visitor from channel A sees 7 pages on the website and a visitor from channel B bounces from the 1st page, channel a gets more credit.

The last step is applying custom credits to each source/channel. This is one of my favorite part of analytics. If one channel is more important to you, you can assign more credit to it than other channels. If one is less important or not important at all, you can give less credits to them. If you feel that your display advertising retargeting campaign is more important than the other source because it brought the customer back to the website, you can assign it more credits. For e.g. I have given 1.5 times more credits to Display campaign than other campaigns. This was just for example, you have to do a lot of experimenting to arrive at this number. 

You also have the option to make custom channel grouping for Branded Keywords, Unbranded keywords, Facebook, Referral source etc. It makes sense to give less credits to the branded keywords and more to unbranded keywords. But then again, it is your choice and totally dependent on your business profile. For eCommerce business, you can give more credits to the conversion which came from cart abandonment emails than the normal newsletter.

There are many possibilities, all you have to do is experiment, collect data and then take the decision. As I told in the introduction that attribution model for every eCommerce store is different, you need to work on it and find out what fits your business. There is no perfect model in google analytics, you have to work out on your own model, it will at least be better than the last click model and will help you take a better decision while dividing your marketing budget.

Do comment below if you have used Attribution modeling for eCommerce store and made better decisions with it or if you have any questions related to Model Attribution Tool in Google Analytics.

Shetul Majithiya , Senior Digital Marketing Executive


About Emipro

Being an emerging leader in IT market since 2011, Emipro Technologies Pvt. Ltd. has been providing a wide range of business solutions in Odoo & Magento. We are pleased to have a large pool of contented customers with our meticulous work in the domain of ERP & e-Commerce. Our customers are companies of all sizes ranging from startups to large enterprises who realize that they need a professional internet solution to generate revenue streams, establish proper communication channels, to achieve desired goals and streamline business operations. [....] Read More

Our writings seems informative ?

Subscribe for our Biznote and get more amazing stuff directly to your inbox!

Post Your Review


Your Review has been posted

0 Comment(s)