8 attribution models to measure the effect of email marketing

8 attribution models to measure the effect of email marketing

Ever wondered how you can measure the impact of email marketing on your online conversions? Assigning weighted percentages to the channels that helped make the conversion, also known as the attribution model, will do the trick. But what are the 8 most common attribution models, and how do you use them? We'll tell you more…

To understand attribution models, you have to dig into the customer journey. This is the path your customer has followed from stimulus to purchase. An example path would be: Referral stimulus > website visit > email sign-up > newsletter email #1 > promo email #2 > social > direct traffic > conversion. In this case you have 4 different online touch points: referral, email, social, direct traffic. The question is: which channel has the most impact on the sales conversion? Let's look into that.

1. Last click attribution
The name says it all: 100% of the attribution goes to the last channel your customer interacted with before making the purchase. In this case, direct traffic would get 100% attribution. This is the easiest attribution model, but be aware that this is not fairly weighted. Email and social channels — clearly influencers of the purchase path — are not granted their fair amount of contribution.

2. First click attribution
In this case, the first touch point your customer had with your brand is the touch point that gets 100% attribution. In this case: referral. Same pros and cons as the last-click attribution, but in this case you'll learn more about acquisition of new customers. It tells you nothing about returning customers though.

3. Last non-direct attribution
Because in last-click attribution direct traffic often gets the sale, a variation is sometimes used to attribute the sale to the last non-direct channel. In this case: social. This is where you filter out direct traffic. But on the downside, the remainder of your channels gets zero attribution.

4. Linear attribution
Now we're getting somewhere. This is also a fairly easy attribution model. You divide the percentage of sales equally over all touch points. So in this case, every touch point get the same percentage of the sales. Then you can calculate the cost vs. turnover and calculate ROI per channel. This is a good starter if you are looking at attribution models, but even so: there are more accurate calculations.

5. Time decay attribution
This is not frequently used, but offers some great insights. The concept is simple: the touch points closest to the conversion will get the most attribution and when we move back in time, the attribution diminishes. So when we look back at our earlier example, this could be an accurate insight: Referral stimulus 1% > website visit 3% > email sign-up 6% > email #1 15%  > email #2 20% > social 25%  > direct traffic 30% > conversion.

6. Position based attribution
Here you add more credit to the first and to the last touch point your customer has interacted with. So in this case, referral and direct traffic would get 80%, while the other 20% is divided over the other touch points along the path. This gives a lot of credit to the first and last stimulus, but everything in between is left out in the cold. Email and social with their more programmatic thinking and focus on influencing, are grossly underestimated.

7. Weighted attribution
This attribution model is commonly used, and when done right, it offers valuable insights. Here you weigh all your touch points. So you analyze the contribution per channel, and attribute the right percentage to them: referral is 10%; website & direct is 45%, email is 30% and social is 15%. You can look at historical data, or follow your gut feeling. Either way, it will give you some great insights in sales contribution.

8. Algorythmic attribution
The holy grail of attribution models… but also the most difficult one. This hybrid solution is all about constantly updating and learning more about the channels that work best. In essence, what happens is that you put all your data together, and a smart engine learns you how well each channel performs. To do this, a good setup is essential:
1) use a base model: i.e. position-based attribution
2) select a time window – trace the customer journey back to 70 days before purchase
3) select the engagement credit (conversion, time on site, # page visits,…)
4) apply rules - give extra credit to specific keyword searches or customers who opened all newsletters,…
5) add tracking code to all your channels.

When you have this overview, you can easily monitor and value all your touch points. And when conversions fluctuate, you have your dashboard to find out which channels contribute more.

The key is to constantly learn and adapt to find the model closest to reality. There is no such thing as a 100% accurate attribution model. It is an ongoing strategy to learn more about your customer and which touch points influence his purchase path. And based upon these insights you can make well-funded decision on investing more or less in certain channels.


Start with easy applicable attribution models and evolve to more complex, but accurate models. This way you well continuously get better insights in your marketing efforts.

Posted on
Jun 04, 2014