How to Analyse UTM and Referrer in Amplitude Analytics

When you are trying to analyse your website users, knowing where they have come from (their acquisition channel) is crucial. Whether they arrived from a link on a forum or an ad campaign will often result in different behaviour, likely representing a different type of user, so it’s an important method of segmentation. But how do we see that in Amplitude Analytics?

Capturing UTM + Referrer Data

The good news is that if a user clicked a link, it’s likely we can see where they came from.

To get started, we need to make sure we have configured the right options when implementing Amplitude. Open a chart in Amplitude, and see if you can see the referrer and UTM options to group by:

If you can’t see user properties for referrer and UTM, then you need to follow the instructions here to switch them on. If you’ve integrated Amplitude via segment, follow these instructions instead. In either case, make sure all of UTM, referrer, and GCLID are set to TRUE.

If you had to switch it on, you’ll need to wait a few days before you have some data to analyse.

Analysing by Acquisition Channel

So now we have the data, what can we do with it?

A classic use case is just to look at your visitors and break them down by acquisition channel.

To do this simply group by “utm_source”.

You’ll likely see a good chunk of users with UTM Source “none”. These are users that didn’t click a UTM link (often organic traffic), so we need to look instead at referrer to see where they came from

Group by “referring_domain” to get a nice overview of the site they were on before yours.

Common values here are Google (organic search results), Facebook (links in messages, groups, etc, and other Forums like Quora and Reddit.

If you want to see exactly which page of that site they were on, group instead by referrer.

You can group by these properties in other charts too, like funnel and retention, and create user segments or cohorts that are meaningful for your site.

What does “initial_” mean?

Sometimes a user will visit your site several times but from different sources. The Amplitude cookie ensures that the user is only counted once, but in this case it can be interesting to know the different links they clicked. As such we have the following:

• initial_referring_domain is where they came from the first time
• referring_domain is where they came from most recently

The structure is the same for all properties. This setup enables us to implement what are known as first-touch and last-touch attribution models, and analyse our marketing in different ways.

Does this work for Mobile Apps too?

No! I wish! Mobile attribution, and therefore traffic analysis, is notoriously difficult. Most people will recommend either to implement a mobile marketing platform like AppsFlyer (technical overhead), or to have a survey question in your app onboarding asking how the user heard about your app (user friction).

Need Some Help?

Through my agency Permutable, I help companies get more out of their analytics setup, interpret the data, and make decisions that lead to growth. Please get in touch if you’re interested, or subscribe below to get notified when I post more analytics tutorials.

30 Minutes Free Analytics Coaching!

I’m running a survey to understand more about my readers. In return for completing the survey, I’ll give you 30 minutes of free coaching. Here’s the link.