Django UTM Tracker

Django app for extracting and storing UTM tracking values.

Background

This app has been designed to integrate the standard utm_* querystring parameters that are used by online advertisers with your Django project.

It does not replace analytics (e.g. Google Analytics) and Adwords tracking, but does have one crucial difference - it allows you to assign a specific user to a campaign advert.

This may be useful if you are trying to assess the value of multiple channels / campaigns.

Supported querystring parameters

Parameter	Definition
utm_medium	Identifies what type of link was used.
utm_source	Identifies which site sent the traffic, and is a required parameter.
utm_campaign	Identifies a specific product promotion or strategic campaign.
utm_term	Identifies search terms.
gclid	Identifies a google click, is used for ad tracking in Google Analytics via Google Ads.
aclk	Identifies a Microsoft Ad click (bing), is used for ad tracking.
msclkid	Identifies a Microsoft Ad click (MS ad network), is used for ad tracking.
twclid	Identifies a Twitter Ad click, is used for ad tracking.
fbclid	Identifies a Facebook Ad click, is used for ad tracking.

How it works

The app works as a pair of middleware classes, that extract utm_ values from any incoming request querystring, and then store those parameters against the request.user (if authenticated), or in the request.session (if not).

The following shows this workflow (pseudocode - see test_utm_and_lead_source for a real example):

client = Client()
# first request stashes values, but does not create a LeadSource as user is anonymous
client.get("/?utm_medium=medium&utm_source=source...")
assert utm_values_in_session
assert LeadSource.objects.count() == 0

# subsequent request, with authenticated user, extracts values and stores LeadSource
user = User.objects.create(username="fred")
client.force_login(user, backend=settings.FORCED_AUTH_BACKEND)
client.get("/")
assert not utm_values_in_session
assert LeadSource.objects.count() == 1

Why split the middleware in two?

By splitting the middleware into two classes, we enable the use case where we can track leads without utm_ querystring parameters. For instance, if you have an internal referral program, using a simple token, you can capture this as a LeadSource by adding sentinel values to the request.session:

def referral(request, token):
    # do token handling
    ...
    # medium and source are mandatory for lead source capture
    request.session["utm_medium"] = "referral"
    request.session["utm_source"] = "internal"
    # campaign, term and content are optional fields
    request.session["utm_campaign"] = "july"
    request.session["utm_term"] = token
    request.session["utm_content"] = "buy-me"
    return render(request, "landing_page.html")

Configuration

Add the app to INSTALLED_APPS:

# settings.py
INSTALLED_APPS = [
    ...
    "utm_tracker"
]

and add both middleware classes to MIDDLEWARE:

# settings.py
MIDDLEWARE = [
    ...
    "utm_tracker.middleware.UtmSessionMiddleware",
    "utm_tracker.middleware.LeadSourceMiddleware",
]

The UtmSession middleware must come before LeadSource middleware.