Attribution in SegmentStream — exploring new features in detail

You might ask: why would SegmentStream do that if we claim that attribution is dead? I will answer this question later, but first, let’s take a look at what this feature is and how it works.

How it works

In general, marketing attribution assigns conversions to a channel using different rules. For example, last non-direct click (LNDC) attribution means that the last touch in the user’s conversion path which is not from the direct channel will get the value. 

In this case, any non-direct channel will be considered significant for the LNDC attribution. Conversion will be assigned if this touch happened within the attribution window. 

LNDC attribution can be represented with the following settings: 90-day attribution window, any Source / Medium which is not (direct) / (none) is significant.

The new SegmentStream attribution feature allows users to change the significant traffic source and the attribution window to focus the analysis on specific marketing activities. 

For example, we can define all paid non-branded channels as significant to focus only on Performance Marketing activities. This will look the following way: 90-day attribution window, all clicks with defined Ad Platform but where campaign does not contain brand.  

Report will show the conversion if it had a last click from a significant channel within the attribution window, even if the click was outside of the report period. Below you can see an illustration for last paid non-branded attribution. A user clicked on a Facebook Ads, on a Google Ads branded campaign and converted. Attribution assigns credit to Facebook, as it is the last paid non-branded activity within the attribution window.

You can create your own attribution by clicking on the Plus sign. 

The last click from the significant source within the attribution window will get the credit. 

Why do we need it?

If the attribution is dead, why would we still want to use it? Indeed, this new feature will face the same limitations as classic attributions when it comes to broken cross-device, cross-browser, and cross-cookie customers’ journeys. For this reason, the attribution-based analysis is not recommended as a reliable source of data. However, it still gives a valuable perspective on channel performance. This new functionality allows analysing data through different lenses and provides additional flexibility for performance marketers.

To analyse channel performance, we recommend looking at SegmentStream Modelled Conversions as an indication of traffic quality and additionally at channel-specific attribution.  

Use cases

Case 1: Focus the analysis on significant channels. 

Create a report with attributing value to specific channels only. Below is an example of an Overall Marketing Mix with both paid and non-paid channels where the Ad Platform is not set. Here, we apply three attributions with the same 90-day attribution window:

1. Classic Last non-direct click (aka LNDC), which values non-direct channels;
2. Last paid click, which values all paid activities;
3. Last paid non-brand click, which values paid non-branded activities.

Of course, the total amount of conversions will stay the same, but the distribution is different. Paid attributions give less credit to non-paid channels. Interestingly, Paid Non-Brand attribution valued Google Ads much less than just Paid attribution, and this is because Branded activities are usually very powerful in Google Ads. This way we can choose the perspective for the analysis.

Similar can be done for any channels: paid, paid non-brand, Facebook, Google Ads, TikTok, or even email.

Case 2: Replicate Google and Facebook attributions inside of SegmentStream to have the data in one view.

Of course, SegmentStream is not able to replicate what Google Ads and Facebook Ads will show in their reports because they also apply additional modelling. Facebook Ads uses post-view attribution, Google Ads has data-driven attribution. But the new feature allows us to replicate post-click-based results to analyse what picture will Ad Platform get.

Below is an example of a Facebook-specific report with an applied Source / Medium filter. Here, we apply two attributions:

1. Classic LNDC, which values non-direct channels, 90-day attribution window;
2. Facebook Ads last click, which values Facebook activities and uses a standard 7-day attribution window.

Facebook Ads attribution will assign more credit to the channel which can be seen in the Total row and allow us to analyse more data on deeper levels. 

Case 3: Analyse experiments better. 

During an experiment when we compare standard optimisation with SegmentStream Modelled Conversions optimisation, it is valuable to strip away all other channels and campaigns to focus on the results brought by the Control and Test branch. The new feature allows us to create experiment-specific attribution.

Find an example for an Experiment specific report with an applied Campaign filter below. Here, we apply two attributions:

1. Classic LNDC, which values non-direct channels, 90-day attribution window;
2. Experiment preferred last click, which values clicks from the experiment campaigns;

Experiment Preferred attribution gives more credit to the experiment campaigns despite their place in the path to conversion and allows us to focus our analysis on the experiment results. 

To sum up

The new attribution feature gives additional perspective and allows us to analyse data from different points of view. The main advantage is the ability to focus on a specific marketing activity to make data-driven decisions.