Attribution Models

Our Attribution Models assign credit for conversion events to relevant marketing touchpoints. Northbeam offers 7 different attribution models, which are divided into two categories: Simple Attribution Models and Multi-Touch Attribution Models (MTA).

• Simple Attribution Models
  • First Touch
  • Last Touch
  • Last Non-Direct Touch
• Multi-Touch Attribution Models
  • Linear
  • Clicks-Only
  • Clicks + Modeled Views
  • Clicks + Deterministic Views

---

​

Simple Attribution Models

In Simple Attribution Models, all credit is given to a single touchpoint. These are legacy models that you may be used to seeing. Here are our simple models:

• First Touch
• Last Touch
• Last Non-Direct Touch

​

Multi-Touch Attribution Models (MTAs)

In Multi-Touch Attribution Models, credit is divided across multiple touchpoints. Here are our multi-touch models:

• Linear
• Clicks-Only
• Clicks + Modeled Views
• Clicks + Deterministic Views

​

Summary of Northbeam’s Attribution Models

---

​

First Touch

When to Use First Touch: Anytime you want to give credit to the very first touchpoint (ie, subscriptions, or gut checking MTA data). In this model, full credit is given to the first touchpoint of the customer journey.

---

​

Last Touch and Last Non-Direct Touch

When to use Last Touch: This is the least-used model in our system, but its great for reviewing lift in direct traffic or performance over time. When to use Last Non-Direct Touch: This model compares consistently well with Google Analytics’ “Last Click” and is a strong number to compare to ground data in a familiar environment. It is also great for monitoring Bottom of Funnel performance. In these models, full credit is given to the last touchpoint of the customer journey, but Last Non-Direct Touch does not give credit to Direct if it is the final touchpoint. As an example, let’s take the following customer journey that resulted in a $90 order. This model is great to understand the impact of your lower funnel channels — Branded Search, Organic Search, and Email/SMS to name a few.

---

​

Linear

When to use Linear: Typically for channels that are in the middle of the funnel (i.e. organic social). Linear divides credit equally across all channels.

---

​

Clicks-Only

When to use Clicks-Only: Great for day-to-day decision making. This will be more conservative, but makes for a very useful prospecting compass. This proprietary model is similar to linear, where credit is divided equally between all touchpoints.

​

You may be asking yourself

• “What’s an Upper Funnel Touchpoint?”
• “What’s a Lower Funnel Touchpoint?”
• “How does the credit allocation work?”
• “Why isn’t credit given to Lower Funnel Touchpoints?”

Let’s dive in to your questions:

​

Lower Funnel Channels

• Direct Visits (Labeled as “Organic” in the dashboard)
• Organic Search
• Paid Branded Search
• Email or SMS

​

Upper Funnel Channels

• Any channel that is NOT a lower funnel channel (ex. Paid Social, Paid Search (excluding Brand), PMax, Organic Social, Affiliate, Influencer, Display, etc)

​

Clicks-Only: Example 1

Let’s take the following journey and assume we’re on the Accrual accounting mode, where credit is given to the following touchpoints.

1. Jan 1 - Facebook Ad click
2. Jan 2 - TikTok Ad click
3. Jan 3 - Google Branded Search Ad click, resulting in a $90 purchase.

Explanation: Although the Branded Search pushed the conversion, Branded Search does not get credit with Clicks-Only because this is a Lower Funnel Channel. Therefore, credit is distributed equally to Facebook and TikTok.

​

Clicks-Only: Example 2

Here’s another example. Let’s say the entire journey is made up of Lower Funnel Touchpoints:

1. Jan 1 - Google Branded Search Ad click
2. Jan 2 - Direct Visit
3. Jan 3 - Email click, resulting in a $90 purchase.

Explanation: Because all the touchpoints are Lower Funnel, credit is distributed evenly. Credit is only pulled when an Upper Funnel touchpoint is recognized in the journey.

---

​

Clicks + Modeled Views

Our Northbeam attribution models leverage machine learning to ingest your data and stitch together customer journeys so you can start accounting for view-through data when analyzing attribution decisions. Clicks + Modeled Views provides directionality on which ads within a channel are driving the most results and which ones are ineffective. Utilizing the Clicks + Modeled Views model in conjunction with Northbeam’s Clicks-Only attribution model can help you fill in any “click-through” gaps, particularly for channels like Social and Video that are highly view-oriented.

​

What exactly is the Clicks + Modeled Views Attribution Model?

Clicks + Modeled Views is Northbeam’s proprietary machine learning model that accounts for impression and ad engagement data when making attribution decisions. This model looks to split attribution credit on a multi-touch basis. it reduces attribution credit to any direct, organic search, paid branded search, or email/SMS touchpoints in the customer’s journey, and gives appropriate credit to Views at any point throughout the customer’s journey. We built this model because unattributed traffic is not helpful for marketers when trying to analyze which channels are powering their highest-converting paths. Although our Clicks-Only model is our most accurate model because clicks are hard data, there are many conversions that won’t be captured: clicks don’t represent the entire picture. We understand that many customers eventually convert via direct visits after watching a TikTok, YouTube, or Facebook ad. This is why our Clicks + Modeled Views model tries to assign proper credit to any ad views throughout a conversion path. This can help make your data actionable by giving you a more complete picture of what’s going on with your campaigns.

​

How does Clicks + Modeled Views work?

At a high level, our model tracks engagement metrics (impressions, video views, etc.) and looks for correlations between those metrics and any corresponding organic spikes. Our model then makes an educated guess using machine learning to assign organic traffic and conversions to the ad campaign that were mostly likely to prompt that behavior. Please note the attribution window for Clicks + Modeled Views is 1 day view only, regardless of which option you select.

​

When do you use Clicks + Modeled Views?

Keep in mind, our Clicks + Modeled Views model takes 25-30 days to “learn” from the historical data collected by our tracking pixels. One of the most common questions we get is: Which attribution model is the best one to use to make media buying decisions? The answer is: Clicks + Modeled Views offers directionality of which views are driving incremental revenue. Look for big gaps between revenue reported in Clicks Only vs. Clicks + Modeled Views to better understand what assisted organic conversions. We recommend using the Model Comparison tool in Northbeam to do this via the hamburger icon on the top right of your Northbeam dashboard next to the maintenance alerts icon. Let’s say you notice that Clicks + Modeled Views captures a revenue attributed to YouTube ads that weren’t captured in Clicks Only. This could mean that YouTube ads are driving revenue, but we should try and validate these results before dramatically scaling spend. One way to do that is to conduct post-purchase surveys and track where incremental revenue is coming from. You could also slowly ramp up spend on YouTube while keeping an eye on blended revenue numbers to see if they’re highly correlated. If post-purchase surveys are also indicating YouTube as a leading channel, then you can be reasonably confident that YouTube ads are having a positive and additive impact on your business.

​

What is the benefit of looking at the Clicks + Modeled Views model?

Ultimately, the Clicks + Modeled Views can help you better allocate your marketing dollars. You spend less on conversion ads if more people view them, don’t click, and then later purchase. It is great for understanding the full impact of prospecting performance, and also useful for monthly reporting on prospecting channels.

---

​

Clicks + Deterministic Views

Clicks + Deterministic Views (C+DV) is Northbeam’s most advanced multi-touch attribution model. It builds on the Clicks-Only framework and extends beyond Clicks + Modeled Views by incorporating platform-verified impression and view data from participating ad platforms. While Clicks + Modeled Views uses machine learning to infer the impact of ad views, Clicks + Deterministic Views only assigns view credit when an impression is explicitly verified by the ad platform and deterministically linked to a conversion. At a high level, this means:

• Combines first-party paid click data with platform-verified ad views to measure the full impact of paid media.
• Uses deterministic matching only. Views are credited only when explicitly verified by the ad platform and linked to a conversion.
• Excludes lower funnel touchpoints (Direct, Organic, Branded Search, Email, SMS) from receiving credit when paid clicks or verified views are present.
• Assigns view credit at the platform level, then distributes that credit across individual verified view touchpoints from that platform.
• Provides clearer insight into view-heavy channels like Paid Social, Video, and CTV that last-click models miss.

When to use Clicks + Deterministic Views:
Use this model when you want the most accurate view of upper funnel impact and when participating ad platforms are enabled for deterministic view matching.