Fortify Your Measurement Toolkit with Rule-Based Attribution

October 16, 2017
Contributor: Chris Pemberton

Use four questions to maximize value from rule-based attribution analysis

When a retired couple planned their dream vacation to Europe recently they touched the boutique hotel where they ultimately stayed many times throughout their buying journey of researching and finally booking the room. They clicked on a banner ad, sent an email query and even liked the hotel on Facebook before they finally decided it was the right choice. For the hotel’s marketing analyst seeking to better understand the customer journey of similar customers, the task of attributing credit to the various channels was a complex and confusing exercise. Should she credit the banner ad more or the Facebook Like? What role did the email play since it came in the middle? Should she use first-click, time-decay or open-assist-closer? What is the Opener-Assist-Closer model?

“Use a methodology guided by four questions to ensure you obtain value from your rule-based attribution analysis,” says Lizzy Foo Kune, Principal Research Analyst, Gartner for Marketers.

Attribution analysis approaches

Attribution modeling can be a prohibitively time- and resource-intensive activity. However, advanced, algorithmic approaches aren’t the only methods you can use to understand how various events drive your audience to action. There are two approaches:

Rule-based attribution analysis

Basic attribution methods assign fixed weights to events along a sequence leading up to a success event. This approach to attribution analysis uses rule-based heuristic models such as linear, time-decay and opener-assist-closer methods. It does not depend on statistical algorithms, nor does it necessarily require collecting data at the user level. Most marketing analysts have the skills required to perform rule-based attribution analysis in-house. Because the rules are straightforward, this analysis is easy to understand and explain to internal stakeholders.

Multi-touch attribution modeling

Multi-touch attribution applies advanced statistics and machine-learning algorithms to estimate weights for events. These models are highly complex, time- and resource-intensive, and analytically demanding. The costs associated with advanced attribution typically reach into the mid six figures (and more), as marketers must align internal support and management requirements with professional services and software.

While this method is accepted to be more data driven and, thus, more accurate, it is limited by its complexity. Models can be difficult to understand and act upon if you are not a data scientist.

Example rule-based attribution methods include:

  • First or last-click: Allocate all credit for a conversion to either a first or last touch along the path
  • Linear: Attribute equal credit to every event prior to conversion
  • Time decay: Attribute greatest credit to final event and decreasing credit moving farther away from conversion
  • Opener-Assist-Closer: Assign equal credit to first and last touch and equal credit to channel in between

Ask four questions to guide your approach
  1. What’s my attribution goal?
    The goal you identify informs what models you’ll select for your rule-based attribution analysis.  A linear model is good for understanding the customer journey and would be appropriate place to start for the boutique hotel marketing analyst trying to understand the path to purchase for a retired couple segment.
  2. What data can I realistically obtain?
    The key word is realistically. Make use of whatever data you have available and recognize that activities like in-store showrooming and word-of-mouth discussion play a role in driving conversion, but they can’t realistically be represented in rule-based attribution models.
  3. What’s the right lookback window?
    A lookback window is the period of time prior to conversation that an attribution analysis considers events and allocates credit among them. Your lookback window should align with your market’s purchase dynamics. Start by understanding the purchase cycle of your product.
  4. How will I validate the analysis?
    Use A/B split tests to refine your analysis, examining lift in your test group when compared with your control group. Continue to test and refine your analysis and compare results from different points in time. Look to establish strong coordination between your marketing organization and your media team.


“Use this research to add simple, easy-to-use rule-based attribution analysis to your marketing measurement arsenal,” says Foo-Kune.

Get Smarter

Client Research
Gartner for Marketers clients can read more in When and How to Use Rule-Based Marketing Attribution Analysis by Lizzy Foo-Kune.

Marketing Transit Map
Explore the Digital Marketing Transit Map to see what other opportunities may bolster your marketing strategy.

You may also be interested in
“I use Gartner to bolster my confidence in decision making.”

Stay smarter.