ellipsis flag icon-blogicon-check icon-comments icon-email icon-error icon-facebook icon-follow-comment icon-googleicon-hamburger icon-imedia-blog icon-imediaicon-instagramicon-left-arrow icon-linked-in icon-linked icon-linkedin icon-multi-page-view icon-person icon-print icon-right-arrow icon-save icon-searchicon-share-arrow icon-single-page-view icon-tag icon-twitter icon-unfollow icon-upload icon-valid icon-video-play icon-views icon-website icon-youtubelogo-imedia-white logo-imedia logo-mediaWhite review-star thumbs_down thumbs_up

How to choose the right attribution model

How to choose the right attribution model Ted Rooke

Like most successful marketers, you are running multi-channel campaigns in order to maximize the performance of your digital marketing efforts. As you begin to read results, you are starting to see one channel out-perform another, and you begin optimizing your budgets into the channels that work best. Then it hits you: How are you crediting conversion activities by channel? Are you de-duping conversion events across all media channels or are you double-counting the same event across two or more channels?  Which ads are getting the credit? Is search getting all the credit because it's the last touch point? Should it get the credit? Is direct load really driving such a big share of actions? 

If you are beginning to ponder these questions (and you probably are -- it's why you're reading this article), then what you're really saying is, "I need an attribution model!" Attribution models are tools that allow marketers to assess the relative value of each digital marketing channel in delivering a goal action (such as a sale, lead, or video view) in relation to each other, not in individual silos. By understanding how each channel impacts one another, you can best assess where to invest your marketing energy and dollars to drive success.

There are many attribution models to choose from. These are four primary models:

Last click attribution

All of the conversion event value goes to the placement delivering the last click. This is the most common model employed, and it's the default method of attribution we have used since conversion measurement began in the early days of internet marketing.

First click attribution

All of the conversion event value goes to the placement delivering the first click. Basically the same as last-click, but it delivers credit for the initial engagement and not the last.

Equal weighting attribution

Every touch point leading to the conversion gets an equal share of the credit. I like to call this the "good parent" model because just as we love each of our kids equally, we "love" each of our channels equally.

Custom credit attribution

This model takes multiple factors into account and determines the proper credit share to give to each placement based on customized valuation of various factors impacting media performance. This becomes the basis for most algorithmic-based models, and hones in on the unique levers that drive user action forward.

In order to begin quantifying attribution, you must first determine the model that best suits your needs. Your specific approach to digital marketing will help identify which model addresses your needs. Below are some common examples of various digital marketing types and the corresponding attribution model for each.

The novice or small business owner: Last click attribution

For marketers new to digital marketing, and especially for the small business owner who is plying just one or maybe two digital channels within the mix, last-click attribution is still your best bet for understanding channel performance. Your investment (both in terms of time and dollars) probably does not warrant getting deep into performance analytics just yet. By understanding the digital channel, or specific placements (such as a keyword, an ad, or tweet) that drive desired actions, you are probably ahead of your competition already. Capitalize and maximize your performance using last-click attribution, and you can go far and beat your competition.

The branding marketer: Equal weighting attribution

For brand marketers, conversion measurement is a nice add-on, but it's not the primary engine that drives media value. These marketers tend to focus on traditional metrics such as reach and frequency, message recall, and purchase intent. However with digital marketing, even brand marketers can't escape from conversion measurement, as performance tracking is ingrained into all digital campaigns. For marketers wishing to track conversions while optimizing to traditional metrics, this model allows for each channel to get equal conversion credit, thus making your branding metrics more visible and the more evaluable differentiators in performance measurement.

The social media marketer: First click attribution

For social media marketers, the goal tends to be more about engagement, and less about actions such as direct sales. Also, as social media speaks to consumers who are "friends" or "followers," these consumers are pre-disposed to liking your brand already. As such, these consumers may also be in your email database, and they will definitely find your site via direct load or organic brand searches. When they convert to a sale, social media is probably not getting the credit for the conversion. If instead we leverage first-click attribution, we can begin to see which channels are driving the initial interest, something that social media is especially good at. First-click attribution helps us understand what channels motivate consumers to begin an engagement with a brand, and social media is a strong channel for motivating action.

The direct marketer: Custom credit attribution

For the direct marketer whose goal is maximizing conversion performance of each dollar spent, this model provides the most value in showcasing channel and placement level performance. As the name implies, this model is fully customizable to your needs and each channel or placement's performance is graded on a scale based upon:

  • When each touch point occurred (both in terms of how long before the conversion event, as well as in what sequence the touches occurred)

  • What type of interaction occurred (i.e., an impression, click, video view, or rich media interaction)

  • What size and type of media it was (i.e., rich media expandable, text link, or a video)

This model can be very granular, and as such each placement can be uniquely optimized based on the value it does or does not deliver. Cross-campaign optimizations based on a customized attribution model will net you the most returns for your integrated campaigns.

So what type of marketer are you? By leveraging the right model for your specific needs, you can best showcase your efforts, while also providing valuable cross-tactic insights into your digital performance.

Ted Rooke is vice president of media services at Response Mine Interactive.

On Twitter? Follow iMedia Connection at @iMediaTweet.

"A picture of the back" image via Shutterstock.

Ted Rooke is vice president of media services for Response Mine Interactive (RMI) and is charged with the expansion of the firm’s media expertise across the digital services division. Prior to RMI, Terence executed marketing initiatives...

View full biography


to leave comments.

Commenter: Dino Mahi

2012, June 23

It is interesting to note your perspective on the utility of Custom Attribution Modelling only for Direct Marketing Scenarios.

Increasingly, we find clients coming to us to help them use such custom modelling for 'upper funnel' channels or 'Digital Outreach' Activities such as Social Media (paid and owned), Blogs, Webinars, Videos etc. Some of these clients spend 6 figure monthly budgets on such activities and increasingly their challenge is to use Attribution Modelling for identifying channels that drive their upper funnel metrics

That being said, I think another dimension to consider (over and about the type of marketer and specific business need) before deciding a modelling approach would be to do an internal assessment of their data and governance maturity. Custom Attribution Modelling can provide powerful insights but may not be for everyone from a purely operational standpoint

Commenter: Erik Carlson

2012, June 20

This is probably tied more closely with the custom credit attribution, but what about the horseshoe model that gives more credit to the first and last touch while acknowledging the middle touches?