The Quick Guide to Social Media Attribution Models
Since the inception of social marketing, social marketers have struggled to measure and prove the true value of social. Only 13% of CMOs say they can effectively quantify the value of social, a sentiment manifested a million different ways, but primarily in questions like:
- “What is the ROI of social?”
- “How much value does social actually drive for our brand?”
- “What was the ROI of that social post?”
- “Does social generate revenue?”
These questions are much more difficult to answer for social than for other digital marketing channels which are closed-loop systems that connect directly to revenue.
Social is not a closed-loop system. That’s why these questions have continued to plague us. In order to answer them, we need to be able to connect the value of social actions to real business goals:
- Business value
They key to accomplishing this is attribution, which is the process of assigning credit for business outcomes to marketing channels and campaigns.
Nota Bene: At Simply Measured, we’re investing deeply in attribution solutions to help you answer these difficult questions. We want to bring greater sophistication to your social measurement so that you can prove more tangible success, drive meaningful improvement in the business, and garner more resources for your team.
Attribution is a complicated topic with a lot of different approaches, but developing an understanding of attribution modeling will give you the tools you need to start answering those questions and more tangibly prove and improve the value of your social programs.
The Social Marketer’s Guide to Social Media ROI
The first thing to note about attribution modeling is that there are many different ways to do it. There is no “correct” way to do it. Everyone has a different opinion of the “best” attribution model, but only you and your colleagues can determine what makes sense for your organization.
Roughly speaking, there are three primary types of attribution models. You have options within each of these different approaches. We’ll walk through some of the most common here.
1. Single Touch
Single source attribution models assign all of the credit to a single touchpoint, and thus a specific source. These models are the most common, the easiest to use, and, unfortunately, also the least accurate.
Last-Touch: This model applies all of the credit for a conversion to the source of traffic for the session on which the conversion happened. This is the most commonly used attribution model, but also one that is deeply flawed.
Consider a scenario in which a consumer comes to your site for the first time…
…via a social post you’ve made…
— Neumos (@Neumos) April 8, 2016
…and your follower reshared:
— SEATTLEITE (@SeattleiteMag) April 13, 2016
Then they come back to your site the next day by typing the URL into their browser…
…and converting. In this model, that direct visit would get all of the credit, despite the fact that it was really social that drove interest. Last-touch is especially tough on social, given that social typically is one of the earlier touchpoints a consumer has before converting.
First-Touch: This model applies all of the credit for a conversion to the source traffic for the first visit a consumer ever has to the site.
Using the previous example, this model would give credit to social, but what if someone discovered your site via search…
…and then liked your brand…
…and subsequently came to your site seven times by clicking on your social posts…
…and THEN converted? All of the credit would go to that first search touchpoint.
Last Social-Touch: This model applies all of the credit for a conversion to the last social touchpoint a visitor had before converting.
We advocate that social marketers use this model so that they can maximize the amount of conversions attributable to social (which is historically under-attributed as a channel).
Now, obviously, this model is flawed much like those above, and we’re admittedly deeply biased towards social, but we want social marketers to be able to show their largest impact as a starting point before getting more nuanced about attribution modeling. This is not unlike Google advocating for a last AdWords-touch model (which it does).
Attribution modeling gets more sophisticated (and more accurate) by considering all the touchpoints a visitor has with you before converting. Multi-touch models attempt to understand the entire customer journey, instead of reducing attribution to a single touchpoint.
Equal Weight or Linear: This model divides credit for a conversion evenly across every touchpoint a visitor has before converting.
This model is more accurate because it understands that a single touchpoint is rarely responsible for a conversion. A single touchpoint is just a part of the larger customer journey. But the truth is, not all touchpoints are created equal. The first touchpoint and last touchpoint do typically drive greater conversion consideration.
Position-Based or Starter/Player/Closer: This model improves upon the Equal Weight or Linear model by giving a fixed amount of credit to the first touchpoint and the last touchpoint for a visitor, and dividing the remaining credit evenly among the other touchpoints.
Most commonly, 40% credit is given to the first and last touchpoints, and the remaining 20% is given to the other touchpoints, but many organizations will use custom values for each of these touchpoints.
Time Decay: This model recognizes that later touchpoints typically are lower in the funnel and are more likely to drive conversions. Thus, it divides credit across multiple touchpoints based on how recently they happened prior to the conversion.
For example, if a consumer first came to a site via search…
…and then three weeks came back twice within a few hours from two social posts…
…and converted, almost all of the credit would be shared evenly between those two social touches, with a small amount going to the first search touchpoint.
Custom: This model is used by organizations who have deep insight into their customer journey.
In this model, custom values are assigned to each touchpoint based on what the brand knows about the efficacy of different channels and journey stages.
This model is only possible with a deep historical knowledge of how consumers convert across many channels.
This type of attribution modeling eschews hard rules. Instead, it tries to use all of the available data that a brand has to dynamically determine how much credit each touchpoint should get.
In this model, there is no set weight for any touchpoint. The values for each touchpoint will change over time as the algorithm refines itself based on new data that comes in every day. This is ultimately the most accurate method of attribution, but also the most difficult to institute. It requires a large corpus of data and (typically) an expensive tool that must be implemented. We don’t suggest using this model unless you have significant experience with data modeling and attribution.
Now you have insight into the primary attribution used by marketers. If you don’t yet have a social attribution program in your organization, it probably makes sense to start with last-click or last-social-touch, so that you can start measuring social’s impact on business goals and effectively quantify.
After mastering that and getting a baseline understanding of how social is driving value, you can move on to more sophisticated models to obtain greater insight into the customer journey through social, and ultimately drive even more value from your social programs. Attribution is paramount to measuring and increasing the impact of your social programs.
Get everything you need to analyze the metrics that matter
Social is your Secret WeaponDownload
Brewster is Head of Attribution at Simply Measured, where he builds products to help marketers measure and increase the business impact of their social efforts. He joined Simply Measured through its acquisition of Inside Social, a venture-backed social analytics startup where he was CEO and Co-Founder. A reformed financier, Brewster worked in equity sales and corporate finance before realizing that zeroes and ones are way more interesting in code than financial models. He is a Seattle native and is passionate about Seattle sports teams and contributing to the startup ecosystem.