A look at attribution models + predictive analysis to measure influencer marketing ROI.
Any brand involved in influencer marketing today is trying to determine if influencer marketing is working. In 2014, an astounding 38% of marketers admitted they were yet to adopt an attribution model (emarketer) for their marketing efforts, let alone their influencer channels.
The good news is that the times are changing. Now, in 2017, we’re learning that 57% of marketers will allocate a large portion of their time to cross-channel attribution and measurement. The methods discussed in this tip-sheet cover attribution models that our clients use in order to measure the success of each influencer in their campaign.
Measure Each Influencer’s Success Individually
When working with influencers, you want to make sure that each influencer’s ability to drive conversions is measured individually. This allows you to test the success of different audience verticals, behaviors, and affinities within each influencer’s audience. After you know that information, then you can double down on the good influencers, and move on from the not-so-good. Think of it the same way you’d treat audiences on linkedin/facebook ads, or ad groups on adwords. So, whichever attribution model(s) you use, treat each influencer’s performance as it’s own.
Use Last-Click Attribution
This is one of the more popular, albeit least fair, attribution channels that marketers use. This is because 100% of the attribution is applied to the ad that drove the last click. In the case of influencers, if you have 4 influencers who launched a video, look at how many immediate sign-ups each influencer drove over the course of the videos launch (typically 30-40 days). The problem with last-click attribution is that it doesn’t factor in the longer-term effects of the campaign.
Use Predictive Analysis (With Last-Click Attribution)
Use predictive analysis to measure the ROI each influencer will have 3, 6, and 12 months down the road. Many of our clients have found that this approach allows them to see the long-tail success they have with influencers, attribute a better value to each influencer’s content, and then factor exactly how much they can spend each month on influencers.
Add predictive analysis to the last-click attribution model you’re using to better measure the true impact each influencer had on your conversion goal. This is how some of our brands do this:
Based on the last-click model mentioned above, you know the direct CPA (i.e. click-to-purchase) you have from each influencer (see bottom right).
To take a step further, calculate a media value for each piece of content. Do this by calculating 1) the order value of each sign-up through Google Analytics, or a similar tool 2) and an estimation of how many times they will purchase again over the next 12 months (which is titled “long-term purchases” in the chart below).
HubSpot also has a few methods of predictive analysis that you can use to help attribute “long-term purchases”.
Run a Survey
Run a survey on your site. One of our clients ran a survey on their homepage and shopping cart simply asking how people found out about them and learned some surprising results. The results of 93 surveys showed that for every direct sign-up that came from the click-through link on the YouTube video, three more people had discovered them from YouTube, but hadn't used the click-through link to purchase. This was after running a campaign with 11 influencers, spanning over 1.8mm views on YouTube. While it didn’t allow them to attribute success to specific influencers (beyond their last click model), it did show that YouTube is a great channel for them.
With a tool like SurveyMonkey you can embed a survey on your website. Setup your survey to simply ask “How did you find out about us?” then include selections for every marketing channel you have in your mix. We recommend that when providing options for social media channels, you separate each (i.e. YouTube, Facebook, Instagram) instead of combining them (YouTube/Facebook/Instagram or just Social Media).
Obviously last-click attribution is an important model to have for all channels, however, it shouldn’t be the only one. A multi-channel attribution model can help measure the halo-effect influencer content has on other, sometimes more direct channels like FB ads.
A multi-channel attribution model assigns the value to multiple channels that a consumer touched in their customer journey. A few of those models are below:
We have found that influencer content contributes to the success of other marketing efforts as well; not just improving brand lift, but the success of other marketing channels conversions. However, there are some gaps currently in influencer marketing that make it difficult to fully attribute conversions in a multi-channel attribution model, especially if you’re looking at clicks. Remarketing is a key component of the multi-channel attribution model and while you can still attribute every conversion to a brand-owned landing page and measure its impact on other channels, there are limitations to view/impression attribution on the content itself.
Being that influencers typically post content on their owned properties, social media juggernauts like YouTube and Facebook don’t actually allow brands to drop a pixel on their influencer partner’s content. There are some workarounds for this, so, please let us know if you’re interested in learning more.
As social media platforms continue to grow, the value of non-click, multi-channel attribution modelling only increases. New platforms like Instagram and Snapchat aren’t oriented on click-based marketing. So, if not via clicks, how are brands attributing social platforms, and thus influencer campaigns? With video, marketers are now placing great value on “view through” or instances where consumers see ads and don’t click through.
As you can see from the chart above, in 2016, 46% - nearly half - of marketers said they attributed at least 50% weight on viewthrough conversions.