INFLUENCER MEASUREMENT: THE GOOD, THE BAD & THE TOTALLY FAKE

It’s ok if you missed yesterday’s webinar (we know you wanted to come but that last minute call popped up) because we have the recording! Listen to a CLEVER panel of influencer marketing experts—Kristy Sammis (Founder), Alison Kimble (Product Manager), and Brandon Farley (Director of Analytics)—walk us through their take on influencer metrics, how to spot fake followers, and the right way to measure program success. Catch up on all the action here.

We received many questions during the session that we wanted to share with you as additional reference, below. You can also always download our white paper here or reach out to Alison directly.

IMPRESSIONS

Q: What’s CLEVER’s formula for determining actual impressions?

The CLEVER formula is based on both social media algorithms and influencer-reported-and-verified data:
10% of potential impressions on blogs (monthly page views)
25% of potential impressions on Instagram
10% of potential impressions on Facebook
15% of potential impressions on Pinterest
5% of potential impressions on Twitter

The number of video views on YouTube (counted at the 30 second mark, not estimated)

Q: If actual impressions are determined by a formula, wouldn’t it be possible to calculate actual impressions at the onset of the program (vs. at the end)?

Almost! Once you have selected your influencers, it is possible to calculate your potential impressions (and therefore actual impressions) on most channels, EXCEPT for Twitter and YouTube:

* Potential impressions on Twitter include secondary impressions. Primary impressions represent the followers of your program influencers. Secondary impressions represent the followers of those who retweet your program influencers. So in addition to your program influencers’ following, the potential impressions also include the followers of those who retweet program posts, which can’t be predicted ahead of time.

* On YouTube, actual impressions are not estimated, but rather, the number of video views.

Q: If all the influencers that you are working with are able to share screenshots of how posts perform, wouldn’t that be the most accurate number?

We only recommend self-reporting for small programs with a few influencers, if at all. Here’s why:

Unreliability—We ask our network applicants to share information about their reach, and when we gain access to their Google Analytics, we see how unreliable self-reported data can be. When influencers self-report program actuals, we have no way to check.

Scale—you may be able to get a small group of influencers to share their actual post data and collect via email (5–10 influencers). The larger the group, the more issues that will arise such as influencers not following directions or an inbox nightmare.

Lifetime Value—A screenshot is only one small snapshot of the overall reach/engagements a post may garner. There is limitless potential value influencer content can bring and studies show that posts actually generate 2x impressions after the campaign has ended.

INSTAGRAM STORIES

Q: If collecting/analyzing metrics for Instagram posts and comparing campaigns is still somewhat the Wild Wild West, how do you look at and value Instagram story metrics?

Depending on the client’s goals, we would look at success in different ways! Sometimes clients are looking to put their product in relatable, real-life scenarios, in which case we would look at the creativity and reusability of the content. If they are looking for awareness, we would look at follower counts and engagement. If they are trying to drive to a landing page, we would look at clicks.

Q: How are you measuring Instagram stories and quantifying when they string several in a row? Also when they add the stories to a highlight reel, do you think any additional impressions should be added for that?

Currently, because third-party vendors do not have access to Instagram Story metrics, we keep activations small and have influencers self-report. We look at potential impressions (the number of Instagram Followers times the number of stories, not clips), unique actuals, engagements (replies, sticker taps, location taps), and clicks (if they have a swipe-up link, which is available to most influencers with 10K+ followers).

We don’t count additional potential impressions once a story is added to a highlight reel, because we consider it to be the same post.

ENGAGEMENT

Q: Do you count video views toward engagement? How does CLEVER define a video view?
We have counted views as impressions (because they represent eyes) and engagements (because they show interest), depending on the client’s request. Regardless of that distinction, views should be separated out from other impressions to evaluate how gripping a video is. They show a certain level of interaction, beyond scrolling past a post on a feed.

Note: social channels define video views differently.
A view is recorded after:
30 seconds or more on YouTube
3 seconds or more on Facebook
3 seconds or more on Instagram
2 seconds or more with 50% of the video in view on Twitter

Q: What’s CLEVER’s aggregate engagement guarantee and can you explain how you got there?

CLEVER has developed a proprietary formula based on thousands of past programs. Our formula considers factors such as vertical, number of influencers, channel, type of content, etc., to calculate engagements based on similar programs. You can’t compare a recipe series to a campaign about taxes or, say, female incontinence!

We would be more than happy to help you set benchmarks for your specific influencer program!

AUDIENCE DATA

Q: I read in a recent newsletter from Clever that you measure influencer audiences based on publicly available data from Twitter. Do you find that this gives a true indication of audience, especially given that so many influencers are not paying as much attention to Twitter?

We do! Instagram recently made their follower data inaccessible, but prior to that it was possible to analyze audiences of public profiles, where we discovered that — even if an influencer is not very active on Twitter — audience data on Twitter is highly correlated to and representative of audiences on other channels.

FAKE FOLLOWERS

Q: How do you get the follower cluster again?

The clusters are made using the Twitter API to fetch the list of profiles that every audience member is following. This can then be run through a tool like Gephi which identifies connections between the audience members. Gephi can also be used to create the visualization with colors for different clusters. We’ve found this approach is also very useful for brand audiences to understand segmentation of followers.

Q: How can you manually figure out the follower geography?

Though it may not qualify as “manually,” the best way to identify audience geography is to use the Twitter API to “Get-Info” on each follower of an influencer. The location will be one of the fields returned. The challenge then is to standardize the unstructured data. Users can enter any text they want in the location field so the data must be turned into structured form (i.e., NOLA —> New Orleans, LA).

EMV

Q: How does CLEVER calculate EMV

We use the Ayzenberg Group’s 2017 Earned Media Value Index (download their index here), unless a client has their own formula (which we love!).