The Value of a Twitter Follower


by Puja Ramani and Sampada Telang

What we learned from applying the life time value analysis as well as engagement funnel analysis techniques for Twitter.


(Note: if the graphics below are difficult to read, you can download a PDF version of this post here).

Our goal was to estimate the value of a Twitter follower for various publishers on Twitter. ‘Publisher’ refers to any entity with a business objective on Twitter (examples: Amazon, TechCrunch, celebrities or even local businesses). Given that Twitter primarily earns its revenues by charging publishers (via promoted tweets) as opposed to individual profile owners, we will refer to Twitter’s publishers as its ‘customers’.

Our focus was applying the tool not to Twitter’s customers (i.e., publishers) directly, but to Twitter’s customers’ customers (i.e., publishers' followers) in an attempt to quantify how much value Twitter was creating for publishers. We studied a very small set of publishers, so these findings shouldn’t be taken as definitive averages but rather, used directionally.

Value of a tweet: A key source of interaction between the publisher and its Twitter followers is the ongoing stream of tweets that the publisher sends out via its profile. After numerous conversations with Twitter’s publishers, we discovered that the primary source of value created on the Twitter platform is via ongoing tweets, not promoted tweets. Thus, value of a tweet is central to determining value of a follower.

Hence, our analysis was anchored around determining the life-cycle of a tweet, the engagement that an average tweet creates and the total clicks it drives.

1. Summary of key learnings

1.1 Engagement Funnel


For Twitter’s publishers we define the engagement funnel as the total engagement created by a tweet — through follower retweets, favorites etc. and the rate at which that engagement drives clicks.

Exhibit 1: Engagement funnel

We analyzed the conversion funnel for e-commerce and content publishers and discovered the following:
  • Tweet click-through rates are defined based on engagement instead of impressions: The standard definition for ‘click-through rate’ is Total number of clicks divided by Total number of impressions. However, it is difficult to reliably estimate tweet impressions due to the chronological ordering of tweets on a follower’s Twitter home page. The only evidence of impression is the engagement or interaction a follower has with a tweet. Thus, click through for tweets = Total clicks divided by Total tweet engagement.
  • Most engagement on Twitter is in form of clicks and not retweets, favorites etc.: Clicks comprise ~90% of total engagement on Twitter. Surprisingly, this is true not only for e-commerce but also for content publishers. 
  • Content publishers get a higher total number of tweet clicks: This is not surprising, because most followers of content sites want to click on their tweets to stay abreast on news etc., but might be less likely to click on a “deal” tweet from an ecommerce publisher, unless they are really interested in purchasing that item.
  • Follower quality matters more than quantity: We noticed that a larger number of followers doesn’t necessarily result in more clicks. In fact, a lower number of high quality, relevant followers could generate a higher click rate. 
  • Clicks per tweet are dependent on a number of factors such follower quality, tweet quality, and tweet frequency.
  • Publishers range from sophisticated (eg- Amazon) with separate handles for each category of products) to barely Twitter literate -- companies that use the same handle for customer service, marketing and specific deals — resulting in poor targeting, low clicks and high unfollows.
  • The key to improving click-through is effective targeting i.e. create specific handles with a homogenous set of tweets targeted at a specific set of followers (based on interests eg- tech tribe/early adopters or demographics – eg- young women

1.2 Life time value analysis
  • Life time value of a follower depends on the monetization technique of the publisher – ecommerce, SaaS or ad impressions.
  • For each publisher, life time value isn’t as simple as 1/churn rate x value per day, but as the methodology below highlights, needs to take into account several other factors.
  • There is a lot of information asymmetry in social media. There is an opportunity for Twitter and publishers to collaborate, sharing data and analytics, to jointly understand the value of Twitter. More clarity on value creation will certainly enhance Twitter’s ability to capture value (i.e., monetize its user base). At the same time, it will help publishers to use Twitter’s platform much more effectively, thereby increasing their returns.
  • Similarly, an opportunity exists to extend the Twitter API further to understand vital engagement stats. However, currently the Twitter search API is limited and can only provide a subset of tweet data, making it difficult to build tools on top of it.

2. Methodology and analysis conducted

2.1 Publisher classification


We classified publishers into four buckets based on goals on Twitter. For the purposes of our study we only focused on E-commerce and Content publishers.

Exhibit 2: Publisher classification




2.2 Value of a tweet for e-commerce and content

We defined the value for e-commerce players as shown below.

Exhibit 3: Value of a tweet




2.3 Sample engagement funnels

We worked with Twitter to obtain tweet engagement data for select set of publishers. Below are conversion funnels for e-commerce and content. We obtained click data from bitly wherever possible, but this analysis hinges on the validity of that click data. Based on our observation, these funnels are representative of the category. For example, independent of number of followers, the 4 e-commerce publishers we studied, on an average, generate less than 50 clicks per deal tweet.

Exhibit 4: Sample engagement funnels




2.4 Value per deal tweet for E-commerce publishers

To determine the value for e-commerce players – we deconstructed the value per tweet as follows. We vetted the math below by speaking with multiple publishers.

We realized that publishers capture the click and conversion data via Google Analytics but were unwilling to share it, thus we obtained click data from publicly available source such as bitly.

Exhibit 5: Value from a deal tweet


Below is a sample analysis for a daily deals site. For each item above, we used a combination of various inputs – example we obtained click data from bitly, conversion data from analytics company presentation at Web 2.0 and margin data from company 10 Ks. Deal size and Tweet rate were obtained from tracking the publicly available tweet data.

Note: Estimates below are conservative as we’ve only looked at deal tweets and value created through immediate deal fulfillment. What we haven’t included is as follows –
Purchase of additional items beyond those linked to the deal on the tweet
Additional followers obtained due to retweets, favorites etc.
Value obtained through marketing campaign tweets and customer service tweets

Exhibit 6: Sample analysis for a daily deal site





The value is different for different e-commerce players and local businesses due to differences in business models. Example: For every dollar of sale, Amazon needs to factor in inventory management and delivery costs but Groupon does not. Thus, their margins are very different.

Hence, we built out sensitivity tables for different e-commerce players

Exhibit 7: Value from deal tweets to different types of e-commerce publishers on Twitter


Comments

Popular posts from this blog

Quiz Time 129

TCS IT Wiz 2013 Bhubaneswar Prelims

The 5 hour start-up: BrownBagBrain