By Chuck Currin and Arvid Tchivzhel, Mather Economics
Audience segmentation of their readers is a relatively new undertaking for publishers. The publishing business model, historically, has relied heavily on advertising revenue, and the principal audience information that a publisher possessed was focused on characteristics valuable to their advertisers. As subscription revenue has become half or more of total revenue, the return on audience analytics and segmentation has grown considerably. The growing share of a publisher’s audience that comes to content from a digital platform has added complexity and additional data on consumption by individual customers as publishers seek to group customers with similar preferences and behaviors into segments. These trends are leading publishers to use big data for customer segmentation; the future of the publishing industry will be dependent on big data analytics.
Until recently, publishers have relied on surveys and third party data vendors to tell them about their audience. This helped them to sell an audience to advertisers, and it helped advertisers feel good about the ROI of the reach of their messages. However, with tracking tools measuring every click and scroll on a webpage, and with cheaper technology costs, advertisers are demanding much more detail about the audience they want to reach. Additionally, as publishers face pressure to retain and acquire more subscription revenue, targeted retention and acquisition campaigns are critical for efficiently managing subscriber yield. It is clear that a one-size-fits-all approach on subscription management does not work.
With big data tools and techniques to profile their readers, publishers can now segment their audiences based on “revealed preference”. This is distinguished from surveys and third party data (“stated preference”) by the fact that data on actual measured behavior is driving the customer profile. Metrics such as time on site, visit frequency, login frequency, content breadth, content preference, device preference, article page count, scroll depth and scroll velocity are measured by a number of clickstream trackers to determine online visitor engagement and what specific content engages that visitor. Additionally, if a subscriber logs in and consumes content online as a known user, all the “offline” data previously isolated to customer data warehouses can be added to the segmentation attributes of each user. Lastly, statistical and predictive modeling allows segmentation to use metrics like churn probability, acquisition probability, long-term customer lifetime value, product choice probability and price elasticity. This concept of combining detailed data for each person is commonly referred to as first party data. The ROI from this type of data for both advertisers and publishers is incredibly high when leveraged properly.
All of these metrics, probabilities, indexes and attributes may seem like a dense forest, but if they can be simplified and tailored for the specific market conditions, both advertisers and publishers can benefit. For example, publishers can combine all the engagement metrics into a single index (1-100) and then group them into logical thresholds to design easy-to-understand marketing campaigns.
The pie graph here shows how both subscribers and anonymous visitors can be grouped into six easy-to-understand segments. For example, for the “experimenting” group, a publisher might have a strategy of raising the engagement level through email marketing while the “fully engaged” segment might receive targeted offers and upgrade campaigns for premium products.
The chart here shows how overlaying churn probability and engagement can help build strategies to target specific subscribers with personalized messaging and campaigns.
Subscribers with low engagement and low churn need to be moved into higher engagement categories, while subscribers with high engagement and high churn should be prioritized for stop-saves and 1:1 communication. Also, this same data can be used to build products and prices that work best for managing long-term volume and yield.
These are just two examples of how publishers can segment and communicate with their current subscribers. The same type of data can be used to enrich the audience profile for advertisers to target ad campaigns. Using segmentation on key attributes that are valuable to advertisers - including content preference and behavior, as well as offline data such as paid subscription price, zip+4, demographics, email-opt-in and delivery days - allows precise targeting and much higher value for both the advertiser and publisher (and possibly even the user, if the ad was relevant to them).
The latest development in first party data is identifying users and audiences with “NAP” (Name, Address, Phone number). The actual personal information is not shared with the advertiser or any outside party, but if the publisher can confirm that a particular audience is not just an anonymous browser or cookie, advertisers are ready to pay top dollar to target their message to a “NAP” user. Hooking this audience segment into a data management platform and into programmatic ad exchanges also improves the yield for ads not sold directly to advertisers.
Audience segmentation, married to the publisher’s first party data, allows for more effective monetization of the website and audiences from advertising revenue with valuable first party (and NAP) data providing a much higher ROI to all parties. Once an audience is segmented publishers are able to much more effectively allocate their marketing resources, which enables upselling or cross-selling products to specific readers. This segmentation also provides the ability to send more personalized offers and provide for improved reader retention and loyalty. Readers are much more receptive to, and appreciative of, marketing messages specific to their interests and purchasing history.