At Content Insights, we are continuously working on perfecting the heart and soul of our content analytics solution, the algorithm called the Content Performance Indicator (CPI).
Like all great things, the development of the CPI has happened thanks to the collective efforts of many people – data engineers, data scientists, mathematicians, product developers, visionaries, and real editors, but also our awesome clients, who provide their invaluable feedback and help us adjust our product to the actual needs of publishers and content professionals.
The CPI algorithm recently reached its third, most precise version yet.
We also added a traffic perspective with unique segmentation enabled, new account & organization management, and more.
What is Content Performance Indicator (CPI)?
Our content intelligence tool (Content Insights) relies on the Content Performance Indicator (CPI), a complex algorithm that measures real human behavior, not just browser events. It is currently unrivaled on the content analytics market.
Our CPI takes into consideration dozens of different content performance metrics and examines their relations. It weighs them according to three behavior models:
Exposure CPI values volume. It expresses how good your articles are at attracting wider audiences, and which articles are more effective at bringing new audiences to your website. Exposure CPI is often favored by publishers who operate on an advertising business model.
Engagement CPI helps you to understand which content is connecting well with your audience, and which generates the most interest from your readers. It measures how efficient articles are at influencing readers to stay engaged with the content, both within the article, but also more broadly across the website. Engagement CPI is ideal for discovering native advertising opportunities and measuring your success in that department.
Loyalty CPI measures how good your content is at attracting highly-engaged, returning readers. This behavioral model looks at how articles are contributing to the overall loyalty of your reader base on the website.
You can configure your business model within the Content Insights app settings and increase/decrease the importance of each specific CPI behavior. This way, you’ll get access to the right kinds of insights which will support your growth.
How CPI is calculated
Besides its precision, the value of CPI lies in the fact it actually quantifies human behavior. You can observe CPI scores on article, author, topic, or section level. This score is always presented in a single number between 0 and 1000.
Interpreting CPI scores is pretty straightforward:
- CPI score of 500 = baseline
- CPI score < 500 = underperforming
- CPI score > 900 = exceptional performance
Why is this scoring system useful? Well, with just one glance, you can understand what’s going on with your content. You can, for instance, identify articles that need promotion because they are performing well but lack exposure, or pinpoint potential gaps and flaws in your current paywall strategy and see which articles could be turned into premium ones.
Content Insights collects all the necessary data about the traffic and the audience from the publishing platform (your domain) and behavior signals from social networks for each article. This data is then converted to raw metrics (e.g. Pageviews, Bounce, Visits, Visitors) and scored metrics (Read Depth, Page Depth, Attention Time, etc).
The result of normalization is always a number between 1-1000, achieved by distributing scores non-linearly around the median which is set to 500, to serve as a baseline for comparison. The median is specifically calculated for your website.
Introducing two CPI perspectives: article-oriented and traffic-oriented
Now, this was just a brief explanation of how CPI is generally calculated. But, let’s take a closer look on how the two calculations differ depending on the chosen perspective.
1. Article-oriented perspective
When it comes to the article-oriented perspective, CPI values for all dimensions were calculated as the average value of an article’s CPI for the observed dimensions in the selected time period. So, CPI is always calculated on the article level and its values are distributed across authors, sections, and topics by taking in mind your chosen time period.
This perspective allows you to analyze your website by putting your content in the focus.
In this perspective, we compare articles published in the last 30 days to pre-calculate the CPI. So, if you want to see how one author performed in a specific time period, the algorithm calculates the average of all the CPI values the articles by this author generated across the selected days.
2. Traffic-oriented perspective
Segmentation is a common feature in both content analytics and web analytics solutions. Segments let you isolate and analyze various subsets of data according to set criteria and filters. This way, you can make more sense of the data that’s available to you.
At Content Insights, we created unique segmentation that offers great value to publishers, and it is enabled within the traffic-oriented perspective where CPI is calculated ad hoc, calculated depending on the chosen dimension and set time period.
Here are some of the ways you can use this perspective.
a) Switch views and analyze data on a granular level
Thanks to segmentation in Content Insights, it’s possible to easily switch views and discover new, actionable insights which would otherwise stay hidden.
But, in contrast to other analytics solutions out there, we don’t just offer you filters so that you can dig through the data more easily. We offer real information that can possibly change the course of your strategy.
The reason why our segmentation capability is unique lies in the fact we recalculate the CPI values incredibly fast for each segment you create.
You can segment your traffic in the following ways:
- Reader Type: Subscribed / Anonymous / Registered
- Article Type: Free / Premium / Preview
- Channel: AMP / FIA / Native Mobile
- Referrer Type
- Device Type
An important thing to mention here regarding outliers: in order for CPI to be calculated, the observed dimension (author/article/topic/section) has to be identified as qualified. It can happen that your selected segment has articles which fall below the threshold, meaning you will not see them in your report or you won’t get access to insights you were searching for. However, this is rarely the case, especially with well-established websites.
b) Keep each member of your team data-informed
So, why else is segmentation in Content Insights a big thing?
Firstly – more so than ever before – it invites different team members of your media organization to explore insights in the app, and get a more realistic view of the results they managed to achieve. Segmentation extracts more value from the data, so there is something for everyone. Professionals responsible for different operations can enjoy reports that are highly relevant to them, and cross-team collaboration becomes better.
Your newsroom or organization will become democratized and each team member more data-informed. There will be less frustration with playing the guessing game with insufficient data, and more excitement with testing hypotheses that have a great chance of working out.
c) Understand the value of different channels and organize your budget the right way
All this is great in theory, but let’s see how segmentation can help with a specific example. At some point, you may want to evaluate the value of your Facebook traffic. Your editor might be skeptical about the importance of social media in helping the visibility of your publication, while your social media manager might want to know if their efforts are paying off.
In the Content Insights app, you can choose to segment Facebook traffic, and the CPI values for all the articles that match your criteria will be recalculated. You will get an incredibly precise and insightful report on the article, author, topic or section performance of that traffic segment.
This is invaluable for those in different roles within your media organization. You can fine-tune your strategy, analyze what type of content resonates with your audience and on which platforms, and allocate your content distribution budget for the biggest ROI.
Depending on your business model, you will put different things in focus and look predominantly at Engagement CPI, Loyalty CPI, or Exposure CPI. For instance, if you operate on subscriptions, you’re going to be most concerned about Engagement CPI and Loyalty CPI. If you see your Facebook traffic brings a lot of engaged readers to your publication, you should figure out how to make the most of your presence there and amplify it further.
So, with segmentation you will not just discover the amount of traffic coming from let’s say, Facebook, but the actual quality of that traffic in terms of engagement, exposure, and loyalty. And all that in just a few clicks.
d) Understand different reader types
In the Content Insights app, we recognize three different reader types:
Our data study has shown that subscribers are on average 34.5% more engaged compared to occasional readers, and this shouldn’t come as much of a surprise. Subscribers also spend more time reading stories and they read bigger chunks of content compared to people who are not subscribers. In addition, they are better promoters of stories as they enjoy sharing content on their social media profiles.
Knowing that different reader types interact and engage differently with your content can help you develop strategies to convert occasional readers into habitual ones, and then habitual ones into loyal readers.
Our data science research, which included the analysis of around 20 TB of content performance data, has proven that the loyal reader base (as defined by our Loyalty CPI algorithm) is the base of readers who have the highest probability of becoming subscribers.
By using the reader type segment, you can better understand who your readers are, and what type of content delights them and makes them stick around. A variety of behavioral metrics such as Article Reads, Attention Time, Read Depth, and Page Depth, will enable you to dig deeper and truly comprehend what people expect from you- and how to build a stronger relationship with them.
Of course, this is just the tip of the iceberg when it comes to the things you can do with our unique segmentation feature. If you want to explore all the other possibilities or find out how we improved the performance of the Content Insights platform, contact us at firstname.lastname@example.org, we are happy to talk to you and hear what your needs are in the domain of content intelligence.