We’re in a Three Bears situation with editorial analytics. First we had metrics which were too simple (page views, likes), then we were exposed to volumes of data too enormous for any regular person to process, let alone extract meaning from. Now we’re finding a balance between ease of use and depth of data and it is – as Goldilocks herself might have said – just right.
At Content Insights we’ve spent years developing the CPI tool to ensure that we are able to extract the most useful – and actionable – insights from the sea of data, and that that information is presented in way which is easy to understand and to incorporate into your workflows, whether you’re an editor, blogger or marketer.
What’s CPI, then?
Well, for starters we’re not talking the Consumer Price Index.
Oh, no, no, no.
CPI refers to Content Performance Indicator and it’s the engine room of the Content Insights tool.
To say that the algorithm deals with a lot of data is putting it mildly: in fact CI looks at all the data that is collected from the publishing platform and social media activity across the site in question. It then examines the relationships and ratios between certain indicators to produce both an article score (CPI), and other readings.
A Single Score
Once the data wizards inhabiting the CPI algorithm do their thing, you’re able to see a score. This will be between 0 and 1000 (500 being – perhaps unsurprisingly – average). It can relate to an author (as compared to other authors), articles (compared to other articles), topics (compared to – you’ve guessed it – other topics) and so on, and so on. It’s a great way to get an instant overview of how things are going.
So, how do we calculate this? (It’s a great question)
We examine over 100 metrics and look at the relationships between them in order to provide scores which are easy to comprehend, but based on sound data and data analysis. The key difference is CPI acts as the processor of all this data, so what you get are the actionable insights and the analysis, not reams and reams of data.
Why relationships and ratios? Simple.
Here’s a rudimentary example: say you’re comparing two pieces of content.
Article A has 10,000 page impressions. Article B has 100.
Article A has an average read time of 15 seconds. Article B has one of 2.5 minutes.
Judged on the first metrics, Article A is the more successful of the two. Judged on attention time alone, Article B is the clear winner, right? Well, not really….
Article A is a 200 word article about England’s recent World Cup penalty shoot out.
Article B is a longform piece about differing managerial styles of England and Colombia in the same match.
Even looking at the ratios between page impressions and average read time, we can start to get a more detailed understanding about how the content in functioning. Article A gains only minimal engagement, but it is a very short article and it does very well on exposure, as it’s designed to do. Article B has far fewer page impressions, but boy do those readers stick with the content. Both have their place, and both are effective in their own areas, but not when compared against each other using only simple metrics. In CPI terms, they might both score highly.
See what we mean? Now, imagine what looking at over 100 different metrics, and the relationships between those metrics might reveal. Exciting, isn’t it?
So what can CPI do for you?
Content serves different purposes, which is why a singular approach to analytics is often problematic: articles seeking to drive subscriptions will report differently to those hoping to drive successful native advertising campaigns.
CPI is organised around three main objectives: Engagement, Exposure and Loyalty.
You might be interested in Engagement CPI if you want to find out which authors are a good match for your audience, which articles aren’t getting the kind of attention you’d expect, and can help ascertain which parts of the website are a good fit for native advertising.
Exposure CPI can determine the effectiveness of social media efforts and also identify which content might be suitable for premium display ads. Exposure CPI in correlation with Engagement CPI help to identify articles which have clickbait headlines.
Loyalty CPI is calculated slightly differently because it studies reader behaviour, not content performance. The data labs team have defined loyalty as the ‘most sequentially highly engaged readers’ and if you’re looking at what these super fans are doing, you’ll get a good idea of how to attract more readers of this type into the fold – vital if you’re pursuing a pivot-to-paid approach.