Introducing CI Labs: a chat with our data scientists

CI Labs is a newly formed part of the Content Insights team. This week our roving reporter, Em Kuntze, sat down for a chat with Ognjen Zelenbabić, one of our data scientists, to find out what exactly the CI Labs squad do in their data bunker* and how the blended metrics that form the CPI tool are in a constant state of refinement.

So, Ognjen, you’re part of a three-strong team at Content Insights called the CI Lab. What exactly do you do there?

As this is a data-driven company, it was clear that we needed to create a department where the sole purpose is to analyse the data and try to find new insights and hidden patterns which will improve our calculations of the CPI model, as well as our clients’ understanding of what they’re seeing. We’re all data analysts and data scientists: guys who are experienced working with data and extracting information from it.

But everyone knows how to use analytics tools these days, don’t they? Why do we need a team of expert analysts looking over our shoulders? 

As we pointed out in last week’s blog, many people know how to open an analytics tool and fumble about in it, but far fewer know what to do with the data they uncover. In the analytics world it’s very difficult to actually draw conclusions from the metrics you might be using, especially if they aren’t very precise or concrete.

“Many people know how to open an analytics tool and fumble about in it, but far fewer know what to do with the data they uncover”

You need to be able to looks at trends over the longer term as well as being able to cross-reference different metrics to draw useful conclusions. All of that takes a great deal of time and understanding of data analysis, and it’s easy to get it wrong.

This is exactly what I’ve found from personal experience with a tool like Google Analytics. I have to admit, I don’t really know how to read the data…

And that’s not unusual for a journalist. You’re not alone!

Real-time analytics
Everyone’s chasing real-time analytics, but what are they learning, and what is it doing to journalism?

If you set up Google Analytics by yourself you will be served with a bunch of different statistics and information. You might understand the questions, but not necessarily what to do with the answers – the data that you are being served. 

The difference between CI and the rest of these analytics platforms is that we aim to provide insights, so the user doesn’t have to interpret the results themselves. You will get the answers rather than raw data.

In my experience, the same problem exists in all industries: when companies establish departments for business intelligence and analytics, the result is always the same: they stun managers with a bunch of statistics, data and reports and then those managers try to analyze the data and draw out some conclusions. Why would you expect people trained in business development, say, to know what to do with so-called ‘Big Data’. The whole thing is bizarre.  

When you put it like that, it seems a bit backwards…

It is! Data scientists, like the ones I work with at CI Labs, have made a profession out of analyzing data – we can interpret data and make recommendations. Content Insights, at its very core, is an improvement on the whole concept of data utilization: don’t waste time looking at numbers, just ask the questions you need answering.

And the CPI part of Content Insights gives you that information, distilled into something which is understandable and easy to monitor?

Absolutely. The intention of CPI is to look at the relationship between multiple metrics – precisely which metrics we look at differs from business to business – and essentially collate them into a single reading. So, if you like, it’s a single metric that takes into consideration many, many calculations and gives you a quick performance overview: whatever you did yesterday, did you do it well? Did you achieve what you wanted? And if you want to drill down into what went right or wrong, the rest of the tool allows you to do that. 

Presumably those overviews are going to look different depending on your industry aren’t they? The needs of a content marketing company are going to be different to that of a newsroom, right?

Exactly. CPI takes into account those different business models. For example, if your main goal is just exposure – where you want as many users as you can possibly reach – then the CPI would be calculated differently to a current affairs website. It absolutely has to be relevant to whatever kind of editorial you’re creating.

“We’ve planned a bunch of really exciting stuff that I can’t wait to get my hands on! Cross-relating the weather with the number of article reads…”

If your goal is reader loyalty, then monitoring what users are writing about you and how many of them are writing about you is obviously important to get a clear understanding of your success. Here, you’d be more interested in knowing if and how your readership reads and reacts to your content than how many clicked on a headline, something which sounds simple, but in fact is where some other analytics systems fall short.

Does this get refined the more you use the tool?

Insights are refined all the time. The more data you collect, the more precise your calculations will be. So the more historical data you have, the better.

So this is what the CI Labs is focused on? The refining of these questions?

Predominately my work is based on revising the calculation of CPI to recommend additional metrics to improve calculations to make them more reliable.

UX and analyticsTalking stats: can UX gain anything from analytics?

Is that something that you will always refine? It sounds like there’s never going to be a time when it’s ‘done’?

Exactly. It’s a continuous process. It doesn’t mean that the current calculation is wrong – it just means that we want to constantly improve. We’ve planned a bunch of really exciting stuff that I can’t wait to get my hands on! Cross-relating the weather with the number of article reads…

Hold up: cross-relating the weather and article reads?! Please explain a little more…

Well, you can imagine that most articles are read during the workday rather than at the weekend, so take that a step further and imagine how the weather might affect attention spans. If the sun is shining, do you think someone will stay at home and read something? Programming this stuff is almost infinite process: it’s a veritable lake of possibilities. You just need to be constantly thinking about other factors in reading behavior.

Wow. That’s incredibly specific. So here I am thinking that you are involved in the very technical side of how long somebody’s on a site and the click rate through and movements of mouse on page, and you’re talking about weather and external factors. That gets into the psychology of reading, doesn’t it?

Yeah, well, you can’t be purely technical. You need to open your mind and get into the philosophy and think about what people like and what people do and get the complete context, not simply the numbers. Once you put human behaviour into the mix, you start to get information and readings that really mean something – info that is really useful for writers and editors. It’s an exciting process to be a part of.

To find out more about Content Insights and what it could do for your content strategy, get in touch with us at www.contentinsights.com.

*We’ve told Em they work in a bunker. She believed us. It’s up to you readers to decide…

Schedule live demo now!
Content Insights: the content intelligence solution for the real world. Let us know when you’re free to take the demo.
request live demo
Close
Request Live Demo


Close
Sign Up For Free Trial

We would like to inform you that we track visits using Google Analytics, Facebook Pixels, Hotjar, LinkedIn and Mixpanel cookie trackers.

You can learn more about it in our Privacy Policy. Confirm you are comfortable with it.