It seems like everything has an analytics dashboard nowadays. At both Lexagle (a contract management software company) and NUS Developer Groups, analytics on how our platforms were being used was a major request from clients and my product teams had devoted significant amounts of time to it. Once, out of sheer boredom, I poked around Slack for a while and was surprised to discover that even they had analytics too!

Why do we even need analytics?

For a while, I was very skeptical of the utility of analytics dashboards. At some point, everywhere I turned my head, numbers were being shoved down my throat. As analytics features became more and more ubiquitous in the products I saw, I couldn’t help but feel that companies were just jumping on the bandwagon, implementing analytics only because it looked cool and shiny.

The turning point came for me when I started work on a project in NUS Developer Groups and learnt that the organisation had specifically requested an analytics feature. I realised that product managers weren’t just working on analytics dashboards willy-nilly; they were doing it because clients had asked for it. As someone who favours a user-centric approach to product development and design, it suddenly all made sense. Users wanted to dive deeper into how their work processes were being carried out, and they wanted concrete evidence to back it up. Product managers had to step up and address users' needs.

For business-to-consumer products, analytics might just be a fun way to learn more about your own usage and face the reality of just how enthusiastic you are about the products you enjoy (think Spotify Wrapped). But it doesn’t stop there: for many business-to-business products, analytics makes even more sense. For workflow and operations-focused software like Lexagle, data on the usage of products that have been integrated into firms’ daily routines gives a lot of insight into the company’s efficiency. It can help management understand where bottlenecks occur, whether work is being distributed evenly among staff, or even whether the company is utilising its subscriptions fully. (Slack Analytics, I’m looking at you!)

Why do Product Managers need to get involved? Isn’t analytics just numbers?

Well, yes. Most of the grunt work seems to be coming from maintaining and pulling the data itself. But as I found out when I got my hands dirty in some product work for analytics features, there is much more to building a native analytics feature beyond the SQL magic that developers are running behind the scenes.

I’m sure that everyone has encountered at least one convoluted graph in their life that they can’t make head or tail of. Therein lies the power of product managers: to make numbers make sense for everyone.

What an analytics dashboard should look like actually has several details that a product manager needs to decide on. For starters, the overall layout of the dashboard is up for debate: the “widget” style is popular, displaying each chart in its own little container, but certain situations might warrant different arrangements. We then delve a little deeper into what each chart should look like, such as selecting an appropriate chart type to represent each metric, and what should go on the axes of each chart. Sometimes, users are very particular on what they want to know, but when they aren’t, product managers might even have to choose what metrics to include in the dashboard.

Also, not all data is perfect in reality. Accounting for real-world quirks gave rise to several tough questions during my own journey in managing analytics features. How do you present a readable, accurate chart for data that is recorded in different units (such as purchase value, which can be recorded in different currencies depending on where the supplier is from)? What happens if no data is recorded for a particular metric, or recording stops halfway? Is the ideal time filter for one metric also the best for another, or should we apply the same time filter across all metrics in the dashboard for uniformity? Because every product and client is unique, analytics feature work is never boring—there will always be new and interesting problems to solve!

At each juncture, the key question that I ask myself as an analytics feature owner is, “Would this be intuitive for users?” This doesn’t just apply to user flows and actions, but also considers how users might interpret the information on the charts they view. I want to make sure that users can get the information they need from one glance at a chart, without needing to perform mental gymnastics. One rule of thumb that stands out to me from my days at Lexagle is to aim for "an analytics feature that even your grandparents can use”.

This sounds fun! How can I learn more?

One thing that was very helpful for me in thinking about how to build an analytics feature was a strong foundation in data. Fundamental SQL knowledge allowed me to have a slightly more realistic picture of how developers might go about making my ideas come to life. More importantly, although I was never the one actually writing and executing the commands, knowing SQL structured my thinking on how data should be aggregated and represented visually, to achieve my ultimate goal of drawing the link from numbers to real-world insights. For product managers with non-technical backgrounds, learning SQL on the job is not unheard of. But if you would like to get a head start, SQL Murder Mystery is a fun and digestible way to pick up the basics.

Another way to sharpen your analytics skills, no coding required, is to simply get out there and observe! Every chart I see, even my Goodreads reading stats, is an opportunity to ask myself, “does knowing this data add value?” I’m also inclined to question several other things, such as the accuracy and readability of the chart design, or if filtering and sorting is intuitive. As I look at different approaches and experience their pain points for myself, I also learn a little more about what makes for an impactful, usable analytics feature.

User-facing analytics seems like it’s here to stay, and product managers have the exciting task of defining what that looks and feels like. At the end of the day, a good analytics dashboard should be able to transform all that raw data in your systems into clear, actionable insights for users, even the most data-illiterate ones.

About the Author