The 4.5 Steps to Building A Better Data Product

How to avoid the dashboard trap by crafting your product around the patterns of data consumption

I did not know the key to delivering better data products when I was hired as a product manager at Teem. Rather, I learned this from years of research which included some significant trial and error.

This post isn’t my story per se, but rather an outline of how to craft a better data product, and some surrounding context so you understand where I’m coming from. It’s up to you to take this and apply it into your own microcosm.

tl;dr

Humans consume data in a specific sequence and you need to craft your product around that pattern. Analogous to patterns found in nature — think how oceanic waves all follow a pattern, yet are individually unique — humans consume data in the same sequence, yet each person maneuvers each sequential step differently.

Humans consume data in a specific sequence and you need to craft your product around that pattern

Failure led to a little slice of enlightenment

In 2015, I was hired to build my company’s data analytics platform and product offering. At Teem we make Workplace Experience software and our data was supposed to help customers gain insight into their real estate, workplace effectiveness, and wasted resources.

We began by researching with our customers and asking what data they wanted. “Everything!” they said. So we worked hard to create a product they could use. After months of work by our engineering team, we delivered dashboards of everything — visitor management, space utilization, recaptured time, peak meeting time — everything that we thought our users wanted. Customers upgraded their accounts and praised the product. We thought it was a great success; and near term metrics confirmed it was.

Typical metrics confirmed our product was a success; but our customers weren’t using the data.

There was only one problem: our customers weren’t using the data. They liked all the charts, but they didn’t know what to do with the information. Rather than improving on the dashboard with incremental changes I decided to start over. This time instead of asking what our customers wanted, we focused on the business decisions and the data gaps in their process. In this research, we discovered a pattern that people use every time they encounter and learn from new information. I call this pattern “The 4 Steps of Data Consumption.”

The 4 Steps of Data Consumption

The steps are intuitive and somewhat obvious in hindsight:

1 – Education

2 – Exploration

3 – Monitoring

4 – Action

I’ll dive into each, and use an analogy of the weather to help explain the concept: when getting ready for the day, we look at a weather forecast — but what we actually want to know is if we need to pack an umbrella. As users we have overlooked all of the knowledge we’ve acquired needed to interpret the forecast, and even then, a rainy forecast doesn’t keep us dry; the action we take on that knowledge does.

1 – EDUCATION

Our data journey begins when we’re children. We learn about clouds, we learn what temperature is, we learn about the water cycle and different types of precipitation.

In parallel, users begin by learning about the data — they need to comprehend what they’re looking at. They need to understand the underlying assumptions, calculations, and limitations. If your users are new to the data, the burden is on you to educate them. You cannot expect them to understand the data as well as you do and if you skip this step, they won’t trust your product and its value will be lost on them.

2 – EXPLORATION

The next step also begins in our childhood. We learn to apply our weather knowledge to our physical comfort. Remember your first bad sunburn, or getting soaked in the rain? Once this happens, we connect our learning to our experiences.

Once your user understands the data, they consummate that learning phase by exploring the data for themselves. This is ad-hoc analysis and data exploration. They learn how to apply the previous data lessons to their specific microcosm. There’s actually an entire industry of tools dedicated to helping your users do this; they are called business intelligence (BI) tools.

3 – MONITORING

This step is easy, our favorite weather app monitors the weather for us — all we need to do is read the forecast. In a weather forecast, we’re given a small dashboard of all we need: temperature, cloud cover, and chance of precipitation. A forecast is such a commonplace element in our lives that we forget the experiences we’ve had to help us interpret what it’s saying, know how it will affect us, and to trust it enough to act upon it. Even still, a change from fahrenheit to celsius can confuse the majority of American adults.

Analogously, once your user understands this specific data, and how it applies to their microcosm, then exploration is no longer needed. Rather than perform the same ad hoc analysis each morning, they evolve and automate it into charts to be monitored. Dashboards only work if the user fully understands what is being monitored.

Dashboards often fail because they don’t consider the entire pattern of consumption for users. These product owners argue that because the data user created the charts, they should understand the data, but that’s naive. Just like a forecast, dashboards are built by one with the intent of being consumed by many.

Your users are likely more familiar with rain than they are with your data.

4 – ACTION

This is the most valuable step. We need to know if we should bring an umbrella; that’s the entire purpose of the forecast! Our education, exploration, and monitoring are worthless if they don’t help us stay dry when we want to.

A rainy forecast doesn’t keep us dry; the action we take on that knowledge does.

Your users aren’t interested in looking at dashboards for entertainment — that’s what Netflix is for. Users look at dashboards for the sole purpose of knowing what action they can take to make an impact.

For example: is revenue dropping? OK, let’s do something about it. Let’s track down why…maybe it’s that qualified leads have dropped and revenue is a lagging indicator of our funnel. If that’s the case, now we know we need to strategize about how to increase the number of marketing qualified leads.

Taking action requires deep understanding of the data, and most users never get here. If you can craft a data product that helps users take specific actions that will result in an expected outcome, you have created immense value.

Step 4.5 – REPEAT

As your users successfully consume data and take action on that data, they receive a new perspective and the data takes on a new permutation. When the user takes action the data changes and users repeat their journey through the consumption pattern at an accelerated rate. This time education and exploration are much quicker, but they still happen.

Taking action requires deep understanding of the data, and most users never get here.

Conclusion

The product we’ve built has taken me two years of testing and iterating to understand what will and won’t work for my users. I then had to convince my executive team to take a leap of faith and let me craft an experience that is completely different than what any other company is doing. Over time, the feedback of our customers has been overwhelmingly in favor of the transition, and this time they have taken actions to prove it.

Your product will be different than what we’re building at Teem. But if you want your product to make a lasting impact on your users, craft the experience around the 4 steps of data consumption.

Special thanks to my favorite architect Morgan Williams for constantly helping me refine my thoughts and providing the illustrations — and to zach holmquist who I’m forever grateful to for recruiting me back in 2014 and letting me join Teem (EventBoard back then).

Similar Posts