[Editor’s note: This interview has been edited for clarity.]

Most companies (probably all, if we’re being honest) struggle to make sense of their metrics.

The problem isn’t that we don’t have enough data—or even the right data—to make informed decisions. In fact, it’s the very opposite: It’s 2018 and we’re all drowning in our own data.

We sat down with Jason Amunwa—SaaS growth consultant, product manager, and data design vigilante—to talk about metrics, data design, and the very human anxieties behind our need for numbers.

Hi Jason, thanks for taking the time. Before we dive into talking about data, can you introduce yourself and tell us a bit about what you’re working on now?

Hi everyone! My name is Jason Amunwa.

My background is in marketing and product management. I was in the advertising game for quite a while and then I made the jump to product management about seven years ago. Since then I've worked on a lot of startups and a lot of products, from their beginning all the way through to them being acquired.

Now I'm a product manager at Voalte, which is a secure communication solution provider for healthcare organizations that makes it easier for doctors and nurses to collaborate.

What is it about this work that piques your interest?

I find that my I tend to come alive in places where data meets design. Recently, that interest has been channeled into questions about how analytics and metrics are presented to people.

I find that a really interesting area, because we're living in this time where we just have so much data available to us right now. We're struggling for meaning. It's a “water, water everywhere, but not a drop to drink” situation.

The more I talk to the founders of the startups or the leaders of companies, the more I find that everybody is looking for meaning—for that golden insight that will inform them and let them say "hey, I'm making the right decision!"

So they'll start with a super simple question like “how’s the website doing?” and then ask for reports. And they get back these CSV files, or pipe delimited files, or log data—all of which adds up to a big kind of shrug.          

Do you feel like this problem is pretty universal?

Absolutely. Any time analytics comes up, the eye roll always happens and there's that sound of like: "Ugh. Analytics."

And I really sympathize with that feeling. When the boss walks in the door and asks how the company is doing—well, you’ve got to answer that question. But with the tools that we use today, it can take hours to answer that question effectively. And there is a lot of human effort required to shape and design the data into a presentable form.

That always makes me think: How awesome would the world be if we could interpret our data easily?

It's like having a dirty telescope: We've got the wonder of the heavens out in front of us and we can point our telescope at it all, but if you've got a dirty lens then you're ultimately distorting what you see or missing things entirely. You're bound to make inaccurate conclusions unless you polish up that lens.

How do you do that? How do you polish up that lens and look at data more accurately?

With words. Not numbers.

Part of what makes analytics such an exciting subject right now is that we're finally at the point where the technology is starting to be able to synthesize data—run it through a computer and have the computer look at the data and then tell you things about that set of numbers in plain English.

Logs files, big tables of numbers, pie charts, graphs—all these things became popular because that's what the technology was able to produce.

But when most people look at analytics tools, you can literally watch their eyes glaze over. The numbers don’t really have any meaning to them.

Ultimately, I would love analytics to build off the proliferation of text synthesis technology that’s happening. I want to get to the point where people aren’t logging into a tool or looking at a dashboard every single day. I want to be able to get text updates saying "hey, you got a spike of users from that ads campaign. You should try rerunning it with these keywords or try spending more money on this demographic."

That would be amazing.

Yeah. And you really don't even need to know the numbers. If you know that today your website is doing the best it's done in the last three months, and it's doing well because of what you did on Twitter yesterday and you got retweeted by Neil Patel, etc.—do you really need to know how many tweets or page views you got? No. You have all the information you need to take action right there in words.

Can you explain the difference between data-driven and data-informed decision making?

Basically, data is just the raw material out of which we manufacture meaning.

A lot of people point to data and say "it's data, so therefore it is true." And that's not necessarily the case. It’s like this one episode of The Office, where Michael Scott ends up driving a car into a lake because the GPS wasn’t up to date. And his defense was “the GPS said to go straight.”

Being data-driven involves looking at a number and carrying out action based on the number alone. Being data-informed means factoring numbers into your decisions, but also taking into account things like timing or resource availability or anything else that is not captured in the raw data. It involves adding a layer of interpretation onto the numbers.

Because nobody has a perfect, all-encompassing set of data. Being data-informed involves recognizing that reality, and then adjusting your decisions to take that into account. Trust but verify. Don’t drive your car into the lake just because the GPS is saying so. Dig deeper and find out why it's telling you to go that direction.

How does design factor into data-informed decision making?

So, I actually really love Google Analytics. It's super robust. And it's probably the most popular analytics tool on the planet. It's installed on so many websites and it's kind of seen as the gold standard in measuring web activity. I love it.

But, I also hate it. I hate using it. It takes so much effort and so many hours to surface up anything meaningful.

This is where I think design can be improved to do a much better job of presenting data. A lot of people tend to think of design as just the pretty wrapping around the engineering. But when you have smartly designed presentation, it changes the way people use a tool and interpret the data.

It's sort of like the difference between reading what someone said and hearing them say it in their own voice. When you hear somebody speaking in their own voice, it conveys so much more information than just the words that they're saying: They might be tired, they might be excited, they may be fairly mad about the words that they're saying, their voice may rise and fall. All of that is additional information that conveys qualitative meaning.

Design makes the story that you’re telling about the data much more compelling, and helps people pull meaning out of that data more easily.       

What sort of design elements are particularly useful for data storytelling ?

Anything that gives context. Things like the use of color (green and red in particular), the hierarchy of the layout, directional indicators—all of these things help people look at the data visualization and immediately understand "oh, man, things are going really bad right now—we need to take action" or "things are all good—we don't need to take any action right now. Let's move on to other things."

I see such a need for this in the healthcare industry. People are stressed, they're processing a lot of information at any given time, and they're being called upon to make very, very quick decisions with serious consequences for other people's health. And the whole time, they’re working off these really dense charts or Excel spreadsheets.

Again, design isn’t just pretty wrapping. Better analytics design really matters, especially in high-stress, high-impact environments like a hospital or an emergency room.

Can you think of any tools with good analytics design that tell the story well?

I can’t think of anything that doesn’t require a certain amount of human input. There's no tool that I know of where you just pour your data in and it just tells you excellent stories.

I have seen some tools that are moving in the right direction. For instance, Tableau lets you visualize your data any way you want, which certainly facilitates better data design—but it’s not automated. Nothing is quite there yet.

The funny thing is, data analysis sounds at first like a really cold, technical profession. But in actuality, analyzing data still a highly human-dependent endeavor. You still need a person to look at the data, analyze it, and then tell the story to people so that they understand what the is data implying.

There are some blogs that do data storytelling well. People laugh at me for saying this, but I think FiveThirtyEight does a really good job of data storytelling. Yes, they missed the mark with the 2016 election, but they were telling a story based on the data that they were collecting. In that case, I just don't think they had all of the data.

Let’s backtrack a bit… How do you choose which metrics to look at in the first place?

There's no one-size-fits-all answer. I know that's an easy expression to toss out, but it’s true. “Businesses” aren't just an amorphous, homogenous blob. Businesses are all different and they're also all at different stages of their lifecycle or growth stage.

Start ups, for example, really need validation that their product is meeting a need, that there is demand for their product, and that they are acquiring the fuel to get the engine running. Since they need that affirmation, metrics that measure user growth are typically good indicators of success for a company at the startup stage.

On the other end of the spectrum, established companies like Fortune 500s need to retain and upsell, so the metrics they look at typically focus on that. Ultimately, it really comes down to what your business strategy looks like.

And the “right” metrics change over time. The metrics that correlate with revenue when you're a tiny startup most likely will not be the same metrics that correlate with revenue when you've grown to 50 employees or 500 employees.

It's important to regularly reevaluate the metrics you’re using to measure success. I find that many businesses figure out the metrics that they want to pay attention to, and then they pay attention to them for too long. So they get to a point where the business has actually progressed to a different phase, and they’re focusing on the wrong metrics even though they used to be the right metrics.

Is it possible to focus on the right metrics for the wrong reasons?

Yes, definitely.

For example,  I worked on an analytics product called Filament, which was designed to help content marketers better understand how their content was performing. Essentially, we were pulling data from Google Analytics and a few other sources, and then marrying that together to present content marketers with an “engagement score.”

We ran into trouble when Twitter decided to alter their API and make it so people couldn’t access their tweet counts for individual pages without shelling out hundreds of dollars a month.

I’m sure that was a great business move for Twitter, but it kind of screwed everybody that was measuring tweet counts as a signal of engagement. Anyway, we (Filament) sent out an announcement explaining that we were pulling tweet counts out of our scoring algorithm because of X,Y, Z reasons.

People were really mad at us. They said things like “well, this thing’s useless now” and “how am I supposed to know how well I’m doing if I don’t know how many tweets my article got?”

Saying that an algorithm is useless without tweet counts is kind of like saying “we didn't register that you breathed in the last minute, so you must be dead.” Tweets are just one signal in a vast sea of signals for this amorphous concept that we call user engagement.

But people really got fixated on the numbers. They liked going to their website and seeing that those tweet numbers were going up and up and up. And I think that the fact that people get so fixated on the metrics is an indicator of how hungry we are for meaning.

That’s a very human anxiety, isn’t it? Is that part of what fuels your passion for this subject?

Yeah, that’s one of the reasons why I'm so interested in analytics and data design. At the end of the day, it’s about allowing people to find meaning and find truth in what they're doing. “Am I doing better or worse than yesterday?”

I find that’s a pretty common desire. People always want to know if they're doing better or worse than the day before.

Big picture, I think about a world in which anybody can make sense of large sets of data quickly, without the need for a statistics degree or a lot of experience in analyzing data.

Making analytics more accessible would make everyone more empowered to make more informed decisions. And that seems like a pretty good goal to work towards.

So we have to keep polishing everybody's telescope lenses so that they can see clearly. And I think the way that we do that is by finding the most intuitive ways to present data to people so that they can make sense of it.

That means no pie charts, no graphs, no big tables of numbers. It should come down that one line that tells you "hey, your website is doing awesome today. Whatever you did yesterday, you should keep doing that—because that is the path to growth."

There's kind of a philosophical twinge to that, which makes me feel good.

...
In his own words: Jason is a little robot that converts tacos into somewhat useful insights about analytics, design, user experience, and the business of building great software. Part product manager, part SaaS growth consultant, he will visibly perk up whenever you talk about space, AI, technology trends, and tacos. You can find him at www.growthlook.com.