The "Shape" of Data

One conversation I find myself having frequently lately is about the “shape of data.” As we show new companies what we’re doing with Variance and talk to them about their customer and prospect data, I often talk about my own experience getting my arms around the data our marketing and product generates. We were pumping all our product data through Segment and then using various tools to catch that data and send it to Slack and elsewhere in various forms. As I got my hands dirty wiring all this stuff together, I got a real feel for what was moving through these pipelines and how to harness them to help drive growth.

Seeing the data in its raw form helped give me a better sense of its “shape” than any chart or score could have. To that end, I think many companies are chasing down the wrong goals when they seek the perfect chart or score before they understand what’s living underneath. There are two lessons in this story that are worth exploring: the value of tinkering and finding the right level of abstraction.‍

The Value of Tinkering

I am a tinkerer and an entrepreneur, so I’m obviously deeply biased in this regard. But one of the lessons I’ve learned over and over in my life is that nothing beats getting your hands dirty. I can point to any number of occasions—including just this weekend—when I thought I had a good handle on something and then decided to check that understanding against the reality by playing with the thing and realized that it wasn’t quite as clear as I had initially believed. 

For me, this has happened frequently with technology. I remember spinning up my first virtual server and gaining a completely different understanding of how building websites and applications would be different moving forward. I then had a very similar experience the first time I played with serverless computing from AWS and Google Cloud. Instead of even having to spin up a server, I could just write some javascript that would only run when a specific URL was called. A few months ago, I got Redis, an in-memory datastore, running locally to get a better feel for how it works. Each time what came out of the tinkering was a realignment of my thinking and a recognition of new possibilities. I’m not an expert in any of these things, but by putting my hands on them, I was able to give myself a much more high-fidelity mental model than I had prior.

Working with marketing and product event data over the last few years has been another one of those eye-opening transitions. As I dove deep into how the data is generated and transmitted, I started to see all new possibilities for how the go-to-market could work. Instead of relying on waiting for a renewal or a customer complaint to look into product usage, we could get the signals in real-time. Since platforms like Segment do identity resolution, we could also carry these signals over to our marketing site and use them to understand where and how our prospects and customers were digging deeper or running into roadblocks. I got all this in theory before I submerged myself, but what I found was a large gap between the theory and the practice. Lately, I’ve taken to using the scuba emoji (🤿) to indicate immersion into a new world of data I hadn’t explored.

Finding the Right Level of Abstraction

One way to think about tinkering is as shifting your level of abstraction. When you’re talking about something you’ve never touched, you have a model in your brain for how it works that is naturally removed from reality. That’s fine with most things. I know, for instance, that I put gas in my car, which fuels the engine that then moves the wheels as I hit the gas pedal. If something goes wrong in that process, I also know that I can bring the car to someone whose understanding of what happens in between those steps is many levels deeper. When you tinker, however, you start to fill in the gaps yourself, and often that opens up new possibilities.

What does this have to do with customer data? A lot, I think. It’s my opinion, both from my own experience and many conversations over the last few months, that there are lots of folks who are living a little too removed from the reality of their data flowing underneath. The problem here is two-fold: first, they might be misinterpreting the meaning behind the charts, graphs, and scores they’re producing, and second, they haven’t opened up a whole new space of possibilities that would emerge from having their arms around the fantastic data they may already have.

To that end, I had a great conversation with an ex-sales person recently who is now heading up growth at a hot startup. He made the point that coming from sales, where everything is about understanding your prospect and customer to help you win the deal, he was amazed to see trial product data for the first time. Here was a treasure trove of actual information about how a prospect was using their product that he could use to help get a deal to the finish line. Instead of asking sixty discovery questions, he could dig into how the prospect was already using their product to help put together the most compelling pitch possible to win the larger business. Seeing and feeling the data sent his head spinning with possibilities. 

The Path Forward

You need to immerse yourself in the micro before you can even pretend to know what’s important about the macro.

- Josh Reich, The Micro to Macro Edition

Now, of course, not everyone wants or needs the raw feed, and as you understand things better, it’s critical to find shortcuts and abstractions that make it easier to use day-by-day and hour-by-hour. But the flip side is also true: often, there are many more people in an organization than you might expect that have a stomach to dig in and find ways to drive transformational growth. When given the opportunity, the folks on the ground who spend the most time speaking to prospects and customers often have a whole different view on what’s important than managers, who are many layers removed. So helping them feel the shape and recognize the trends is critical. It can also give them a whole new avenue to tinker and new ways to identify what works and what doesn’t for prospects and customers. One simple use case I’ve found helpful for Variance is as a way to test different customer communications before they become automations. I can see how different companies and teams are using the product, reach out with notes and questions, and then if it makes sense, turn the ones that get the best response into something more automated in the future.

In the end, what I’m most excited about with what we’re doing is giving everyone in the organization the opportunity to find new paths to growth using the goldmine of first-party data many companies are currently underutilizing. If we can help turn more members of the team into tinkerers we will have been successful.