Our CMO Cassidy Shield and I recently got a chance to sit down with Kate Strachnyi who runs DATAcated to discuss all things related to data storytelling. It was a great conversation with a lot of audience participation, but it also ran for an hour. Below you can check out the TL;DR, and/or watch the entire video if you want more!
What really is data storytelling?
Informally, it’s what someone comes back with when you say, “What am I looking at?” or “Can you talk me through the numbers?” A little more formally, a data story is a combination of visualization, narrative, and context that helps people understand and act on what’s happening in the data. It can happen externally and can be used to communicate, but people also do it internally all of the time. We, humans, understand via stories.
How can data storytelling help bridge the gap from all the data we’re collecting to actually communicating the key insights to those that need to hear them in a timely fashion?
The only way people understand anything is through stories, that’s just how we’re wired. So if data is going to influence us and our decision-making, it has to be in the form of a data story. There’s two ways organizations are creating data stories today: 1) Try and train people up and improve their “data literacy” so they can create their own stories or 2) Hire a bunch of analysts to write all of the stories. Unfortunately, no one is having much success with either of these approaches. The company was formed to explore a third way: scaling and augmenting the analyst function with technology, specifically automated data storytelling.
Let’s talk about Lexio, how does it help with organizations get more value from their data?
Roughly 30% of most organizations today engage with BI. Lexio is designed to be a great experience for the other 70%. To do that, we knew it had to be Simple, Personal, and Proactive.
Why is the current adoption low, and how does Lexio help?
Business users often don’t understand dashboards, or they often come away with a shallow or wrong interpretation of the data. Neither result is satisfying or useful for business users, so adoption is low. Another challenge is culture. If you want to drive data adoption in your org, you need the culture, top-down. Too many people see top-down behavior and think, “If the folks above me in the org chart aren’t making data-informed decisions, why would I?”
How do we actually move towards a stronger, more data-driven culture?
It has to start from the top. Executive leaders who want a data-driven culture but don’t practice what they preach will continue to find themselves disappointed. Just like you can’t start with a dishonest CEO and build out a culture of honesty, you can’t start with “from the gut” CEO and expect to build a data-driven culture. Regularity is also very important. As Ben Horowitz says, “What you do is who you are.” If you only “do” data once a month or quarter, you’re never going to be a data-driven org. Everyone in your org needs to be aware of the stories in the data on a daily or weekly basis. This is why it’s so important to get the stories out of the data in a way that people can quickly consume!
How do you decide the message to communicate and know what to leave out?
Having a message at all is critical. Too often, analysts are afraid to include their own recommendations or insights. Beyond that, audiences are always interested in knowing what drove a particular change, and almost always uninterested in learning the “story of the data” (“Well, I had to join these two tables…”) The better you understand the audience and their priorities, the better you’ll be able to decide what to include and leave out.
Are Lexio’s stories created by AI?
Yes! Every story is written automatically by Lexio’s AI. It decides which information to include, how to structure the story, and how to express it all in natural English. Lexio’s AI works in two “modes”. The first is when Lexio is looking for interesting things happening in the data, which it publishes to the newsfeed as “Notable Insights”; this is super helpful because these stories can be pushed directly to users without them needing to sign in at all. The second mode is when a user wants to explore or answer a specific question, and Lexio is generating stories for them on demand.
What brings prospects to Lexio?
We hear two themes over and over again:
- Analysts contacting us to say, “I need to replicate myself”. Analysts know how critical they are as the data “middle man” for their org, but the org’s need outweighs their ability to provide.
- Executives contacting us to say that their approach to BI isn’t working. This gives you a sense of the scale and immediacy of the problem: not a lot of BI tools being bought by C-suite.
Is Lexio in English for now?
Yes, but it will eventually be multilingual. We have internal Hackathon projects that generate stories in Mandarin, but we haven’t made the commitment yet to productionizing an approach. Broadly speaking, there’s two paths we could take: generate natively in other languages, or continue to generate in English and then use cloud translation services to provide stories in other languages.
How do I go from having data to actually seeing it in Lexio?
When it comes to getting data into the system, we follow the standard BI paradigms, and we’re always working to reduce the friction it takes to get data into the tool. One big difference between Lexio and traditional BI is that Lexio doesn’t use any effort to build dashboards. Once an analyst sets up the initial KPIs and Topics that matter most to people, Lexio is able to write about those in any way a business user would like.
What kinds of stories can it tell? Predictions and correlations?
Lexio focuses on operational metrics: KPIs that regularly change over time. For those metrics, Lexio can track them (discuss their recent performance), break them down by dimensions, or compare them across entities (e.g. East Coast v. West Coast.) Lexio does predictions and projections now, and it will do correlations soon. Lexio will never optimize for pure analytical power; we always look at more advanced statistics / analytics through a lens of “How can this improve stories for a busy user whose day job is not data?”
If you’re interested in hearing more (or seeing my son who fell offscreen and needed comforting!) you can check out the whole conversation below. Cassidy and I are always happy to talk more as well, connect with us through LinkedIn or email directly (Nate Nichols or Cassidy Shield)
Thanks to the audience for attending and asking great questions!