Season 2, Episode 10
Data storytelling is evolving just as quickly as the business problems that it’s trying to solve. One of the newest forms of data storytelling is Natural Language Generation (NLG), which transforms data into plain English.
Join Lyndsee Manna, EVP of Global Business Development & Strategic Partnerships at Arria NLG, and host Chris Hemphill as they discuss the possibilities that come from using this new tool to improve everything from investments to marketing engagement.
1:39 Lyndsee Manna’s Data Literacy Mission
5:23 How to Become Data Driven
7:35 What’s Holding Teams Back from Data Literacy?
12:19 How Technology Restricts Data Literacy
14:50 How Natural Language Generation (NLG) Tells Data Stories
19:34 Who is Doing Data Storytelling Well?
22:29 What NLG Insights Actually Look Like
30:37 Spend Less Time Combing Through Data
Subscribe to receive emails when new episodes are released.
Thank you for subscribing!
Please check your inbox to confirm.
VP, Applied AI & Growth
EVP, Global Business Development & Strategic Partnerships
Chris Hemphill (00:00):
Hello, healthcare. The worst thing that can happen in healthcare marketing. Your programs are resonating. You’ve activated your patient base. You’ve even driven volume to service lines. Yet, no one believes you.
Chris Hemphill (00:14):
Data storytelling is the difference between falling flat or securing the budget and trust that’s needed to grow your strategy and team. You can learn about data storytelling with books like Nancy Duarte’s, Data Story, or we’ve even made a video on the subject, if you click above. However, the future of data storytelling is evolving in ways that I didn’t even expect. Leaders are becoming more sophisticated and there are even newer technologies like natural language generation that tell data stories for us. Whoa.
Chris Hemphill (00:52):
So, does that mean our spreadsheets are going to start talking to us? Maybe. We spoke with Lyndsee Manna, the EVP of business development at Arria in order to find out. Lyndsee works on the leading edge of natural language generation. But she also has a quest to help everyone become more data driven, more data literate. We talked with her about what’s holding people back, the future of tech and what metrics to discuss and to avoid when speaking with healthcare leaders.
Chris Hemphill (01:21):
What we’re going to be talking about are some foundational steps and some newer technologies that are going to make telling those stories a lot more easy and effective. But before we get all the way into that, Lyndsee, I thought it would be good just to learn a little bit more about your background. Could you talk about what’s fueled this path and why you’re now interested in data storytelling in general?
Lyndsee Manna (01:45):
Well, I think ultimately, I’ve always been very passionate about helping other people, that’s the constant. How can I make a difference? What is my purpose today on this planet? So how can I help others? So, yes, as a musician, I started going to university and majored in music. So as a musician, we’re sharing our gift with others, the gift of music, and ultimately that’s to make other people feel the joy that we feel, and to share that gift and to give a little piece of yourself away every time, when you’re a performer or a musician.
Lyndsee Manna (02:21):
So certainly there’s a constant, but I quickly, in my first year of university, fell in love with economics and analytics and math. And so I’ve always been a lover of math. And we were talking before this session, a lot of mathematicians are musicians. There is a direct relationship between math and music. I think about eight out of 10 people that I ask here at Arria, and I ask out in the field that our analysts either have a deep love for music or are musicians themselves. So that’s an interesting connection there.
Lyndsee Manna (02:52):
But I’d say my path really was very much catapulted by where I came from. My parents are serial entrepreneurs. If that is a true label and definition, that is who they are. And so that’s been around me my whole life. So although I had a passion for business and analytics and economics, I used everything that I learned there and applied in the business space, helping to bring companies from startup to public many times and helping build businesses and fulfilling lots of different roles throughout the way. So my journey has been unique, but I am very fortunate and blessed to be surrounded by the best of the best in terms of building businesses and making them successful.
Lyndsee Manna (03:47):
Certainly a unique story, but always focused in technology and in software that really can be a change agent for paving a new path and a new way to do things. So prior to Arria, I actually built a long-term care pharmacy. And then before that I was part of a company called Diligent or Diligent Board Books, Board Books, as it may have been known. And that was the first of its kind digital board, board paper. And that was in the Enron times and Sarbanes-Oxley. So, always new, innovative tech, changing the way people do things, empowering people to have more time and make a difference. And that’s really what data storytelling empowers. And we’ll be able to talk more about that today.
Chris Hemphill (04:35):
Cool. I just want to first pause and say, if you are a musician, I want to hear about it. If you’re watching and you’re a musician, let us know what instrument you play. I’m just curious.
Chris Hemphill (04:47):
But second, one thing I wanted to emphasize is, we’re talking about a transition from art into math and data. And I think that a lot of people might look at what it takes to become data driven or they might look at that as a halo, but maybe if somebody doesn’t have a math background or a science or technology background, they might be intimidated by that. And I think it’s just really important for folks to hear from someone who has gone from the art side to the data side, and as people strive to be data driven, I just think it’s important to get those kinds of stories out and appreciate that.
Lyndsee Manna (05:23):
Yeah. Well, real quick, let’s talk about that connection. Because what makes humans human is our ability to be creative, is our ability to be artistic, is our ability to communicate through language to one another. And so what’s fascinating to me now about where technology is and specifically artificial intelligence, is that yes, data, math, analytics, science, the ability to code, those are all important skills, but we also need writers. People that know how to communicate. People that can recreate a story, a story. And that is a very creative skill.
Lyndsee Manna (06:11):
So if you are going to make a humanlike technology, something that is not so artificially human, but very human, and as human as it can get, you need people that are artistic to contribute to how that virtual digital human is going to behave. So at Arria, we really lean on creative writing and creative writers, which is a skill of the arts to recreate, to recreate how a human would communicate to another human. So there is a place for everyone. And certainly as we get into artificial intelligence where you’re trying to create humanlike experiences, you need those creative skills.
Chris Hemphill (06:54):
Great. And I like the emphasis on the fact that there is a place for everyone. When we look, when we take a step back and ask ourselves about what’s holding us back from being data driven, or what’s holding us back from being able to tell effective data stories, there’s kind of two, two overarching parts I thought it would be interesting to get into one.
Chris Hemphill (07:16):
One part is the fact that your profile on LinkedIn, it focuses on the passion about data literacy. So I’m curious, what’s holding organizations back. And as you know, we have a heavily in the marketing audience, but what’s holding people or organizations back from a data literacy perspective and being able to tell effective data store and convince or communicate with leadership about data stories?
Lyndsee Manna (07:44):
I think it’s a, it’s a bridge of accessibility, is really what’s holding people back. The data is there. It’s there. Now, I can access that data, but not everybody can have a level playing field when it comes to understanding or comprehension, the ability to analyze that data, the ability to understand in a way that is in line with my level of understanding. Different constituents have different kind of baselines and skill levels.
Lyndsee Manna (08:19):
So I think it’s that bridge to facilitate understanding and comprehension, and democratizing that ability, democratizing accessibility to that bridge. That’s the key. So can your data tell an effective story? Can your data communicate to everyone within the organization, regardless of skill level, regardless of knowledge level, in a way that they can understand and then do something with? So understanding, comprehension, now I’m knowledgeable, and I can take that data because I understand it now. I can take that data and do something with it. It’s actionable now. It’s intelligence that now I can change the path of my department or change a marketing initiative, that can help us achieve our goals.
Lyndsee Manna (09:16):
So I think that the stopgap right now for many companies is, do I have a way? Do I have a way to give everybody in my organization access to that? It’s that bridge. It’s accessibility to that information in a way that they can consume it, in a way that everyone can consume it. And not everybody can consume hundreds of thousands of rows and millions of… Or hundreds of thousands of columns and millions of rows. That’s a lot, that’s really overwhelming. Not everybody’s a data analyst. So how do we make that accessible to everybody? That’s the trick.
Chris Hemphill (09:52):
And I know that we’re probably planning on getting deeper on this in a minute, but just curious about any examples that you’ve seen that demonstrate data literacy challenges.
Lyndsee Manna (10:04):
We talk about internal applications and today, of course, my expertise is data to language, which is known as NLG, so data to language. Internal use cases within an organization is going to help us understand all different aspects of the organization, financial. Certainly we’ve got a lot of marketers on the line here today. So what’s driving my campaign. Where are the leads coming from? What can I do more of? How can I impact my marketing initiatives? How do I actually do that? There’s a lot of data that’s going to tell us where to go and what to do. And so internally, I’m going to use that as a powerful tool to guide my next, most intelligent business decision, or decision as a manager or leader, and then externally, externally facing, it’s also true one to one marketing.
Lyndsee Manna (10:59):
So if we kind of stick with the marketing thread because that’s a lot of our audience today, I want to give a very specific message to this person. I know a lot about that person. How do I constantly change that message with the data that I have, so it feels like it’s a message just for me? I’m a consumer. I want to receive something that’s written just for Lyndsee Manna, not a marketing message that’s written for a bucket of people or a large population.
Lyndsee Manna (11:26):
So either way, if you think about it, whether I’m an internal inside the organization consumer of information, in words that I can understand, or I’m outside of the organization, it’s meeting people where they are. So meeting people that I can identify with myself, if you’re giving me marketing content and I’m outside, or if I’m internal to the organization, meeting me with information that’s relevant to me that’s useful to me and that I can understand.
Chris Hemphill (11:55):
Now, we get into a new territory when we think about like how we individualize our presentations and communications, data stories to meet… let’s say that we want to give a presentation to our chief financial officer or some somebody who’s going to have a different perspective and different metrics they’re focused on than us. We were talking about the data literacy limitations, but what about any kind of technology limitations in being able to tell those stories? Well, what are some common limitations that you see in terms of, well, hey, I have an understanding on how to tell these stories, but it just doesn’t work in my organization. What are some of the tech barriers that you’ve seen?
Lyndsee Manna (12:36):
Wow. Well, I’d say historically it was organized data. I was talking about data accessibility and that bridge to it. If it’s not there to begin with, we have a problem. I think more and more companies are data focused and focused on collecting data and organizing that data. Historically, we had a lot of companies that weren’t collecting the data, weren’t organizing that data. So it starts with data accessibility, and I’m not talking accessibility like comprehension. I’m talking about having it in the first place.
Lyndsee Manna (13:12):
So making sure you’re collecting the data and that it’s organized, is going to really turbocharge your ability to use more powerful technologies, like natural language generation, to communicate that to internal and external constituents, and then let it work for you. Get the data really working for you. So having it and having it organized and clean, should be a best practice for any company that’s collecting data. Collect a lot of it, as much as you’d like, but make sure that it’s organized and structured so that you can use it.
Chris Hemphill (13:44):
Perfect. And I love that example because in healthcare, what you’re hitting on, even if a health system, let’s say they have two instances of the same electronic medical record, even those might not be able to talk to each other. So the types of insights that would be valuable across the enterprise, aren’t available until there’s a whole lot of connection that happens. But at very start, getting all the data together, I mean, that’s what we need to do to be able to drive insights and predictions and things like that, so that touches on one of the biggest challenges that people have when focused on that.
Lyndsee Manna (14:17):
Yeah. And if you’re using two different EMR systems and they don’t talk to each other, that’s okay, but let’s make sure that you’re pulling that data into a unified database. So fine, they don’t talk to each other, no big deal, but you want to get all that data in one place so that you can use it. Because the data coming from both EMR systems are going to help you tell the full picture if you’re using two.
Chris Hemphill (14:39):
Indeed. So that opens the door for us to dig a little bit deeper on the new stuff that y’all are working on. And this three letter acronym you’ve been dropping NLG, natural language generation. I can’t wait to dig a little bit deeper on that.
Chris Hemphill (14:59):
So just curious, let’s say there’s a situation where for a given use case or for a given department, there’s a data set that’s available. Could you talk to us about, get into the basics on what natural language generation actually is and how it relates to data storytelling?
Lyndsee Manna (15:21):
Absolutely. It blew my mind when I saw it for the first time, and I know you referenced another talk that I did where I know exactly where I was when I saw it for the first time. Essentially, what natural language generation is, is you’re taking structured data and you’re turning it into words, so into written language, any language, but into written words.
Lyndsee Manna (15:43):
What fascinated me as an economist, as a lover of math, is I had no idea that my brain was doing mathematical calculations to physically say a word. And if I think about it too hard right now, it’s hard to speak because you’re like, wow, my brain is actually… But what we’re doing is we’re retrieving data, what humans do is I’m retrieving data from my database, which is my brain. That’s where my data’s coming from, right now. And I am mathematically calculating how to tell you what I know, what’s in my data base, in a way that you, Chris, and the audience today will understand. And my best way to communicate that is not by drawing pictures, is not by doing a silly dance, but is by speaking to you and telling you in sentences what I know. So I’m going to transfer my knowledge, that’s in my database to you. And that’s what I’m doing in words.
Lyndsee Manna (16:43):
And so that’s exactly the process that Arria replicates with our technology. That’s exactly what natural language generation is. So we’re taking data, with our software, that’s in a database, not Lyndsee’s brain, but another database, and we’re doing data analytics to that, data to create more data, because more data helps you tell a better story. And then we’re doing language math, some people call it language analytics. And this is the math that I never knew existed. And if I could go back in time, I would take lots of classes in university on this, because I just love it. But I don’t need to because I’m here at Arria, but it’s the math that we do to say words.
Lyndsee Manna (17:25):
And the easiest way to understand this is to go back in time, and you might not remember, when you were a toddler where you said something was big or small. Mommy that’s big. That was your brain doing a mathematical calculation and associating that mathematical result with a word. So very simple, straightforward, maybe a one year old, a one and a half year old says big or small. But as we get older, those mathematical calculations get more complex and we use word like over performed or outperformed or underperformed, or beat the trend, or whatever it was. So we start to use words that are more complex to describe what’s happening in data or in something that we know.
Lyndsee Manna (18:11):
And then not only do we choose the right words, but we decide how to put those words together, not only in sentences, but in entire thought streams. So all of us at some point learned about summarizing, having three paragraphs to state the facts and then summarizing again at the end, everybody learned how to write an essay. Our brains do that, too. As humans, when we communicate, I say, well, I want to tell Chris and this audience, these things, and this is how I’m going to communicate those things. And then I’m going to summarize it. And I’m going to say it in just a way that Chris and the audience can understand, because I know who I’m speaking to. So I decide what I’m going to say, how I’m going to say it and how I’m going to deliver that message so that you can understand my data story.
Lyndsee Manna (18:56):
And so as humans, we do that. And with Arria’s software, we replicate that process so that now you can get your data talking. Now, your data can automatically tell its data story in language that you can understand. So it can communicate what’s happened, how it’s happened, why it’s happened and in a way in which whomever we’re delivering that story to can really understand that message and then do something with it. So you’re really giving your data a voice.
Chris Hemphill (19:27):
I think that’s a good lead into a question that we got from Ellen in the audience, which was just around given this technology, I’m curious about who’s doing this well and effectively, and maybe even examples of what those data stories, these word streams and generation process, if there’s examples that you can share on what the actual output is on some of these important use cases.
Lyndsee Manna (19:54):
I’ll talk about specific use cases, but I think those that deserve acknowledgement are doing it. I think we have to start by getting started. And I think that that’s often the most… it’s the hardest thing to do is say, “Wow, I’ve got all this data. And I have so much that I want to learn from this data. So much I want to be able to communicate.” And it can feel overwhelming. Anytime you’re learning a new technology, adopting a new piece of software to add to your stack, a lot of people get so overwhelmed that they don’t do anything at all.
Lyndsee Manna (20:26):
The people that do it most effectively today, certainly have seen a huge pickup in the pharmaceutical industry, which is tangential to healthcare. And the ability to understand a lot of pharmaceutical customers that we have here at Arria, top global pharmaceutical companies, are using it to start on their financial data, understanding their business, understanding what’s driving revenue, where to focus their time, where to focus their energy, what to do more of what to do less of.
Lyndsee Manna (20:54):
So high level executive dashboarding right inside of business intelligence tools. It’s organized data, it’s accessible, it’s clean and it’s very NLG ready. So a lot of our customers will start somewhere where they have a business intelligence dashboard. So I would recommend for our healthcare teams that are there, if you’re using a BI tool, we’ve got add-ins for every major BI tool and out of the box in three clicks, you can get your data talking. So I would say start somewhere where you have very organized data. That’s where our customers are most successful is when they quickly add us to a business intelligence dashboard. It’s easy, it’s fast. And you get a lot of value very quickly.
Chris Hemphill (21:32):
Kind of makes me wonder, let’s say that I have this dashboard that’s telling me, I’m going to just give kind of a marketing example that, and I’m going to say things like volume and things like that. I’ve launched a program to reach out to people who might need cardiology services. And over time, let’s say that on my dashboard, I would probably have a line graph that shows the fluctuations in volume over time. And if I’m getting too excessive, then I would have all kinds of different things that break down various zip codes and things like that. But the questions that I’m going to have, and the questions that my leadership is going to have are, is this campaign, are these programs, is this outreach performing well? Are we driving the volume that we expect?
Chris Hemphill (22:27):
So I’m curious, if there’s kind of, when we go down to where it’s been used, what kinds of insights or what some of these stories that it’s been able to tell people who are using it.
Lyndsee Manna (22:42):
I think the most exciting thing about our technology is the human behavioral tendencies of a software that’s trying to be a human. We’re replicating the human process of analytics. So, when I’m looking at that line graph for the cardiology service, and I have a program or a campaign that’s looking to bring in volumes of leads, and possible buyers of that cardiology service. And I see a line that’s going up. I see a line that’s going up. The first thing we do, that’s not where we stop. That one picture, that one line graph, that’s not what humans look at. What I do as a human is I look at all of the data.
Lyndsee Manna (23:31):
So it’s not just one measure, one dimensional. We don’t think in a two dimensional way. How humans think is every piece of data that we can find, why? Why is that line going up? What were the attractors and detractors to the slope of that line? Maybe the slope isn’t as steep, but it’s still inclining. That takes time. That takes time to understand what actually drove that line to go up. What caused the slope to change. And we’ll get into performance in a moment, because I want to talk about, you said the word performance. Is it performing well? We’ll get there in a second.
Lyndsee Manna (24:08):
But just that alone, the NLG technology does that work for the human. So now, I’m empowered with understanding not only the fact that the line went up, easy. I could see that. The numbers are going up. That’s very easy to see and a picture helps us see that. But now, why? Exactly why is that line moving up? And how did that compare to the same time last year? Or how did that compare when we ran this campaign or this program to a similar body of people and at another dat? And so instantly now, to be able to understand in language, data storytelling in language, the ability to understand what’s driving that helps us pay attention to what’s working and what’s not working.
Lyndsee Manna (24:56):
Now, I do want to articulate, too, what happened? Why or how did that happen? And then that third piece, so what? Who cares? So what? Who cares? What’s next? So why is that line moving? What caused it to move? And now what do I do about it? That subjective piece, we can create words to drive those subjective pieces.
Lyndsee Manna (25:16):
But I want to talk about performing well really quickly, because you said it and it’s something that I’m very passionate about. I oversee the sales and marketing teams here at Arria, and it’s just a pet peeve of mine. When we get high clickthrough rates, the team that’s responsible for that goes, “We’re doing so good. This is a success. We’re performing well.” Wait a second. Clickthrough rates does not equal performing well. Clickthrough rates means something’s working, but performance ultimately, when you’re a marketeer, equals revenue. Yes.
Lyndsee Manna (25:55):
So performing well equals conversion. So if something that I’m doing is performing well, that means it’s turning into dollars from a marketing perspective. That means it converted to money in the bank. So when you look at a trend line of a campaign, you have to cross correlate that to the ultimate outcome, which is dollars in the bank. So again, you’re not just thinking about one piece, but you have to cross correlate that with all of the other information that you know, and I think that’s power comes, is when you can actually say this is working, we know why it’s working. We know who the buyer is and it’s converting. And tying those pieces together, it’s not a two-dimensional story. It’s really a story pulling in all the different data that you have access to.
Lyndsee Manna (26:45):
And I think that becomes the power. Here at Arria, we’re using data from Marketo. We’re using data from our social media tools. We’re using data from Google analytics and we’re using data from Salesforce. We pull that all together inside of our business intelligence tool to tell a comprehensive, connected data story. And that’s what I think becomes very powerful, Chris. It’s not just about that one line graph.
Chris Hemphill (27:11):
Excellent. So when we think about that, and again, the line graph I was talking about was in the healthcare niche. I’m looking at some chart, but really there’s all kinds of different things that influence where that line is going and all that. What would one of those data, what would the stream say? How would it word one of those insights that tells us why? Just curious.
Lyndsee Manna (27:37):
It’ll say something to effect of, this line is trending upwards at either a faster or slower rate. If you want to pull in another time period compared to the same time last year, this increase in trend, or increase in sales, say it’s sales data or engagement data, was mostly driven or largely contributed by these things. This is what drove that line to go up. And these were the detractors that prevented a steeper line from occurring.
Lyndsee Manna (28:12):
What analysts will do, and this is obviously the analytics that are programmed into our software, is we’re going to look at what caused that line to go up and what caused that line from not going up as much as it could have. Because you not only have to pay attention to what’s working, but you always have to pay attention to what’s not working. Because in what’s not working is an opportunity to make more things work.
Lyndsee Manna (28:36):
So, always, we’re going to look at the top and bottom. And then also part of that story would be something potentially that’s harder to see. Sometimes, for example, and maybe you’re running a campaign and we’ve had this happen here at Arria, that is very, very effective from a marketing perspective, but you have a very low spend there. So let’s call it a Google Ad campaign. And you just haven’t put a lot of dollars into it, but the dollars that you did put into it performed very, very well. But it wasn’t your largest reason why that line went up and it wasn’t your big detractor. It was this little piece of information that could be easily missed. It’s in that gray space and you might not know to pay attention to it.
Lyndsee Manna (29:22):
So there’s also a lot of algorithms that we have built into our software to pay attention to the things that you might miss in the middle and say, “Well, this campaign performed well, however, your spend was low. So you might want to take a look at that.” So it’s really about churning through that data, doing the analytics that humans know how to do on that data, but having our software do it, and point out the insights that are important, to impact whatever your objective is.
Lyndsee Manna (29:54):
High level, it’s going to talk about what’s going on. The trendline is increasing. It’s going to talk about what attributed to that, what detracted from that, and then some other key things you might want to pay attention to.
Chris Hemphill (30:05):
Lyndsee Manna (30:05):
And it’s looking at all the data, not just what’s in that line chart.
Chris Hemphill (30:08):
Lyndsee, I got to say, I’m really thankful that you came to discuss this. And one thing that just reading about your profile and background and all the different things that you’ve done, I feel like there’s a reason, something that you might have wanted to stick with the audience by having this conversation, by talking about data literacy and by talking about digging down into the why of how different metrics perform and things like that. So just curious if there’s any kind of final thoughts that you’d like for people to reflect on for the rest of this week?
Lyndsee Manna (30:46):
Well, I said in the beginning, I’ve always had a passion for helping other people, so I think that we need to be stewards of our data and think about how can we leverage what we have access to, how can we empower others with that? I think the final message would be a message of empowerment. Let the data help you help yourself so that you can do more, so that you can have more time spending time doing things that are humanlike, like building relationships, like being conduits of change in your organization, and less time combing through data. There’s plenty of technology that can do that.
Lyndsee Manna (31:30):
So you want to help people to understand, help people have comprehension, so you can be a change agent within your organization. And ultimately you’re helping other people that are struggling to understand the data, you’re helping your business perform better. And you’re helping yourself to spend more time doing the things that humans should be doing and humans love to do. So I guess that would be my last message, Chris, a message of get empowered, let your data work for you, so you can do more.
Chris Hemphill (32:03):
Well, apparently your message falls on very welcoming ears. Amy Hand, thank you for praising the discussion and things like that. I think I want to reemphasize a point that you had made earlier about the fact that being data driven is for everyone. You don’t have to have gone and gotten your masters in math to be able to communicate, tell data stories and help organizations make the right investments when it comes to growing patient care. Lyndsee, super appreciate you coming in and joining us.
Lyndsee Manna (32:41):
Thanks, Chris. Excited to be here and happy that you had me and hopefully we get to do it again soon.
Chris Hemphill (32:47):
We hope you’ve had fun learning about the future of data storytelling with Lyndsee and me. If you want to learn about them now, you might want to check this video out. Ryan Younger, the VP of marketing at Virtua Health, shared the reports and the metrics and the techniques, strategies that he used to help get their executive team excited about running AI driven campaigns across more service line.
Chris Hemphill (33:10):
Data storytelling is one of my favorite topics, but if you have another topic that you’d like for us to discuss, let us know in the comments below. Again, we appreciate you spending some time with us, and we’re going to keep the conversation going by talking with healthcare leaders at the intersection of data and healthcare strategy. So, until next time, hello.
Find the Clarity You’ve Been Missing
Learn how Actium Health is driving improved quality, outcomes, and revenue for innovative health systems nationwide.