Web Masters #43: Robbie Allen


Automated Insights - Wikipedia

Robbie Allen:

They would bring in all of these speakers during lunch to give talks, and so I got to hear all sorts of just fascinating entrepreneurs come in and give their story. But the one thing that I heard pretty consistently is, “Follow your passion.” If you’re not going to be passionate about the thing that you’re working on, when the going gets tough, you won’t put in the time to keep going. And so, that really resonated with me because anytime I could get passionate about something I would do whatever it took to be successful. That was how I had gotten to the point I was in my career.

But anyway, I decided to make a list. All right, what are the things that I’m passionate about? And the list was: web programming, doing interesting things with data, and UNC Basketball. Those were the three things. I wrote a list and that’s what I came up with. So was there something I could do with that? Web 2.0 was just getting rolling, this was 2007, 2008. And so I thought, “Well, there’s nobody that’s really done much with statistics on the web.” You have your standard box score and all of that, but no one really had interesting graphics and interesting stats that they were presenting.

And so I thought, “All right, well, let’s do that. I’ll build something that ultimately became StatSheet, where I’ll just do all sorts of interesting statistics and graphics, essentially, to help satisfy my curiosity about UNC Basketball, but the nice thing since I’m programming it, I can make it available for every college basketball team.” And that’s how StatSheet got started.

Aaron Dinin:

You just heard Robbie Allen describe how he built a company called StatSheet Inc. It started as his way of combining passions for the web, data processing, and college basketball. StatSheet would go on to create a huge network of sports websites, probably one of the largest in the world. And the reason they could do it was because all their content was automated. It was created with a custom natural language generation algorithm that enabled computers to write like humans.

Unfortunately for Robbie, sports websites filled with computer generated content wasn’t a great business opportunity, but automated content creation, well, that was a much bigger business opportunity. Robbie recognized it, and the result was a major pivot into a company called Automated Insights, the pioneer of automated online content generation. Are you ready to hear the story? Let’s get dialed in.

[INTRO]

Aaron Dinin:

Well, hello, and welcome to another episode of Web Masters, the podcast that explores entrepreneurship by talking with some of the internet’s most impactful early innovators. I’m your host, Aaron Dinin. I’m a serial entrepreneur, I teach entrepreneurship at Duke University, and I create lots of content about entrepreneurship that I post online. I mention this because today’s guest founded a company that automates online content generation. So yeah, is this the guy who’s going to put me out of a job? We’re going to find out right after I tell you about our sponsor.

Web Masters wouldn’t be possible without the generous support of Latona’s. Latona’s is a boutique mergers and acquisitions company that helps people buy and sell cashflow positive internet businesses and digital assets. I’m talking things like eCommerce stores, SaaS apps, Amazon FBAs, domain portfolios, and yes, I suppose even enormous content networks created by automated writing bots. If you’ve got a business like any of those, its profitable, and you’re interested in selling it, then contact the team at Latona’s so they can help you find the best deal. Or if you’re hoping to buy a profitable internet business, be sure you check out the Latona’s website where they’ve got listings for all the companies currently being sold. That website is latonas.com. L-A-T-O-N-A-S.com.

There’s a decent chance you’ve never heard of Automated Insights, and that’s okay. Unless you’re looking to produce enormous amounts of long form content using standardized data sets, you’re not really their target customer. So before I tell you more about the company’s founding CEO, Robbie Allen, and how he built the business, I’m going to let him tell you what Automated Insights does.

Robbie Allen:

Automated Insights is an automated writing technology company. And so what we would produce are stories that sound like a person wrote them, but the input was data. And so instead of having a writer look at data and then write a summary describing that data, we would create algorithms, programs essentially, that would take in that data, again look for what were the interesting patterns or anomalies in the data, and then describe it in language form.

Eventually, we morphed the company, because one of the challenges with that is we were deterministic. That is we had to essentially hard code the algorithms, it wasn’t a learning system. And again, there’s some reasons for that. When you’re generating a earnings report like what we started to do for the Associated Press, and we’re describing Cisco’s earnings reports, it has to be 100% accurate and the language has to be perfect because we’re essentially replacing what an AP writer would have written.

The problem with things like machine learning is that it’s not deterministic. In fact, you can’t be entirely sure what it’s going to produce. And in fact, it could produce something that’s not coherent. And so for the use cases that we had it had to be precise. And so what that required is programmers and data scientists at Automated Insights to craft algorithms that knew what to look for and knew how to describe it. Now, again, we had lots of tooling so that it sounded very… It was highly variable and it didn’t sound formulaic. But essentially, that’s what we provide to companies.

Aaron Dinin:

So, sure, even though you may never have heard of Automated Insights, there’s a decent chance you’ve read something created by their software, and probably didn’t even know it. How cool and/or creepy is that? And as interesting as the technology is, I think Robbie’s entrepreneurial story is even more interesting because it doesn’t follow what we might call a typical path.

Robbie Allen:

Even though I’ve started several companies, I didn’t grow up in a very entrepreneurial household. In fact, both my parents worked in manufacturing. And so I didn’t think about starting a company until I moved to Silicon Valley many years later.

Aaron Dinin:

If you’ve been listening to Web Masters for a while, you’ve probably heard me point out a common pattern with lots of guests on the show. The majority have had some sort of entrepreneurial influence as a child. Usually, it’s a parent or close family member who started a small business or something like that. This early exposure to entrepreneurship indirectly taught my guests that they too could take the interesting market opportunities they’re discovering throughout the course of their day-to-day lives, and turn them into businesses.

In Robbie’s case, his ultimate career trajectory is unusual because he wasn’t even exposed to entrepreneurship until he was already well into his professional career. The result is that Robbie was solving difficult problems from a young age, but he didn’t understand that the solutions he was creating for himself were solutions other people might want too.

Robbie Allen:

In 1995, I was a freshman taking computer science, and I was just kind of messing around, there was this thing called Java that was kind of popular at the time, playing around with Mosaic and Netscape. I became a system administrator in some of the computer labs in college, and so I had plenty of time just to mess around and play around with the different applications at the time. But I was starting to use email, and there was, I think Pine was my email application of choice. Back then it was a command line driven email client.

And I thought, “It’d be really neat if I could do this but on the web, right?” And again, this was 1995. And so actually wrote a web-based email client that was driven on the back end, it was just essentially a web interface to Pine back in 1995, and that was my first web programming.

Aaron Dinin:

And did you share that with anyone, try to turn it into a business?

Robbie Allen:

That was just for me, that wasn’t even something that dawned on me as an option. I was just building something because I loved the web even in 1995 and 96. And I was getting tired of using Pine for email, and I figured is there a way to just do it via the web? I didn’t even think, “This would be kind of a neat thing to productize and maybe launch to the masses.” That would have been pretty cool to do back then.

Aaron Dinin:

For reference, Hotmail and RocketMail, the two companies largely recognized as the first web-based email services were both launched in 1996. So if Robbie was developing a web-based email platform in 1995, well, I’d say yeah, it would have been pretty cool to launch it to the masses back then, but he didn’t because it didn’t occur to him. Instead, Robbie followed a more traditional career path. He took summer internships during college, and that started him on his professional programming career.

Robbie Allen:

By junior year I did an internship at IBM in their Networking Hardware division, which they no longer have, but it was essentially their router business. This was in Research Triangle Park, North Carolina. And then I had heard about another company that was right down the road in RTP called Cisco, again this was 1997. I heard that they were an up and coming company, but at that point they were actually pretty large. But I did an internship the next summer at Cisco.

And then, then this was middle of the dot-com boom, Cisco told me that they didn’t care that I didn’t have a degree yet, they needed programmers as fast as they could find them in Silicon Valley. And so they said, “Do you want to move to Silicon Valley, and you can finish your degree in your spare time?” And I was like, “Cool. Yeah, that sounds like something I’d be up for.” And so I moved out to San Jose and started my corporate career.

Aaron Dinin:

Okay, so you moved to Silicon Valley, and I’m guessing that’s where you got exposed to entrepreneurship?

Robbie Allen:

Yeah, so I started writing some technical books. I ended up writing 10 books for O’Reilly Media over a five year period. And so that got me into the conference circuit.

Aaron Dinin:

O’Reilly Media, for those of you who don’t know is one of the leading publishers of technical instruction books, so books on things like programming languages and networking infrastructure, and other sorts of nerdy developery, engineery-type things. The fact that Robbie wrote 10 of them, and he started when he was in his 20s, that’s well, pretty impressive. They’re usually written by leading technology experts.

Robbie Allen:

A lot of people think that you write books because you know so much, I decided that I wanted to write books so I could learn, because there’s no better way to learn than to write about something. And so this was back when I was sort of super eager and interested in learning new stuff, kind of an insatiable appetite for learning new things. And so I just started writing many books on lots of server technologies, things like DNS, DHCP, server administration. And again, while I was doing that I was also doing a lot of programming around that.

Aaron Dinin:

That seems kind of intense. I feel like most people read books to learn about certain subjects, not write them.

Robbie Allen:

I was, again, in Silicon Valley during the dot-com boom and bust, and just the crazy brain I had at the time was I lived about, during rush hour it might have been 45 minutes to get home. And so my thinking was, “Well, actually, why don’t I just sleep at work maybe three nights a week? And so that way I don’t have to waste all that time driving to and from work. I can just work straight up till midnight, go to sleep, get up at six, and then I’ve reclaimed a couple hours of productivity, because I didn’t have to do all that driving.”

Aaron Dinin:

You might hear these stories and think Robbie is a really talented and dedicated Cisco employee, I’d argue they’re more like signs of Robbie’s innate entrepreneurialness. He’s self-educating. He’s optimizing for productivity time. And through his book deals, he’s figure out alternate ways to generate income beyond his main job. He just hasn’t quite figured out how to turn these skills into a new venture, but he was about to.

Robbie Allen:

I got to know some startups and advising them, and eventually I was like, “You know what, I think I could do that.” I hadn’t really thought about it before but there’s no reason why I couldn’t go start a company at some point. And so I actually decided, again, this is a sort of the engineer in me, I thought, “Well, I don’t know anything about starting a company so I need to go learn about it.” I didn’t know that that’s not really how entrepreneurship works. But I always wanted to go to MIT because I’m in computer science, and so why don’t I go to MIT and learn about entrepreneurship there and then I’ll come out and start my own company. And so that’s essentially what I did. I started a graduate program at MIT in 2004.

Aaron Dinin:

As you heard at the beginning of this episode, MIT was where Robbie got exposed to lots of entrepreneurs who instructed him to follow his passion. And that led to his list of three passions: the web, data, and college basketball.

Robbie Allen:

After I had finished with MIT, I was like, “Okay, I’ve got this fancy degree now. I should know what I need to know to go start a company, what am I going to do?” And one of the great things about MIT is they would bring in all of these speakers during lunch to give talks. And so, I got to hear all sorts of just fascinating entrepreneurs come in and give their story. But the one thing that I heard pretty consistently is, “Follow your passion.” And so that really resonated with me, because anytime I could get passionate about something, I would do whatever it took to be successful, right? That was how I had gotten to the point I was in my career. But anyway, I decided to make a list. All right, what are the things that I’m passionate about? And the list was web programming, doing interesting things with data, and UNC Basketball.

Aaron Dinin:

Robbie’s list inspired the company you also heard about at the beginning of this episode, a company called StatSheet Inc. It began as a company dedicated to providing statistics for sporting events, but Robbie quickly realized that wasn’t as big or lucrative a market as he’d initially thought.

Robbie Allen:

What I learned pretty early on was while there are some people that are passionate about sports statistics, it’s actually not a very large audience in internet terms. And in fact, sports in general is not a really large audience in internet terms. That is, if you’re trying to make an internet business that’s subscription-based or ad-based at the time was especially popular. And so I thought, “Well, I mean, I got to do something more than just showing interesting stats or I’m never going to make stat sheet into an interesting business.” By the way, I was still working at Cisco during this time.

But I figured, if I could get StatSheet to be big enough, or interesting enough, or successful enough, then I could leave Cisco and just focus on that. And so essentially, I was working two full time jobs at this point. But I thought, “Well, something else that I’ve done a lot of is writing, right?” I wrote all these books, all these technical books. I actually started the first blog dedicated to UNC Basketball in 2002, called UNC Basketball Update. That’s another little piece of internet trivia that no one cares about. And so I had written a bunch of UNC Basketball recaps.

And so now I had all this data, and I had written all these recaps, and the recaps were kind of formulaic. I mean, it’s essentially you’re talking about which players performed the best. And for some definition of best, that is who had the most points or the most rebounds, or maybe performed better than their average performance. Or did UNC lose when they should have won or won when they should have lost? Those kind of things.

And I thought, “Well, I have all this data, and I’m a programmer, and I’ve done all this writing. Can I just combine all that so that instead of just presenting numbers, which again there’s a limit audience for, could I automate these recaps so that now I can just present the numbers in a more digestible human form in terms of language so that more people would be able to digest that?” And that’s essentially was the very beginning of the automated writing that I did, was taking data, creating the scenarios that we could essentially identify, what were the most interesting things that occurred in a game, and then expressing it in English.

Aaron Dinin:

How were you thinking about turning those long form stat summaries into a business?

Robbie Allen:

Well, I started off with just trying to do advertising based. It was again, 2007, eight, and nine and internet advertising was still a very common mode of trying to monetize content. And so we thought, “Well, again, if we can generate lots of content, then that makes more places for us to have impressions.” What we thought at the time, this was right around the time, SB Nation, if you’re familiar with them and some other sites, where they started doing team-oriented websites where they would instead of having espn.com, there would be a site that they would stand up specifically for one team.

Robbie Allen:

And I thought, “Well, actually, we could do that. I mean, they have to have hundreds of writers to do that, but we don’t need the hundreds of writers. Why don’t we stand up team-specific sites and then we could populate them with all the stats that we had, and as well as all this content that we’re producing?” And so we started down this path of creating literally hundreds of websites, hundreds of Twitter handles, where we would just start pumping this content on the websites through Twitter.

And again, we created team-specific Twitter handles. This was in 2008, nine, I got a little bit of trouble with Twitter over that. That was some of the first time sports content was being sent through Twitter. But the whole idea is can we create a large advertising-based business that’s maybe similar to something like SB Nation, but it’s a fraction of the cost because we don’t need all the writers?

Aaron Dinin:

And so initially, it sounds like you were trying to do an automated content SEO type of play. How successful was that strategy?

Robbie Allen:

Well, it was kind of a mixed bag. It was we got written up in the Thanksgiving edition of The New York Times, I think this was maybe 2010. And that was a huge deal. They said the robot writers are coming for our jobs. And that’s when I started to see that, at least in the press, that this was going to have a lot of legs, that is the notion that couldn’t we automate the writing process? What I found, and this wasn’t something I thought going in, but journalists loved to hear about that. They loved to write about that story. And what I found is… I mean, it makes sense, right? People like to talk about their own profession, and this was a profession that maybe they were a little nervous might be at risk due to this new technology.

And so we always had just a ton of success getting in the press and getting these stories written up about what we were doing, because journalists were just fascinated that there might be something out there that could replicate what they do. But anyway, we did all of that. It was pretty clear that even though sports was, there was a lot of passionate people around, we found that what we were doing, that is taking data and generating quantitative analysis out of that data applies to more than just sports.

And so maybe there’s a bigger play here that instead of us trying to scrape for these sort of internet digital dimes through advertising, what if we became more of a B2B company and focused on taking large companies’ data and turning it into quantitative analysis? And they probably would be willing to pay pretty well for that relative to what we were seeing in terms of internet advertising.

Aaron Dinin:

So I’m guessing that’s basically the point where you started transitioning into the B2B company that became Automated Insights. Is that right?

Robbie Allen:

Yeah. So when I raised the series A, we had decided at that point to change the name of the company. So we changed it from StatSheet to Automated Insights. We were going to keep statsheet.com up and running, which was probably a mistake, because it just became kind of a distraction even though it was sort of our pet project. We were going to change the name, change the business model entirely, change essentially everything we were doing. And when we raised that series A, that’s kind of what marked that happening.

And it was a pretty big deal for us at the time, because normally, by the time you get to your series A you’ve answered a lot of questions about your market and all that. We were going to enter a brand new market. Now what we had going for us at the time was a lot of people were interested, including investors around this technology. That always captured the attention of a lot of people. So we were able to continue to get the benefit of the doubt even though we hadn’t answered some of the questions around our market just because we had this compelling new technology.

Aaron Dinin:

In the entrepreneurial world, when a company makes an enormous shift in business model and customer focus, we call that a pivot. You’ve probably heard the term before. Honestly, most of the things people call pivots aren’t really pivots, they’re either minor shifts where a company maybe starts targeting a slightly different market with the same product or they’re basically just entirely different companies. What Robbie did with StatSheet to Automated Insights is like a textbook example of an actual pivot.

Robbie realized that his technology and the core concept of his business weren’t being used to their fullest potential, so he repurposed that technology to produce a different solution he was able to then sell to a different market using a new business model. In other words, the core concept behind Robbie’s innovation never changed. He was still creating automated texts, but he completely changed how he monetized on that innovation. While this type of complete shift in the company’s focus might seem extreme, according to Robbie it’s actually pretty natural.

Robbie Allen:

So I’m definitely in the camp of when you’re just starting a new business, and I hear it all the time from entrepreneurs, “All right, we got this figured out. We’re doing this playbook, it’s one to 10. And at the 10, we’re going to be in great position. And in three years we’re going to be ready to do this or that.” And I mean, it’s great to have a plan, but don’t go in thinking that you’re going to follow this plan and nothing’s going to change along the way. It’s almost inevitable something’s going to change.

You’re going to learn things that it’s going to cause you to maybe make absolutely small changes, but in some cases like it’s been with me, significant changes, all the way to include changing the name of the company and the entire business model. And so I just, I’ve seen this play out so many times. It’s just so rare that an entrepreneur starts with an idea and they end with that exact same idea. It’s almost always the case that you’re going to change it in small and large ways along the way.

Aaron Dinin:

Out of curiosity, do you have any advice for entrepreneurs that can help them know when it’s time to pivot?

Robbie Allen:

It’s hard because for me I tend to be a little bit of the grass is greener on the other side. It’s always challenging when you’re in the trenches, and especially if you don’t have product market fit. And a customer maybe asks for something that you thought about before that’s a bit of a tangent for you, but you didn’t want to do it because it’s not in your core business now, but you’ve heard it a couple times. And you’re like, “Oh, man, maybe we’d get product market fit if we just did that thing now.”

And so I get this question pretty frequently around, “Well, how do you know when to make a pivot?” Since I’ve done it so many times. Part of it is just my nature. I tend to want to pivot until it’s just obvious that you’re hitting the thing. But there’s also a lot of risk in that. Maybe you just needed another six months in your current thing. And so unfortunately, there’s not a clear answer as to well, these are the three things that happen right before you need to pivot.

It’s much more of an intuition and kind of a gut feel unfortunately. Again, I’m an engineer, these are the aspects of entrepreneurship I don’t like, like how much luck is involved, and the fact that it’s not always clear which decision you should make based on the data. Sometimes you just kind of have to go with your gut feeling on these things.

Aaron Dinin:

So Robbie went with his gut feeling when he pivoted from StatSheet to Automated Insights, and that turned out to be a great decision. It was proven when they got one of their first big customers.

Robbie Allen:

One of our big early successes was with Yahoo. In fact, this is probably the biggest success on terms of a visibility standpoint. We started producing fantasy football recaps for them. And this was just a perfect use of the technology. So Yahoo was the largest fantasy player at the time, and what they would do is they’d have several million fantasy football players in their fantasy application, and they would play head to head every week. And then we would generate a recap that sounds like a reporter wrote a recap analyzing why your team beat my team, and the boneheaded move that Robbie made by sitting a certain player when he just went off for 50 points or something like that.

And again, this was a perfect use case because it really factored in the scale advantage that Automated Insights have. You can never crowdsource writing to do all of that analysis for fantasy football. There’s not enough people to generate eight million recaps every single week, but for us once we had it down, it’s as easy for us to produce 1,000 as it is a million. And so what we really started to try to find are, are there other types of use cases like that where you’d love to be able to write about certain things, but it just doesn’t make sense from a scale component? And so that’s where we started focusing the business.

Aaron Dinin:

But even though Automated Insights had a lot of early success with Yahoo, getting customers still wasn’t easy because of the uniqueness of what they were doing.

Robbie Allen:

We had to go so far as branding a new industry. And so natural language generation, which is it is a component of artificial intelligence that’s been around a while that even predates what I was doing at StatSheet a little bit in the academic world. I mean, it’s not been heavily sought after space, but you definitely can find pockets of activity around NLG. But we had to brand this in the corporate world, right? And so that was challenging to say the least.

Aaron Dinin:

By saying it was challenging, do you mean it was challenging because you were having to first educate the market?

Robbie Allen:

Yes. With us, it wasn’t even that we had to just educate the market, we had to prove to them that this was even possible. So the machine learning era, the latest one, really started 2012, 2013. We were out selling, “Hey, we have automated writing ability in 2010.” And the first questions we inevitably get is, “What? You have what?” There is a disbelief that what we were selling was even possible, and that’s just not a great place to be because it’s great if you can say, “Well, we have these features and this is what we can enable you to do, and this is why we’re better than the competition.” We weren’t even there. We were at the point of, “No, it actually does work, and let me tell you how.”

So we could always get a lot of interest, much like on the journalist side. We could get conversations with customers, but again, there was almost this disbelief, because autonomous cars, all the machine learning, none of that had even come to the fore yet. This was we were pre all of that, so there is still a lot of people that weren’t even tuned yet that this is the type of technology that’s coming down the road.

Aaron Dinin:

Here, Robbie is describing the downside of something in the entrepreneurial world that’s usually touted as a positive, which is first mover advantage. First mover advantage is the belief that when you’re first into a market you’ll have an easier time capturing lots of customers because you don’t have competitors. While that can be true, being first into a market also has some distinct disadvantages as Robbie found out.

Robbie Allen:

It’s funny because I always like doing new things that have not been done before. I could go out and create a similar technology to lots of other companies, but that just wouldn’t be something I’m interested in. I like doing something that no one’s ever seen. And when I first became an entrepreneur, I thought that was a good attribute because of the first mover advantage. But at least what I found out to be the case, at least for my companies, maybe there’s other entrepreneurs that would disagree is there’s actually a first mover disadvantage. And we came up against that a lot with Automated Insights.

I started StatSheet in 2007. Let’s fast forward to, I don’t know, 2011 or 12, and we’re still trying to figure out who is our ideal target customer. So we had spent five years and hadn’t even really gotten to the point of figuring out the ideal customer because no one had ever done this before. We couldn’t just go in and say, “All right, we’re going to do something that’s similar to Salesforce, but slightly different. And we’re going to go after a similar customer, and we know that profile.” None of that had been figured out.

And so for me, part of the first mover disadvantage is the concern that I’m going to spend all this time and money trying to figure all this out. And I was never worried about like the Googles of the world or any of the larger players. I was more worried that a small startup was going to come in and say, “Ooh, we’ve been following what Robbie has been doing for the last five for six years. We can see where they’re going, so why don’t we just go ahead and start with that?” Or just waiting until we start to get some traction in a market like, “Okay, let’s just start the company there as opposed to having to go through all of these years.” And again, we were still running statsheet.com, which was something that was taking time and resources. And so for me, it was harder because we were constantly trying to figure this stuff out in a brand new market.

Aaron Dinin:

But clearly, you eventually figured it out. So how’d you do it? How did you ultimately start selling the product?

Robbie Allen:

So it took many years before we would have a first conversation in a company and they wouldn’t question it? Or my favorite question of all, we got this so frequently, we’d describe what we do and we’d show them here’s an example of what it produced. And the first question we’d get back, “So who wrote that?” And we’re like, “All right, so if we get that question, then we we’re just not even in the ball. The understanding of what we’re doing is just still not there.”

But I would say, again, the Yahoo implementation did the most for us out of anything we ever did, because what happened was turns out a lot of business people play fantasy sports. And they would get these recaps, and down at the bottom we were so lucky we were able to negotiate in that we would say powered by Automated Insights with the link back to our website. And so business people would read these recaps and they loved it. And especially if we made them snarky, which Yahoo wanted us to do, and/or if we wanted to make jokes about certain things.

And so they’d read this, and they would instantly get that this wasn’t written by a person, right? Because they understand there’s millions of players, there’s no way that this was written and this it says, “Powered by Automated Insights.” They’d click the link to Automated Insights and they’d go to our website. And so sure enough, every year in September, traffic to our website would just be typical wherever it was, and then it would just instantly take off as soon as fantasy football season hit and you had these people reading these recaps and going to the site.

And so if we ever had somebody coming in through Yahoo, we didn’t have to explain that this wasn’t written by a person, they got it. So, it took many years of that. I mean, we had to have enough people seeing those, then the stories about autonomous driving and machine learning and all of that, 2013, 14, 15. Then people started to… Once they got enough of that in their daily diet of reading materials, then it wasn’t like, “Oh, how is that even possible? I don’t get that.” They thought that maybe it wasn’t something they’d heard of before, but it just didn’t sound so foreign, because now they’re hearing that their car is going to be driving itself someday, that this just seemed like yet another thing that was of that ilk.

Aaron Dinin:

And what are some examples of other types of customers beyond sports who wanted automated content generation?

Robbie Allen:

So we got connected with the Associated Press, they ended up becoming an investor in one of our later rounds. And we did automated earnings reports, very similar to sports recaps. The AP would cover maybe 200 companies at most in terms of the quarterly earnings statements, which is one of their most consumed article types, but yet there’s several thousand public companies in the United States. And so all those companies just weren’t getting covered. And in fact, there were all these studies that had been done out of Stanford that showed as soon as an organization like the AP would start covering the quarterly earnings report for a company, their stock price would go up, because now they have all these institutional people that would start following them and would be interested.

And so there was all these interesting things that came out of that, but the AP was another one. We had a whole bunch of different use cases inside of companies, everything from analyzing ratings data, TV ratings data. And that was one of the challenges, frankly. This is another one of those things you hear about that sounds great. It’s awesome if you have the technology that can apply to a bunch of different use cases, except the fact that that’s a big distraction. Well, which one do you focus on? I mean, do you just do the sports one? In fact, this automated writing ability, where does it apply? Well, it could apply to just about every company, right?

And again, that sounds great on the surface, but then when you have a salesperson come in and say, “Okay, where do I go start?” And you’re like, “I don’t know, any company that you can think of could use this. So why don’t you go sell to them?” And so, we spent a lot of time trying to refine our message. We went in and out of different markets. Eventually, Automated Insights really focused on the business intelligence market, because they have tons of data, they traditionally had been generated graphs and things like that. And we thought if we could do it via narrative, that would be more powerful. Really, it was just kind of getting at the core markets that have lots of data that traditionally haven’t had the ability to produce narrative about that data where we could come in and maybe help out with that.

Aaron Dinin:

I’m also curious not just about your customers, but the consumers reading the content. How do they feel about reading computer-generated texts?

Robbie Allen:

Well, something that we wrestled with frequently, some of our clients they said, “All right, we’re going to put like Yahoo did: ‘Powered by Automated Insights’. Let’s just make it clear that this was not written by a person and it was written by a technology.” There were some that wouldn’t put any attribution at all on the page. My promise back then was that I could give you an article. If you didn’t know it was written by a technology, I could show it to you and you wouldn’t think it was written by a technology. It wasn’t something that was obvious that you’d read and be like, “Oh, yeah.”

And then, there were other cases where they would put a human name on it for attribution. Now, what I will say is that in the cases where it was a story… And we used to get this, maybe even some with Yahoo just because we were generating so many. If the end user knew it was written by technology, they were so much more over the top in their critiques than any of the other use cases. And so it was funny sometimes, we’d see maybe there’d be a bad write up, like there’s a journalist that said, “Oh, these are so obvious, and the writing is so poor.” And they would go off because we had a dangling comma or something, because there was some algorithm that was messed up.

And then what I’d do is I’d click through and go to that person’s blog, and it was just filled with grammatical problems. And I was like, “Wow, you spotted one issue in this automated writing, and you call it out as just not being valid at all, and yet I see all sorts of grammatical issues in your writing.” And so it’s just interesting to just kind of think about and see how that played out in terms of how hard people were on technology when they saw that, and how easy they were to dismiss it. But then nine times out of 10, you could show somebody an article and they wouldn’t even know that it was written by technology.

Aaron Dinin:

All right, since we’re comparing this to real writers, I feel like I have to ask, the technology couldn’t have been perfect the entire time, any major issues that cropped up as you were training it to work better? Anything you can share that sticks out in your mind?

Robbie Allen:

There was especially in the Yahoo ones, they would get feedback sometimes because I guess there were some people that thought that Yahoo was writing about their team, their fantasy team, they would tell us that if they didn’t do well, say that they performed the worst in the whole league that week or whatever. And they started getting feedback from some of the users, “Hey, why are you saying such negative things about my team, Yahoo? I’m doing the best I can.”

We would get emails from time to time and then sometimes they would just send those to us. Like there were some times that we’d get emails, like the dad said, “Hey, you sent this recap to my son, this is his first year playing fantasy, he’s doing the best he can. Can you just take it easy on him this year?” So there were all sorts of incidences like that that was just comical. People thought that, “Maybe Automated Insights was writing these stories, and they’re being mean to me.”

Aaron Dinin:

Yikes. Yeah, having built some automated tools I can see how stuff like that becomes an issue. You just can’t account for all possible contexts. So those were examples of the end user responses. And what about the competitors? Specifically I mean, the human writers Automated Insights was replacing, how have they responded over the years?

Robbie Allen:

I never thought that we’re going to replace human writers necessarily. All that we’re going to do is actually replace the things that human writers don’t like to do and aren’t that good at, which is exactly where you want to be with technology. We can have technology that writes the five million fantasy football reports every week. We didn’t have one person argue with us that they missed writing earnings reports every quarter. Generally, reporters don’t like writing that stuff. We should focus on having people write things that are uniquely human, and what humans are good at which is more of the qualitative side of things, and leave the quantitative stuff to computers. I mean, that’s what computers are good at.

Aaron Dinin:

And that’s the fundamental truth of automation. Computers are great at repetitive tasks with clearly defined guidelines and rules. They’re horrible at dealing with ambiguity. In contrast, writers and content creators, people like me, love ambiguity, and we hate repetitive systematic content creation. Automated Insights is a company that handles the repetitive stuff so all the content creators out there can spend more time creating the unique, nuanced and non-automatable content y’all love like this podcast episode, because you did love it, right? And since you did, be sure to share it with other people so they can love it too.

I want to thank Robbie Allen for taking the time to tell his story and the story of Automated Insights. In case you’re wondering, he’s no longer with the company. It was acquired in 2015. He stayed on for another couple of years before deciding after a decade working on the same project, it was time to move on to new challenges. The project he is currently working on by the way, automates the launch of startups. It’s called Startomatic. So in case you’re keeping score at home, Robbie has gone from making my writing career obsolete to making my teaching career obsolete. So yeah, that’s cool.

I suppose if you want to help him do that or see what he’s up to now, you can follow him on Twitter, he’s @RobbieAllen. This podcast is on Twitter too, we are at @WebMastersPod. Feel free to send a message and let us know what you thought of the episode. You can also find me on Twitter, I’m @AaronDinin. That’s A-A-R-O-N-D-I-N-I-N. And I spend lots of time creating 100% non-automated content about startups, entrepreneurship and business over on medium.com. Just search for my name there and you’ll see all of it.

A quick thanks to our audio engineer Ryan Higgs for his help putting together this episode. A thanks to our sponsor Latona’s for their support. Remember, if you’re interested in buying or selling an internet business, be sure to check out latonas.com. And finally, a thanks to all of you for listening. Make sure you’re subscribed to Web Masters wherever you listen to your favorite podcasts, so that you’re sure to get the next episode as soon as it’s released. It’s going to be another great episode. I promise you’ll find out why soon, but until then, well, it’s time for me to sign off.

Leave a Reply

Your email address will not be published. Required fields are marked *