Tasked with growing Auth0’s reach and engagement, systems engineer and marketer Martin Gontovnikas chalked up a long list of trial and error experiments.
But as he explains in this talk from DevRelCon London 2017 the hard work paid off and resulted in a strong audience of loyal customers.
So my name is Martin Gontovnikas, but everybody calls me Gonto. My mommy is pissed off because she really likes Martin, but I’m Gonto, sorry. We’re in the tech world, so my Twitter handle is @mgonto. If you wanna ask me a question or follow me to increase my ego, I’d be happy with any of them. And then the title of the talk is “The Power of Trends.” And it’s, basically, more around what is the experience that we’ve had in Auth0 for both content and evangelism, and this is just a friendly name to get accepted into conferences.
I always say that getting a cool, neat title is kind of a showoff to actually get accepted. So to tell you a little bit about myself, I’m a systems engineer. I’ve been doing software engineering for some time. And then I moved to the dark side, so now I’m actually doing marketing. So I’m this evil person that now tells you to do things and stuff like that. But something that is interesting is that right now I’m working at Auth0.
I’m gonna tell you, shortly, what Auth0 is about so that you get some of the examples that it is. But the main idea is that Auth0 is a SaaS for authentication, so if you don’t want to implement authentication, and you wanna get it done securely, with either username and password, social connections, or enterprise connections, you can use Auth0. And then a very small URL, which is we’re hiding people for DevRel both in Europe as well as in the U.S.
So if you want, go to that URL and we’d be happy to have you on our team. And something interesting for me, for this talk, is that being an engineer, I always say that I’m designed better to use the left part of my brain. That’s the part about analytics, number, math, etc. But for both content marketing as well as the DevRel, we actually had to use both parts of the brain. It’s also about using the right part.
The experiments begin
And to learn why, I basically need to start from the beginning, and I’m gonna tell you a little story about Auth0 and what has been going on with Auth0. I shouldn’t have picked this GIF because the light is killing me with the blue, green, something. Anyway, so when I started at Auth0, our signups were flat at 1,500. So we were having 1,500 signups every month, we weren’t growing, and we were like this.
Like we wanted to be a good company. We were building a good product, and we wanted people to actually see it. We wanted people to try it out, and we wanted people to say, “Hey, this is cool,” “I don’t like this,” or, “I do like this,” or something like that. So in order to start increasing this number of signups, what we said is, “Let’s try to do some experiments. I mean, I’m an engineer, so I’m working now in marketing. I don’t know marketing because I’m an engineer, so let’s try to do experiments, and based on that let’s try to grow.”
So as a background, what is an experiment? And the main idea, to me, is that an experiment is about getting a very big problem, decomposing it into smaller parts. For each of those parts, using both qualitative and quantitative information, create a hypothesis of something that you think you can do to change or to improve, then set up a metric, which is how you’re gonna measure success, set up a goal so that if you get to that goal, for that metric, you’re going to have a success…so we gonna succeed, sorry, or otherwise you won’t.
And then you set some timeframe so that you get try it out in certain timeframe because depending on the experiments, it will be different. And then you run the experiments, you see what’s going on, and you continue and iterate from there. Something interesting to me is that I don’t consider any experiment a failure. You’re gonna see, later on in the presentation, that we’ve tried a lot of experiments and by failing, as long as you actually go deep, try to understand, and try to learn why, that’s not a failure because then you can create a new experiment based on that.
To me, a failed experiment is an experiment in which you don’t check why it was okay, you don’t check why it worked, you don’t check why it didn’t work, so then why did you even try? So the first thing you need to do, to do experiments and for marketing, is understanding your target market. I’m an engineer, so for me, it was easy to understand myself. Well, sometimes sort of. So what do developers do? When do they do it? Why do they…are they doing it?
And for me, it was something like this. Like every time I wanted to do something and I didn’t know, Google is my best friend. I always say that for being a developer, I have to be very, very good at searching on Google. So I was searching modules on MPM, or answers to my questions on Stack Overflow, or blogposts, or GitHub, etc. And then what I was doing is this, I’m very good at doing this, there you go.
So and then if this didn’t work, you know what I was doing? Google it. Googling a lot of solution, trying to see if there’s another snippet of code that I can actually copy and paste, and then try it out again. So since I was doing a lot of this, I said, “Okay, what can we try? What can we start doing? Let’s start trying experiments to try bringing more people to know and realise what Auth0 is.” So the first experiment was Auth0-related blogposts. So every time we were shipping a new feature, we were doing a blogpost.
Or every time, now, that we found a cool use case for Auth0, we were also writing a blogpost. So for this, the idea was that the metric was gonna be the page views and the signups. So how many people see this content? How many people signing up after actually seeing this content? And the timeframe was approximately six weeks to get it to work, and then our plan was to publish, like, four or five articles.
What did we learn?
So we started with this and, yeah, it kind of didn’t work. So we were sad it didn’t work, but at the same time, as I said, in any experiment it can fail. It’s just about learning. So what did we learn? We increased page views 4%, so the number of page views, the number of people who were seeing the content, was actually going up. But we were still flat on signups. We weren’t growing on signups.
So what we did is we served or scuba dived into the data to see what was going on. And what we figured out is that the people that were coming to the blog were already Auth0 users, so we were actually never going to convert them to signups. But once that…some of these very little people that weren’t Auth0 users, they weren’t converting also because it was something related to Auth0. So what we started digging around is, actually, asking on Twitter, doing surveys, asking other people like, “What would you like to read? What would you like to see in the blog? What would interest you about the product?”
And that’s when we came with this second idea, which is, “Let’s do greenfield content. So instead of teaching people how to use Auth0, let’s actually teach them useful stuff. Let’s actually teach them how to authentication to react application, or how to authentication to an Angular application, or how to use Chase and with tokens in a Rovi API or in OJS API.” So the idea here was that we’re going to write a big article, a big piece of content on something that’s useful to developers, and then at a very smaller aside saying, “Hey, if you don’t wanna do it yourself, you can try Auth0. But just try the product. It’s free. If you like it, you continue using it. If not, that’s okay, you can ditch it.”
So for doing that, we went into several tools to start learning what should we be writing about. One of them was Google Trends. So Google Trends is a very good tool to see what people are looking for, and at what that point in time. So this is the curve for React JS, and as you can see it’s growing. And I don’t know why, but in this stage, it is. So React JS is probably a good…an interesting topic to actually be talking about.
Another tool we use is GitHub Trending. So GitHub Trending you can see by language and by timeframe, so month, day, week, etc. You can see what repositories are popular and are being stored or forked. So based on this we were looking at what cool stuff were people doing, and then if something was interesting, but it didn’t get still a lot of attention, or it didn’t get attention but there wasn’t a tutorial, we said, “Let’s actually write. Let’s actually do something about this.”
But other than that, all of the developers are still on Twitter. And even though I feel I don’t have time now to read the timeline, something that we were using is a tool that’s called Nuccel. I think that’s the way, my accent sucks. But it’s N-U-C-C-E-L, and what that tool does is it gives you a daily summary of all of the links that were fav’d and re-Tweeted the most by the people you follow. So if I were following cool developers, probably the links that I get daily are gonna be cool links, and we can learn from that what type of content, or what type of things we want to start creating based on that.
Content, context and communication
So based on all of this, we started to choose content and we said, “Again, we’re gonna do page views and signups as a metric. We’re gonna try it out for a few weeks, and we’re gonna see what happens.” It didn’t work. Again, we were like, “Why doesn’t this work?” So we didn’t increase that much the page views. That was the main thing that we see. There wasn’t that many people, actually, reading our content, but our conversion rate to sign up, so that means that the people who did see the content, were actually signing up.
So that means that these new posts were getting people to sign up and try the platform. The only problem we have is that nobody was actually seeing what we were writing. So, again, thanks to these failed experiments, we were able to understand what happened, and we created a new experiment, which was based on this previous one which is, “What if we actually make sure that actually, people see what we’re doing?”
So one of them was these SEO optimisations, so we’re working now on keywords, which got the keyword using the title, in the subtitle, which had context keywords etc., etc., etc. And the other thing that we did is to create a distribution list. So one of the problems we had, nobody was reading our content because nobody was finding it. It’s like this story where if a tree falls and nobody listens to it, does it make a sound or not? Well, for us, it doesn’t because we weren’t getting signups.
So the idea here is, “Let’s create a distribution list. Let’s search and look for places where we can actually now share this content.” So, for example, that could be Extra Credits, that could be Hacker News, it could be Facebook groups, it could be LinkedIn groups. I actually have one example in here, two examples. One is the actual SEO checklist that we use at Auth0, which is, for example, like, “Is the keyword there? Is the title less than this amount of characters? Is the description…?”
And some of the things seem, like, weird, like, “Why does a title have to be under…?” I think it was 56…yes…and less, like, at most 56 characters. And that’s because if you add more, Google will add a dot, dot, dot into your title. And if you do that, the CTRs, so the number of people who click on that link, lowers a lot. So, basically, we were following all of this, and then once we finished the content, we have the distribution flow. And this is, basically, depending on the type of content, where do we share it?
Like what’s at Reddit, what LinkedIn groups, what Facebook groups and what other places should we be sharing the content, depending on the content type that we’re using? So this was the new experiment that we tried, and, yeah, unicorns, it actually worked. This was the first experiment that we did that worked. What happened here is that we actually increased the page views 10%, and we increased monthly signups 7%. So we were, finally, not at 1,500 signups anymore.
So we were about to pop or open a champagne. We were about to drink wine, drink beer, I know, do everything, but then this happened. I was talking to the CEO, telling him, like, “Hey, it’s working. This is amazing. Blah, blah, blah.” And he told me this interesting phrase, which is that, “It’s a business. It’s a company. And the company wants to have a meaningful exit, and for a company to have a meaningful exit, it has to grow approximately 10% the revenue every month.”
And to grow 10% the revenue every month, how many signups do you think you have to grow? A lot more than 30. A lot…sorry, a lot more than 10. So seven? More than 10. We were sad because that meant we needed to work more. So, in this case, we did three experiments, one of them worked, the previous ones didn’t, so we said, “Why don’t we try doing the same? Why don’t we try doing more experiments?”
If Sherlock Holmes could blog
So how can we enhance these? And now it’s Sherlock Holmes time, and it was, like, “Let’s go, let’s check the data. Let’s see what other things we can do.” And, like, we said, “Okay, now we got it.” And this is an interesting stuff, and this a graph made by our actual data from the blog. And what this is saying is that if a person sees one article, they have a 0.1% conversion rate to sign up. So that means that every 1,000 people that sees a blogpost, one sign up. But if they see two blogposts, 1.2 conversion rate.
And if they see 7 blogposts, it’s 9.8. So for every person that sees seven blogposts, 10 out of 100 actually convert to a signup. So based on these data, which was our data for Auth0 itself, we said, “Okay, what experiments can we do with these to actually explore increasing the conversion?” And what we did when we were analysing the data like this, is we were actually checking a lot of different variables, and plotting them in different cases to see if the conversion rate was pure.
And this was a clear example of that happening. So our ideas were several. These are some of them. So one of them was blogpost series, so we want people to come back. We want people to read more blogposts, so what if we, instead of writing one blogpost, we do a React series and we do seven blogposts where you expect every Thursday there’s gonna be a new blogpost, and a blogpost is gonna total out, another thing about React, but we’re getting people to come back to the blog.
In that case, for example, it was actually helping with conversions. Another example was in newsletters. So we added an email newsletter that people could subscribe, and then they could read the content. So that was actually kind of a failed experiment. We’re still doing it, but we have very, very few people that were actually registering, and we didn’t know why. So when you want to learn why, it’s time to ask questions. So we started asking, like, “What’s going on?”
And people were saying, “I hate email. It sucks.” And I agree because if you send me an email, I’ll take one week to answer it because I hate email. So based on that, we said, “Okay, how can we let people know of new things without email?” And that’s where blog push notifications happen. Now the browsers like Safari, Chrome, Firefox, I don’t know Internet Explorer, sorry, but all of them have push notifications. So now people register to them, they don’t have to enter their email, and they get now a push notification of every piece content that we write.
And then, furthermore, for every article that you read, we also recommend you what other articles you can read based on that. So these were all very simple idea experiments, but we tried, and as I said, newsletter, for example, failed, all of the three actually worked. And thanks to this, we started to grow more. So now we started to increase monthly signups 15%, and we increased our page views 10%. So, not good enough. It’s never good enough, right? It’s capitalism.
Asking the good questions
So how can we increase it? How can we be Sherlock Holmes and start thinking about it? And we had no clue. We didn’t know what to do but, again, capitalism, we needed to do it. So we started asking questions. I think that the best way to find a good answer is asking good questions. It’s sitting down and asking the questions that you need, to find out how you can do this. So one of our questions here is, “How can we jump on developer trends?”
So, for example, Angular was a trend, but then React appeared, and, well, it’s not that much of a trend anymore. Now, React is more of a trend, for example. Then Vue.js appeared, and now Vue.js is kind of starting to be a trend. So how can we jump to these trends, those are examples of frontend only, that are existing and how we jump that wave? Or how can we make people aware of the stuff that we write? How can we make the people who are Vue.js developers, React developers actually know of what we’re writing?
Because the main thing is that our content, it was proven, it was cool, and it was converting. We just needed more and more eyes to see it. So, sorry, this is a shady GIF. But the main idea here is that what we wanted to do is we didn’t know. So we couldn’t use this left part of the brain, this analytical, mathematical thing that we were doing. We actually had to use the right part of the brain, because what we ended up figuring out is that maybe it’s about people. What if it’s about people? What if it’s not about numbers? What if it’s not about this hypothesis?
What about if it’s not like this experiment? So thinking more about it, and thinking more about people, we started to see these things, like, first we started to attend framework conferences, like, Angular conference, React conference etc., and we were trying to talk about authentication and JSON Web Tokens. We would never talk about Auth0, because every time I go to a conference, and I listen about a vendor talk, I hate it. Like, I just leave the room. I don’t wanna see it.
And as a developer, something that’s crazy for me is, like, I go buy clothes and, like, the guys…somebody comes, like, he says, like, “Hey, do you need help?” “No, go away.” So that’s me, and what I want is to research on myself and learn. So what we were trying to do here is teach people. We were trying to teach people what is authentication? What are the OAuth URL or OpenID connect protocols? What is JSON Web Token? And we were trying to add value to people. We actually have an event calendar, that I forgot to open, and we have Wi-Fi, it’s gonna be hard.
Becoming a people person
But the main idea we were having is, “Which are the events that we wanna go by which date? Who can supply? Who needs to apply? Why are we doing that?” And the main thing here, and this is something that I truly, truly think and value, is that you always have to be honest and genuine, and add value to people. That, to me, is something that is very, very, very important because if you’re not genuine, people will realise. When you come to me and you’re trying to sell me something that you don’t believe in, or when you’re trying to tell me something you don’t believe in, that’s never gonna work.
I actually have a good story for this. I was once at ng-conf, it’s an Angular conference in Utah, and I gave a talk, and then I was talking to a guy, and he was asking me, “Hey, I have, like, a pet project, and I actually wanna add Facebook and Twitter. Would you recommend me adding Auth0?” And, of course, what everybody will say, if I work at Auth0, “Yes, Auth0 is great for everything, you have to use it.” And the answer was actually the opposite. It’s, like, “If you only need Facebook and Twitter, and you only need to integrate those SDKs, don’t use Auth0.”
It’s actually, like, just use your own two SDKs. I can actually help you out, and I helped him out to do that, and Auth0 is cool if you wanna have your user store secure, or if you wanna have enterprise providers, but not really just for this. So based…the guy went away, and then two years later he came back. He actually sent me an email and he said, like, “Hey, I’m working at X company, and now we’re actually selling to other companies, and we need to use enterprise providers. And I remember you helped me out, and you told me that for that Auth0 was useful.”
And we actually got a 300K deal out of that by just adding value, being genuine and being honest because that’s what we were trying to do. And if your product is good, you just need people to get to try it out. I always say that our job at marketing is education and also, partly, to get people to try out the platform because if the platform is good, you’re gonna love it, you’re gonna want to use it. And that’s the way it should be. I shouldn’t be selling you…there’s an expression in Spanish where, like, Columbus came and he gave the Indians, like, very nice mirrors in exchange for all of the land. So that’s not what you’re supposed to be doing.
And something that was interesting here is that we actually started to be followed and read by devs that spoke and attended these conferences. So, to me again, the conference is about being genuine and connecting and knowing people. Usually, this time I am. But, usually, I sometimes don’t even go to some of the talks because I can see the talks of lying. To me, a conference is about meeting people, and something that’s interesting is that most conferences have what’s called, “A speaker dinner.”
And imagine who are the speakers of an Angular conference? Well, they’re the creators of Angular or the creator of open source projects that have to do with Angular. And if you genuinely care about them, and knowing other people, because to me that’s DevRel. It’s genuine interest in other people, other cultures and learning new things, you can build relationships with them. And then when you actually start writing good content, and you share it with them, like, “Hey, I just wrote this, what do you think?” They’re actually gonna start sharing your content because it’s cool, and because you built a real relationship with them.
So something that started happening, now, is that the Angular developers were reading more of our content because they were following Brad Green, or all of the creators of Angular, and that was actually pretty neat. So now they were starting to read and follow all of our content, and we were increasing that. And also, since we weren’t talking about Auth0, we were talking about general stuff on authentication, JSON Web Tokens, we actually started to be the go-to questions for authentication.
And now, because our site started to get popular, and because more people were reading it, more people were seeing it, devs wanted to write content for our site. So we created “The Guest Author Program.” And the main idea of this is, if you write content, you can write it at Auth0, you can get it exposed to more than, like, millions peoples because that’s our page views, and we actually pay you for that. We pay you 200 bucks because we did the math of how good a good article is for us, how much revenue it can bring, and we can actually pay up to $200. So that’s what we’re doing now.
And all of these was what got us to, actually, well, even more on traffic, and even more on signups. So going back to the present, and going faster because I’m running out of time, this is what…this is the page that shows, like, what is Auth0, and our ranking, for example, in the U.S. is 2000 sites, and globally it’s 6000. If you think we’re about technology, that’s pretty high gap for being a tech site. Imagine that Facebook is one, and Google are…Facebook and Google are one and second, and they’re seen by everyone.
Always be genuine
So what’s next for us? Going very first. One of them is called the Auth0 Ambassador program. So you probably heard about Google’s GDE, or Microsoft MVP, this is very similar, but the value of being a Google Developer Expert is that they have the Google brand. Auth0 is not as powerful as that, so what value do we give to people? And the main value we actually give to people here is teaching them. So what we’re doing is we’re finding developers that have spoken at Meetups, who want to start speaking at conferences, and all of us actually give them a training on how to write a CFP. We had them practice. We had them write the talks.
And if they get accepted to a talk, as long as they wear an Auth0 T-shirt and they mention we brought them there, we pay their flight tickets and the hotel so they can actually speak at the conference. And this is, again, something that I love because it’s being genuine, we’re being honest, we’re helping developers, and that’s basically what we’re doing here. And it’s good for the company because we’re getting a lot of people exposed to Auth0, it’s good for people because we’re teaching them how to talk, and it’s also good for everyone.
So this program is running right now. And to give you an idea, in November we had 30 events. November has 30 days, so it means that we had approximately one event per day thanks to this program. And now finishing up. If there’s one thing that I want you to remember about this talk is that content distribution is always about good quality content, but it’s also about people. Never forget that. Thank you.
Aja Hammerly shares how Google use friction logs to document where a developer-targeted product is hard to use.