Sean Byrnes is CEO and co-founder of Outlier, he is the leading authority on automated business analysis platforms and outlier data identification. Before creating Outlier, Sean founded Flurry, a highly successful mobile-analytics and advertising platform acquired by Yahoo in 2014.
I have always enjoyed building things, and in college I majored in engineering since I thought that was the best way to spend my life building. I started taking courses in computer science and slowly realized that the most interesting things being built were not physical, but computer systems that were going to change the world. I took a course in machine learning (ML) and was blown away to learn that you can write programs that will learn! Just like people! On the last day of that class, I walked up to the professor and told her that I wanted to spend my life working on machine learning.
That led me to graduate school in machine learning, but at the time of my graduation the industry around ML was still too primitive to make it a career. I started working for software companies, then started software companies myself. In 2015 when I was starting Outlier, the cloud technologies had finally made ML a viable platform to focus on as a founder and I was able to make my dream come true!
In 2014, you founded Flurry, Inc. which was later acquired by Yahoo. What was your mindset at the time of launching Flurry? Did you ever expect it would become such a wild success?
The initial idea behind starting Flurry in 2005 was solving the simple problem of easy access to our email. This was back when a cell phone did very little, it made phone calls and maybe sent text messages. Flurry was started to solve this problem of connectivity, to bring email and news, to the devices we all had in our pockets. Blackberry had proven that mobile messaging was a hit with professionals, so we tried to bring that experience to all consumers.
Flurry was a successful app developer, but we saw a bigger opportunity in opening up our internal analytics platform to other app developers in 2008. At the time, both the iPhone and Android had launched, but neither supported apps yet. We had seen the transformative power of mobile apps through our own Flurry apps, so we believed that a bigger revolution was coming even though Apple had announced that the iPhone would never run apps. At a time when everyone was investing in the mobile web, we decided to go all-in on mobile apps by providing a platform for analytics designed exclusively for mobile apps.
It turned out to be good timing. As Apple and Google made the shift to launching app stores, Flurry was at ground zero of the app revolution with a platform built for mobile apps refined over three years of using it ourselves. Starting at that point in the fall of 2008, Flurry as a business doubled every six months for the next six years.
In 2015, you launched Outlier AI. Could you share the genesis story behind this company?
In 2014, we sold Flurry to Yahoo after a nine-year journey. When it was acquired, we had about 500,000 analytics customers around the world, and I had been travelling and meeting with as many of them as I could. Everywhere I went, every company would ask me the same thing: “Sean, I love this data that you’re giving me. But how do I know what to look for in all this data?” Every company in every location and every vertical was struggling with finding value in their data.
When I started thinking about it, I realized that we had reached a new era of business intelligence. All of our business intelligence tools were designed to answer questions we knew to ask, but the problem was that we didn’t even know what questions to ask! We needed a new generation of products that would find those questions and bring them to us, automatically.
That vision became automated business analysis: the ability for software to uncover emerging problems and opportunities automatically, using AI.
Knowing what question to ask is the biggest problem that enterprises often have. Can you explain how Outlier AI solves this?
Outlier finds insights hiding in your data automatically. You spend a few minutes connecting Outlier to all the different places your data live (databases, cloud tools, etc.), and in return you get a feed of questions about how your customer behavior is changing, how demographics are shifting and how your business operations are changing. These insights are generated automatically, but they have everything you would expect from a human analyst writing a report, including natural-language generation and root-cause analysis. It doesn’t require configuration, training or implementation to get started since it relies entirely on online learning systems.
As a result, Outlier can be deployed extremely quick and is used by typical business users who know nothing about machine learning. They simply connect Outlier to their data and wake up every morning with a set of 3-to-5 insights around things they should be asking about their business. It feels like magic, but it’s just a lot of math.
The most exciting part for us is that Outlier is not only the first ML product our customers have used, in many cases it’s the first business intelligence tool they have used. Traditional business intelligence tools are so hard to set up and configure that they are intimidating to most users. Outlier is so easy it can be up and running with no prior experience.
What machine learning technologies does Outlier AI use?
The Outlier platform uses dozens of machine learning techniques as part of our pipeline, ranging from time-series modeling to unsupervised clustering and recommender systems. In fact, every insight produced by the Outlier system is the result of at least a dozen machine learning systems working together! That’s important, since business users demand a very high level of quality and fidelity, and no single ML technology can produce that quality on its own. Orchestrating that enormously large number of ML systems is extremely difficult, but when you see the results it’s magical.
In fact, there are only a few ML technologies we do not use, such as deep learning (neural networks). A critical part of the Outlier product is explainability, so that the user knows exactly how Outlier found a given insight. This is critical in building trust with users who know nothing of machine learning and need you to explain things in straightforward terms. There is some great work going on to bring explainability to approaches like deep learning, and we might add them as that develops.
You are also an angel investor. What types of companies or entrepreneurs do you invest in?
I invest in companies and entrepreneurs who are solving problems. A problem is the difference between what a person wants or needs and what they can get today. The most important thing I look for is the potential of the problem you are solving and how large the market is that has that problem. You cannot produce a 20x return on $10M of investment if your market size is only $50M, but you can if your market size is $1B. If I can envision your company in 10 years operating at that scale, you will get my attention. That involves doing the following and doing them well:
- Present a plan. Your plan might change, but you need to have a credible long-term plan for getting to the large outcome. If you can’t build a credible plan at the beginning, it’s unlikely you will be able to come up with a new one as the market changes. The plan you present will also serve to identify the key risk factors that your business will face as it grows.
- Show you mean it. You have already been following your plan in building your business, so show off how well you have executed it. Remember, your progress so far is not why they will invest in you, but your progress so far is proof that your plan is credible and that you can execute against plans you create.
- Sell the team. Ten years is a long time. If your company is going to be very successful, it will be a long and difficult road. Your team is critical because it is those people who will steer the company through those hard times and the venture investor needs faith that you can do it. In the end they are investing in you.
- Play to win. If you really have found a big opportunity that can produce 20x returns, it is likely that many others have as well. You need to show a distinct competitive advantage that will allow you to win when faced with dozens of competitors going after the same goal.
In a recent talk you spoke about how entrepreneurs need to sell a new category product, or tell the customer how to use it, but you can’t do both. Can you briefly explain this comment and elaborate on how important this mindset is to becoming successful?
New categories of product are new, so it won’t be part of someone’s daily routine. People are busy, and as a result most of their time is already claimed by products they are already using. As a new category you need to earn time in someone’s routine, and you do that by fitting into their life the way they want you to fit. You can’t tell them how to use it, since that would mean telling them how you fit in, you need to be flexible. Over time, after you prove yourself, you can become part of their daily habit and then you can start telling them how to use the product.
If you are just building a replacement for an existing product, and not a new category, then there is already time in the person’s day set aside for that type of product. In those cases, you want to tell them how to use it to ensure they replace the other products with yours, so that you claim the time before it slips away.
You have spent over 15 years building award winning user interfaces for both consumer and enterprise companies. What differentiates a great interface from the competition?
A great user experience is something that you can pick up and start using immediately, without requiring training or tutorial videos. Too many products focus on what is possible once the user becomes an expert, but most users give up long before becoming experts, so all of those advanced features are never used. The most important skill a product designer can develop is to put themselves in the position of a first-time user over and over again, seeing the product for the first time and identifying all of the confusion and friction that exists.
Another important aspect of a great user interface is what I call “visible differentiation”. In the first 30 seconds of using your product, a person should know exactly what makes it different than anything else. If it takes longer than 30 seconds, or if you have to explain it to them, they will give up and move on long before they see it. People are awash in products today, so you have to move extremely fast to claim their attention and have them understand what makes you stand apart. This is true for everything from consumer products to enterprise software.
When you think about competitive advantages, most people think about complex technologies, but great user experiences are some of the best competitive advantages. It’s very difficult to copy a product with a seamless and magical user experience since it involves so much nuance and detail.
You also advise companies. What types of companies should approach you?
I advise a wide range of companies, ranging from consumer apps to financial services to enterprise software startups. The only criteria I have is that the founders really want to learn and are open to both positive and negative feedback. The most valuable advisers I’ve had in my career cut through all of the noise and gave me their honest and critical assessment, even if it was hard to hear. Many advisers just try to say positive things and build up the founders, and that is valuable as well, but I find that doesn’t help in the long term. I summarized the difference in a blog post (Coaches vs Cheerleaders) about the different kinds of supporters you need in your founder journey.
Thank you for the great interview, entrepreneurs should pay attention to these nuggets of wisdom. Readers who wish to learn more should visit Outlier.