The Artists of Data Science

The ONLY self-development podcast for Data Scientists

Take a Leap of Faith | Alistair Croll

On this episode of The Artists of Data Science, we get a chance to hear from Alistair Croll, a well-established entrepreneur, analyst, and author. He is known for writing Lean Analytics, and for being one of the founders of Coradiant, Year One Labs, and the Strata conference.

He shares some excellent tips one how to ask the right questions when working with data, how to communicate with customers, and the need to be obsessed as an entrepreneur.

Alistair touches on some amazing tips that anyone can use to catapult their success. It was an amazing honor to interview him!

Some notable segments from the show

[11:11] How privacy concerns are being addressed related to data science

[13:39] Incorporating philosophy into data

[14:22] How to compete as an early stage company

[18:30] How unwavering faith can instill the entrepreneurial mindset

[22:35] Identifying the various types of innovation, and their impacts in the organization

[36:56] Music science, and how the digital era has affected the consumption of music

[46:04] The formula for informing and engaging people the right way

[51:38] How to profit off of attention

Alistair's Journey

Alistair’s journey into data science can be traced back into his childhood, when he first began toying with his Apple 2. Then, as a college student, he took a statistics class, which he did well in. This gave Alistair a good stats background, which he relied on when he started his own company, called Coradiant.

Coradiant focused on developing user monitoring products. These products allowed clients to focus on how users navigated websites.

Being in this space, Alistair eventually had the opportunity to write books on data science topics, such as “Complete Web Monitoring” and “Lean Analytics”.

[2:37] I can go back pretty far. I... As a kid, I had an Apple 2 with a 300 baud modem. So I had to, like, spend my summers figuring out how to get that to do things. And then in university, I had a feud with the dean of the business school. It's a long story I won't bore you with. That's to do with student fraud and all kinds of stuff. And we kind of uncovered some stuff that was going on with the University.

And we were running the student council and that caused a lot of discomfort for the dean of the business school. Turns out he was also my stats teacher. And so I had to do really well because if I was going to make any mistakes, he was going to fail me. Right?

So I was like I actually had to open up the books and work hard. And so I got some good stats background there. And my parents are both scientists. So I grew up with the scientific method and thinking about, you know, biases and how to how to understand things properly. And then a few years later, I started this company with some friends called Coradiant. And Coradiant was - It was really user monitoring. So Web analytics shows you what people do on your Web site, but it doesn't show you if they could do it. So, like, maybe the person didn't buy it because the page took forty, forty seconds to load. Maybe the person didn't buy it because they got a five hundred error and there was no JavaScript to tell you that. Right? And so with this product was called TrueSite. It was part of what we call real user monitoring products. But in order to sell it, we started out selling to the technologists - they didn't have the budget or the sort of authority within the organization. So we wound up having to sell to marketers. And so that forced me to get into, you know, speaking analytics to them, even though I was like a networking head. And then we kind of moved from Web analytics - I wrote a book with a a guy named Sean Power called Complete Web Monitoring that got into - like, how do you measure social profiles? And all this other stuff that requires a lot of big data. And then O'Reilly was putting out a book series called Lean - called it based on a lean startup series.

And they asked my co-author Ben Yoskowitz and I to write a book. So we wrote Lean Analytics. Lean - The Lean Startup is an amazing book. It's a book that's launched thousands of ships, that's proverbially speaking, but it is very aspirational.

It's not specific. And we're more like Bob Ross like paint a happy tree, you know, like very boring. Here's the prescriptive stuff. If you're here, do this until it gets this. And so I think that's one of the things that that helped the book catch on. Nobody's more surprised at how far it's gone. I got a mail from someone whose taking university in Madagascar who's like, this is my textbook. But I think that if you are trying to understand the modern world and you're not thinking critically about data and statistics, you are probably being tricked or taken advantage of almost every day. I mean, we need data literacy to survive the modern Internet. And so part of it was business through analytics and web. But a lot of it was just figuring out how navigate today's information heavy world.

Where is the field headed in 2-5 years?

For Alistair, the early days of data science involved the process of ETL (extract, transform, load). But now, he believes that the data science community is starting to see a lot of democratization of those tools. He is starting to see tools that help non-technical people experiment with models.

He also believes that we are going to have the ability to attach models to automated systems that produce new models. This will essentially create an environment where data science can be used to correct and update the system itself. The need for a human to interpret the data will be the exception, not the rule. Instead, data scientists will make sure that the system is still aligned with business goals.

[6:01] I think one of the dirty secrets of data science, at least in the early days, was that 80 percent of what people call data science was just ETL. Just cleaning up data and moving it from one place to another, you know, moving stuff between buckets. We're starting to see a lot of democratization of those tools. And so you're starting to see things like datarobot that will let a non-technical person, experiment with models and kick the tires and so on. But I think one of the things that is going to happen is, when things start out, we use them tentatively. In the early days of cloud computing, you know, that was fine for QA. It was fine for like putting a dev build on there. But you didn't really do it in production.

What will separate great data scientists from the rest of them?

In Alistair’s view, really great data scientists can arrive at a working model much more quickly than their peers. This will be partly due to their intuition, and also due to the role of managing the model rather than building it. The process of anticipating what will make something happen sooner is going to be the mark of a great data scientist.

Alistair also discusses that data scientists who build ethics and trust into their models are going to do really well. In the future, everything's gonna be data driven. Everything's gonna be running on a model. The companies we trust will be the ones that don't squander the trust we've given them.

[8:33] If you'd look at an exponential curve, right, what matters in the exponential curve is the starting number. Right? Because the sooner you start, the better. And the slope of the curve. I think you're going to see that really great data scientists can narrow - it can arrive at a working model much more quickly. Partly through their intuition, and that they will transition from building a model to managing the model. Which is actually a different set of skills, correcting for drift, finding out what could go wrong, are the factors still there, and so on. And I think they're also much more focus on like fast start learning. So instead of needing, you know, millions of compute hours to actually generate a good model, you're going to have data scientists go, oh, I know the model. That's going to be very close to what I need ahead of time.

Key takeaways from the episode

[13:39] The skills that a data scientist should have is the ability to incorporate philosophy into data. Asking questions like “What should the user know?”, or “How could this be misused?” are important. They are what make data science so interesting.

Important soft skills

[45:16] You've got to know what you're willing to fight for, and what you’re willing to compromise on.
Try and understand the customer journey. Ask yourself, “What are the steps that users go through? How do I use those steps to make sure that there's a satisfying experience?”

How to be an intrapreneur

[28:28] For a data scientist to become an intrapreneur, they have to transition to the unknown unknowns. Ask yourself, “What can the data tell me that I don’t know?” and go look in the data for patterns.

How to be an entrepreneur

[19:11] You have to take a leap of faith that this idea that you're absolutely obsessed with is right in the face of criticism. You need incredible humility to learn what other people think and adjust your perception, and then you also have to have this insanely high level of faith to keep you going and believe that you're absolutely right.

Memorable quotes

[14:22] …”as an early stage company, your focus is your biggest currency.”
[22:10] …”crises have a way of accelerating the inevitable.”
[46:04] “...you got to first seek to engage and entertain and then you have the ability to inform people.”
[51:38] …”find a way to capture attention that you can turn into profitable demand better than the competition.”

The one thing that Alistair wants you to learn from his story

[53:26] All my life I have seen people solve for lots of things, such as fame, or profit. All the good things that have happened to me come from “solving for interesting”. Find a way to solve for what's interesting about your company, about your life, about your hobbies, and you will thrive in an attention economy.

From the lightning round

Best advice Alistair's ever recieved

People do things because they want to get laid (perceived as attractive), made (“Made man in the mob”; powerful), paid, or unafraid (reduce risk).

Advice that Alistair would give to his 20 year-old self

Find out how to cultivate a personality that's public and fairly strong. Exercise more. Get a sense of style. Be yourself, and don’t be afraid to be a little larger than life, as long as you aren’t a jerk.

The number one book that Alistair recommendeds you to read

“The Righteous Mind” by Jonathan Haidt

Books and other media mentioned in this episode

“Complete Web Monitoring” by Alistair Croll and Sean Power
“Lean Analytics” by Benjamin Yoskovitz and Alistair Croll
“Propose, Prepare, Present: How to Become a Successful, Effective, and Popular Speaker at Industry Conferences” by Alistair Croll
“Just Evil Enough” a book Alistair is currently working on
“Lean Startup” by Eric Ries

How you can connect with Alistair

Connect with Alistair on LinkedIn
On Twitter
Solve for Interesting

Article Comments