Speaker Spotlight: Using Data to Build a Better Business Faster
Big data is changing the way everything from government organizations to international record labels operate. We can access sophisticated analytics about our businesses from our laptops, making it easier to understand our customers at a granular level — in theory, that is. But for startup and enterprise companies alike, discerning the exact metrics to evaluate, and knowing when it’s time to talk to users face-to-face, can be confusing. That’s where our good friend Alistair Croll comes in.
The Lean Analytics author and Solve for Interesting founder is one of our featured speakers from The Lean Startup Conference in San Francisco. To offer a taste of Alistair’s expertise, we offer an edited interview with this esteemed entrepreneur and author — who knows many important things about web performance, analytics, cloud computing, and business strategy — around best practices for using big data.
Alistair ran Year One Labs using a tailored methodology that put customer discovery first.
“We ran Year One Labs as an incubator more than an accelerator. We chose startups through hackathons that showed they could build new things, rather than for a specific idea. We knew we needed a methodology that would force them to do a lot of customer discovery — in fact, we didn’t let them code for the first few weeks! We recognized that at the start of a startup’s life, it doesn’t know what problem it’s trying to solve, much less how to solve it. Lean Startup fit the bill perfectly.”
Data collection should be built into startups’ DNA.
“Analytics need to be in the entire lifecycle of a product. You have to build data collection into the product roadmap; measure it in real time and in the aggregate; and learn by following up on the data to better understand your market. We often say that in a Lean Startup you aren’t building a product — you’re building a product to figure out what product to build.”
So here are three examples from Alistair of companies that successfully used Lean Startup and Lean Analytics methods to determine the best metrics for their operations.
“Localmind, a Year One Labs company later acquired by AirBnB, had a business around asking questions about a place. One of their big assumptions was that strangers would answer questions. Because we wouldn’t let them code, and because this was a big, risky assumption, they decided to go on Twitter and geofence all tweets coming from Times Square and ask those people questions.
“They figured if you can get an answer from New York, you can get one from anywhere. And it turned out their assumption was valid — over 95% of people would respond to a question like ‘Is there WiFi?’ or ‘Where’s good coffee?’ That was enough to move ahead, in a day, without coding. It’s a great example of minimal data.
“We’ve seen big companies do this too. One global maker of paper products (that I can’t name) noticed a discrepancy between what men and women were searching for on their site, and used this to introduce an entirely new class of personal hygiene product that’s worth millions today. It turns out that unfulfilled searches on a website are an excellent open survey of what customers want you to sell them, but you don’t currently offer.
“Finally, consider David Boyle, who was then the head of analytics at EMI. The label wouldn’t let him see the huge hoard of transactions they had on record, so he decided to grab data himself instead, and ran over a million surveys worldwide about how people consume music. When he brought that data back to the company and its artists, it went well. Here’s a talk of David explaining how the BBC uses analytics today.
“He recently told me that when he showed [EMI artist] Snoop Lion that he had a burgeoning audience of ‘gadgeteers,’ or tech-friendly listeners, Snoop changed his entire attitude to the online world and social media. With these changes underway, David finally got access to the hoard of transactions, letting him to even better analysis.”
The data your organization generates could be its biggest asset.
“Computers today can surface unexpected correlations that marketers can then test for causality. The entire field of growth hacking is really about finding a thing you can change today which causes a change in some desirable metric down the road — and then optimizing that metric.
“But it’s getting even smarter. There are a number of systems that can outperform data scientists on the ‘intuition’ part of analyzing data, deciding what metrics matter automatically. That’s transformative. You only need to look at Tesla’s self-driving cars to see what happens when you can push machine learning and data into a hardware product to see how different the world will be.
“Data is also changing regulations and governance. Essentially, it’s squeezing arbitrage out of industries, because everyone has the same information at the same time. To combat this, many online services are guarding their data jealously, or erecting garden walls around their users, making it harder for upstarts to break in. But that also means that if you can create a valuable data set, your best asset may be the data you generate.
“Consider Shazam, which can tell you what track is playing in seconds on a mobile device. Their service is free — but their data can predict what song will be the next big hit, and that’s invaluable for the music industry, which is where they make their money.”