Many of new businesses are relying on the wealth of data available to them to find new markets and develop targeted products. As data scientists, it is important to understand how startups are innovating ways to utilize data and how we can add to their current systems to improve them . Entrepreneur.com calls the current trend of new businesses, the “golden age of startups”. In other words, they’re going to need a lot of processing power from data scientists to succeed.
On April 12, 2017, Galvanize hosted a meetup for people who wanted to learn about big data in startups. Most of us new to data science including myself probably are not starting our own company… yet. However, that does not make going to a meetup about startups irrelevant to us. Meetups about topics that may be outside of your knowledge provide valuable information about the industry you are getting into.
Below I have outlined the key points that came up during the panel and who I plan to use them at the end of this article.
For more information on the panelists click on the link.
Q: When should you start taking your data strategy into account when building a start-up?
Conclusion: “From the very beginning. You can’t retroactively collect data. You need to figure out: What are the use cases? What do we want to do with the data? What kind of modeling we want?” says Gigi Trencher, the Director of Business Development & Strategic Partnerships at Drawbridge Inc. Julia Macalaster, Head of Strategy & Growth at Def Method, adds, “but for startups it generally doesn’t happen until we had to think of integrations for different software platforms we were using.”
Takeaway: In this case I reframed these questions for myself. When going for an interview with a company, think about what are the use cases for their particular business? What would my client want to do with the data? What kind of modeling would they want? If I do not know these questions, they could make for great follow-up questions after the interview. To Julia’s point, researching platform integrations that the company is planning in the future to help you sound experienced.
Q: What are some key KPIs?
Conclusion: You have to know your users so that you can align your metrics accordingly. It varies across industries and stages that the company it is in. For young companies, for example, you might be more interested in daily, weekly, and monthly users. More developed, sales-focused companies might be looking at customer acquisition cost, which is the initial cost of getting a person to purchase your product for the first time. For companies where services are the product, a good KPI is the mean time to resolve an incident. For much more advanced companies, lifetime value and cost benefit ratios are important.
Takeaway: I saw this a call to action to do some research about KPIs and their applications in different scenarios. Good points to ask myself when suggesting which KPIs to use are: What is the size of the company? How well-developed are they? Who are their users? In case you don’t know, KPI stands for Key Performance Indicator.
Q: What is the advantage of startups?
Conclusion: Trencher says, “The big advantage is that you can scale your models. At bigger start-ups you have to work with what’s given. Startups are lean and mean and they can adjust their pipeline as needed.” It is much easier to step back and look at the whole organization in its entirety, which is much harder to do in large organizations. With smaller companies, it is also sometimes easier to just reboot the whole system and start over without much negative impact
Takeaway: When working at a small startup it is much easier to implement and scrap ideas if they do not work. Larger companies have to work with a much more rigid and often times older data infrastructure that makes larger scale changes most more risky and costly. I see this as a launch pad for learning about the pain points of implementing data models in larger companies by reading through some case studies.
I spoke to Sally Choi, a Senior at New York University, studying Business and Technology Management with a minor in Psychology. She and her friend Jamie Cheng are beginning to attend meetups to learn more about the world of data science. They said that their goal was to attend one a day. When I asked Sally what she hopes to get out of attending meetups, she said, “Even if I don’t know about 15% of it, you get to learn all about the tech used in companies.” I agree, since knowing some of the big software names helps you familiarize yourself with the product before going into an interview.
As Paul Burt, Developer Evangelist at CoreOS, summarized, “Data is just chaos and noise. Information is data that has context and a story attached to it.” As new data scientists, it is important for us to collect the key points at these meetups and then turn it into information that we can use because we will not always be the intended audience.
Below is the link to watch the full panel: