Accelerating Startups with Data Science
Using algorithms to inform strategy
by Armen R. Kherlopian, Ph.D
As an accredited investor, senior executive advocating for leveraging startup innovations, science judge in the Columbia University Entrepreneurship Community, and co-founder of the BAJ Accelerator hosted at the Jacobs Technion-Cornell Institute, I have seen hundreds of startup pitches. During every single one of those pitches there is a question that is on the top of my mind:
Do the founders know how to gain value from data?
The ability of founders to find and understand patterns in data is key to identify risks their teams will face and learn about the markets they seek to serve. Early data provides validation of a startup’s business model, namely, confirmation on how value is actually generated and importantly how it’s captured. It’s not enough just to see metrics like user count or even revenue go up, but also to understand in a nuanced way what parts of a startup’s offering resonate with customers. Successful scaling depends on such insight, and this principle applies both to startups and the largest of enterprises.
How does a startup gain value from data?
Comparing the top of the Fortune 500 list now to the list in the year 2000, notable entrants are companies that have effectively leveraged technology to gain value from data. To do so requires algorithms, and an algorithm is nothing more than a recipe: a sequence of steps that can be executed. Let’s take ride-sharing as an example. When you call for a car by pressing a button on your phone, or even smartwatch, a few algorithms are immediately at play:
- Matching algorithm that helps to connect you to a nearby driver while balancing other customer requests coming in.
- Routing algorithm that assists your driver in finding you and taking you to where you want to go.
- Route-difficulty algorithm that scores how complex to navigate your requested route is.
Although that third algorithm on route difficulty is currently immaterial for customer experience it is actually vital to an upcoming competitive milestone in the market. As autonomous vehicles gain in capability, and as we enter a situation with hybrid manual and autonomous car fleets, ride sharing companies that are best able to score the difficulty of a route will have a competitive advantage. Namely, when you call for a car and if your route is scored by an algorithm as hard, a car with a driver will be dispatched. If your route is scored as easy, a car with no driver could be dispatched. Being able to increase the latter scenario algorithmically can lead to massive commercial lift. Thus, for a company to gain value from data leaders must strategically think through which algorithms to develop and deploy, and critically consider performance milestones of algorithms when setting strategy.
There are indeed all sorts of implications from leveraging data and algorithms, from user sentiment to that of regulatory compliance. Algorithms can even be composites of multiple algorithms leading to technical nuances around software development. The point I want to underscore is that the operational leverage of using algorithms to gain value from data can be tremendous. Furthermore, the fact that we now have algorithms that not only can run sequentially, but that can also learn from data in a dynamic way, is something that has been heralded as being as significant as the development of electricity. For a startup to electrify a market segment though requires the right team.
When should a startup hire a data scientist?
A question that I’m regularly asked by Fortune 500 CXOs (e.g. CIO, CTO, CMO, CEO, and heads of business units), as well as startups, is how to stand up a data science capability. The context is typically during planning of a highly strategic initiative, a systematic threat where people are inspired to rise to the occasion, or when gaining value from data is the difference between winning and losing. I entered data science before the term was in widespread use and have since co-authored a book on the topic, the Field Guide to Data Science. Over the years I’ve realized that data science thrives in data-driven cultures, where decision making includes data use systematically and not just intuition alone. This consideration is salient for startups as it relates to scaling safely and in particular the exact point for a startup to invest in algorithm expertise as per hiring data scientists. The key things to keep in mind when deciding on such critical hires, or co-founders, are what stage a startup is in as related to customer acquisition and what it takes to develop the core product. Let’s take a look at a few stellar startups as examples:
- Exteros makes smart spaces by leveraging computer vision. Here it would not be possible to build the core product without algorithm expertise. As a result, data science was vital at the earliest stages to enable key product functionality in addition to accelerating through securing initial customers and establishing a new product category.
- SuperAnnotate enables high-quality data labeling to improve machine learning model performance. If we consider artificial intelligence as a system taking action, we can consider machine learning as finding patterns to take action on, such as with classifying images in computer vision. As key stakeholders of data labeling include data scientists on the customer side, having more data scientists internally has an added sales benefit of further resonating with customers and so catalyzing further growth.
- Grüv provides a community-driven music experience. The business model here is based on an insight that algorithm-driven music recommendations even at the state-of-the-art level do not provide a great experience. Thus, data science is not immediately essential to the core product build-out but rather in understanding communities that form around the platform. As a result, instead of using algorithms to build the product, algorithms wielded by data scientists can help plan new city launches.
The actual hiring of data scientists is non-trivial, from technical assessments to presenting a diversity of work to keep them engaged. Ensuring the algorithm capability of an individual is a foundational screening step and platforms such as Hackerrank can help as well as assessing activity on Kaggle and Github. Founders have to set an inspiring vision and leverage their own networks, that of investors as well, and leverage support from elite recruiting firms such as Defined Calculus to find such specialty talent. Beyond technical chops, a data scientist should contribute to the story of a startup, whether that’s by adding algorithmic firepower to the team enabling scaling or resonating with customers by providing insights.
Data science represents a formidable way to accelerate startups as algorithms deployed on data responsibly can help build new products and serve customers better. A startup’s strategy should consider it as a way to make an outsized impact. As society grapples with challenges ranging from health and public safety to that of financial access and sustainable energy use, startups represent a key innovation capacity for the world. I’m looking forward to what solutions come next.
About the Author
Dr. Armen Kherlopian is a world-class strategy, innovation, and business leader as demonstrated by repeatedly driving new venture growth via gaining value from data. Moreover, his industry experience is cross-sector including Global Fortune 100 Companies as well as government organizations such as the FDA and NASA.
He holds a Ph.D. in Biophysics with a focus on Machine Learning from Cornell University and completed a fellowship in High-Performance Computing and Artificial Intelligence at Princeton University.
