Select Star: A Startup That Aims to Be Where Every Data Analysis Begins
Q&A with Shinji Kim, Founder and CEO of Select Star
This story is part of the Entrepreneurship of Life series, a collection of interviews with immigrant startup founders, venture capitalists, and tech business leaders in the US.
INTRODUCTION
Rewinding to March 2020, when everything everywhere seemed to be shutting down because of COVID-19, Shinji Kim started her second business and named it “Select Star”. Few things sound more familiar to anyone who knows basic data science: “SELECT *” is the very first command in every SQL query, the most common data analytics activity. Just like this command, Shinji wanted her product – a data management SaaS tool – to be the default place to start for those working with data.
Wearing glasses and a simple ponytail, Korean immigrant Shinji could easily pass for a graduate student. The truth: she is a seasoned serial founder who sold her first company 6 years ago, at just 29. Before that, she had left a well-paying consulting job only 2 years out of college to build a popular mobile game from scratch to 50,000 users – on her own.
Again, she is building something new. Barely two-years-old and six months into its public launch, Select Star already has a roster of happy business customers (among them, global logistics company Pitney Bowes with 11,000+ employees). The startup raised a seed round in March 2020 and (as of May 2022) has 15 team members spread around the world, actively hiring a dozen more.
Earlier this year, I sat down with Shinji to hear her story firsthand. We chatted about topics including:
What value does Select Star provide, and to whom?
Before founding, how did she find a worthy problem to solve?
How does Select Star take on open-source competitors, acquire customers, and keep them happy?
What founder lessons did Shinji take away from her last startup and its sale to a strategic partner?
What is it like to be an Asian female founder in a gender-imbalanced industry?
Let’s dive in.
Shinji Kim is the Founder & CEO of Select Star, an automated data discovery platform that helps organizations understand their data. She also advises and angel invests in early-stage developer tools and B2B SaaS startups. Previously, she was the CEO of Concord Systems, a data infrastructure startup acquired by Akamai Technologies in 2016. She led building Akamai’s new edge computing data platform for IoT (Internet of Things). Prior to Concord, Shinji was the first Product Manager at Yieldmo, one of the largest mobile ad exchange networks. Before Yieldmo, she was analyzing data, building enterprise applications, and advising businesses at Deloitte Consulting, Facebook, Sun Microsystems, and Barclays Capital. Shinji graduated from University of Waterloo in Canada. Her family moved from South Korea to Canada when she was 13, and she currently calls San Francisco home. In her free time, Shinji enjoys her friends’ company, movies and outdoors (skiing, hiking or doing yoga).
(Note: Bolded questions below were from me, and the rest were summarized responses from Shinji.)
I. THE SELECT STAR STORY
Could you describe in layman’s terms what Select Star does, who uses it, and how Select Star provides value to them?
Select Star’s mission is to make data discovery easy by helping anyone find, understand, and use their organization’s data. We provide a SaaS (Software-as-a-Service) solution that automatically generates metadata context — where this data came from, who’s using it inside the company, and how they’re using it.
As organizations grow and use exponentially more data, context for data becomes increasingly hard to access. Imagine you are a data analyst looking into sales, and your SQL query for “sales” returned a dozen results that all look similar. Which of them is relevant? If you work for a micro business and do this occasionally, asking someone might do. But if your company has a hundred people or more, and you run these analyses every week, “asking” quickly becomes cumbersome and eventually infeasible. How can you get key context for your data, such as its freshness, popularity, sources and uses, without asking? Absent such context, finding the right data and using it properly is difficult at best.
Select Star is built to be the solution. We analyze logs of databases to automatically generate crucial insight on data, which then guides proper data usage. When our users search for data, our data popularity tool ranks search results by usage popularity, so that users won’t accidentally cite a table that nobody has touched for ages. In addition, our data lineage function allows users to trace the data flow of any dataset, both upstream (to where it was generated), and downstream (to all business reports and dashboards constructed on this data). This visibility is important for both understanding data and managing data changes to avoid unwanted impact.
Our product can be used by both data producers (e.g., data and software engineers) and data consumers (e.g., data scientists, business analysts, and product managers) in virtually any organization.
[Author note: read Select Star’s blog posts here and here to learn more about its service.]
Before we dive deeper into Select Star, let’s first hear about the path that led you here, as it seems related to what we’ll discuss later. You took on a variety of roles such as strategy consultant, engineer, data scientist, and product manager; you also founded another startup. What is the common thread?
It might not be obvious, but through them all I’ve been a builder my entire life. As early as high school, I was building websites for local small businesses. In college, I studied Software Engineering from University of Waterloo in Canada. Waterloo’s undergraduate co-op program is known for its multiple externships; I did six different in/externships through college, building enterprise applications, statistical models, and data pipelines for business use cases. From these experiences, I started to develop expertise at the intersection of software engineering and data science.
By the time I graduated, I was well trained in how to build things and started to think about what is the right thing to build at the right time?
I chose a new path to learn to find the answer – strategy consulting. I joined Deloitte as a management consultant and worked for two years alongside business analysts in solving business problems. It complemented my technical background and taught me new perspectives.
That said, I started to miss the fun of building products as time went on. As a consultant, I don’t generally get to implement what I recommend. So once my student loan was paid off, I left consulting and started working on an idea I had – building a game app from scratch. I worked by myself, launched the social puzzle game ShufflePix on iOS nine months later, and acquired over 50K users. It’s been almost a decade since, and the game still has a 4.5/5 App Store rating today.
It was fulfilling to go from zero to one alone, but working solo made me miss the team vibe. So I joined Yieldmo, a mobile advertising platform startup as their employee No.20 and first product manager. [Author note: Yieldmo is one of the largest mobile ad providers today, serving companies like CNN, Walmart, and others.] At the time, Yieldmo was struggling with processing a massive volume of user events from our publishers (over 10 billion a day), and we could not find any good solution in the market. So my lead engineer from YieldMo, Alexander Gallego and I figured we would build one ourselves. We co-founded Concord Systems, a high-throughput, low-latency data processing engine to serve companies with a similar challenge as Yieldmo.
What we built at Concord achieved 10x better performance versus the status quo alternative. Two years later, we sold it to Akamai Technologies, who saw a promising future for Concord as part of its own IoT product roadmap. [Author note: Akamai is an industry-leading content delivery network (CDN), cybersecurity, and cloud service company.] After the sale, I led IoT engineering and product at Akamai for another 18 months.
Post acquisition, I started angel investing and advising other founders. I was a startup mentor at Grand Central Tech, Techstars, and First Round Capital. It was fun, but after a while the builder junkie in me got itchy again. When I get involved in something I like to go all-in, but I can’t by sitting in the coach seat – the only option is to get back into the field and play.
I revisited a few data tooling ideas I had developed at Concord but not had a chance to work on. After a lot of research and consideration, the concept of Select Star was born. A new journey of building began!
Out of the ideas you considered, what thought process landed you on Select Star?
I vetted the ideas based on three main criteria.
The first is tangible value-add in solving real problems. In B2B SaaS, it is easy to “solve” an imagined problem. It is also tempting to get excited on high-level thinking alone (“this is a core technology, so it must be applicable everywhere!”). Founders often neglect to ask themselves: could my product address a concrete pain point in a specific industry context? Not only so, could it provide significantly more value to customers in that industry than their existing tools? I could answer yes to both for Select Star.
Second, I wanted to build a user-facing SaaS product rather than an infrastructure platform. I wanted the problem set we are solving to be closer to business use cases and business value. In addition, SaaS products can be developed and iterated in an agile manner based on customer feedback, whereas building infrastructure software tends to takes a lot more time upfront in order for the base foundation to be stable for users to utilize the platform. Agility and short feedback loops are great for getting a new product off the ground.
Finally, I wanted to make sure there was a clear market opportunity. I thought Select Star was very timely given where the data market is today and where it’s heading. I learnt my lesson from Concord, which was born a bit early on the adoption curve. While interest in its technology (distributed stream processing) was growing, not many companies were technically savvy enough or have a strong enough business case (i.e. their data processing volume has not yet strained existing tools) to tap into its full benefits. While Concord had a good run and found a suitable home at Akamai, had it been started a few years later, the growth trajectory may have been different.
This time, I’m determined to build something the industry is ripe for. Today, data analytics and data computation workloads are shifting to the cloud, and this is creating a new wave of data tools. I believe the way most companies are going to consume data in the next 5-10 years calls for a solution like Select Star today.
Locking my eyes on this concept, I held more than a hundred user interviews with people working in different roles related to data, in order to design our initial MVP (Minimum Viable Product).
Fast forward to today, Select Star is being used by both enterprises and fast-growing digital native companies, including Pitney Bowes (11,000+ employees), Opendoor (2,800+ employees), and Faire (~700 employees). Our customers are saving significant time and costs from using Select Star: for example, Pitney Bowes (where 12 teams currently use our service daily) has reported a 67% efficiency increase in their data asset cataloging and is rolling out our solution company-wide.
Let’s talk about competition. Multiple data-driven big tech names have open-sourced their in-house data discovery platforms, such as Amundsen by Lyft and DataHub by LinkedIn; more are speculated to do the same. These platforms are being used by other scaled tech companies (for example, Amundsen is adopted by iRobot and Workday). Some of them have similar service offerings as Select Star’s, such as data popularity and data lineage tools. How does Select Star differentiate itself from these competitors?
Our main differentiation comes from automation and user friendliness, especially for non-technical users. Open-source solutions require a high degree of customization and therefore dedicated engineering resources that many organizations can’t afford. They are also generally not well suited for business users. In contrast, Select Star is designed for “plug and play” by a much broader user base: our customers only need to connect their data sources to gain auto-generated data insights right away, in a format digestible by almost anyone in the organization.
In the early days of Select Star, we delved into existing open-source data discovery platforms including Amundsen, DataHub, Apache Atlas, the platform of Facebook, Uber, Shopify, and more. As part of our market diligence, we wanted to understand how these tools were used in their respective creator’s business case. We also considered whether to build Select Star atop one of these frameworks. What we had found through talking to their users was a market vacuum: while large enterprises can tailor an existing open-source solution to their needs (and they often have to anyway so that it works with their internal systems), for smaller companies without significant engineering resources, customizing and upkeeping an open-source solution is a big burden, and there was no suitable, turnkey product for sale.
We decided to fill the void, and to do so effectively required us to build a new service from the ground up. Automation and ease of use are at our core – for businesses leaning towards “buy” vs. “build” for their IT needs, we want our product to give them as much value as possible while requiring as little effort on their part as possible; we also make our data insight directly usable by their business analysts, product managers, and other non-technical team members, so that the data team don’t need to act as intermediaries between engineering and business operations. Our offering allows customers to add their own custom documentations, tags, and workflow to suit their needs.
The bottom line is, open-source discovery platforms aren’t for everyone, and Select Star exists to be the answer for those looking for a one-click, managed solution.
What has Select Star learnt about its customers so far through product experiments?
Sometimes our customers turn out to have a different feature preference from our initial hypothesis. Listening to them helps us refine our development roadmap, sometimes adding a requested new tool, other times zeroing in on one of several features we tested that had the strongest reception.
That said, we need to be careful not to treat our most active users’ voices as gospel. This may sound counterintuitive, but their feature requests might sometimes go against our own product vision, and a thoughtful balance is needed so that what we build can benefit a broader set of customers in the long run. In fact, we have at times decided to undo changes initially demanded by our early adopters, to the extent they do not fit well with our big picture objectives upon a close review.
What are your strategic priorities over the next 12-18 months?
At this point, we have figured out the base foundations of our offering, which include automated data catalog, data lineage, and usage analysis. For the next phase, we’ll focus on three things:
First, building application features that leverage our core services. For example, based on our data lineage capability, we’ll create automated data change alerts for the affected downstream users, and we can automate the propagation of a documentation update throughout the data pipeline, so as to eliminate repetitive manual work and confusion due to outdated documentation.
Second, supporting more integrations beyond data warehouses and BI tools, so that customers can get these insights and automation end-to-end.
Third, scaling our go-to-market efforts to support more customers, now that Select Star is being used in production by more than 20 companies.
How are you acquiring new customers?
We’re still in early stages, having just launched our general availability last November. So far most of our leads are inbounds, from both my first/second-degree connections and a growing number of people who discovered us through our blog posts. I’ve also been on podcasts and industry events talking about the data analysis pain points that Select Star addresses. We often hear from readers and audiences that this was exactly their headache and they would like to try our solution.
Select Star is a permanently 100% remote startup. Tell us what it is like?
It has been great so far, and I’m excited for what we could become!
Right now we’re 15-person strong; I’m based in San Francisco, and my teammates live in the Bay Area, Boston, New York, Toronto, Virginia, and Poland. We have learnt to collaborate across time zones, and some of us – mostly the Polish members – work together in our Warsaw office regularly. This team setup allows us to support customers globally from Day 1 (we already have several customers in the EU) and facilitates more focus work for everyone with time saved from commute.
That said, bringing the team physically together once in a while remains crucial for team bonding and morale. I tried to find every organic opportunity for my team to meet up – industry conferences are good catalysts for example; we also organize offsites once or twice a year.
As our team grows, collaboration framework and processes will need to be revisited and updated, but I want us to be like GitLab one day and work seamlessly together as a team of thousands.
II. OTHER FOUNDER MUSINGS
Can you tell us about the decision to sell your last startup, Concord Systems, after a two-year run? What made it a good time to sell?
At the outset, my co-founder Alex and I hadn’t anticipated selling Concord in just two years, but by the end of it we had overachieved what we had set out to do, and we found both vision alignment and cultural fit with our acquiror Akamai.
When Akamai approached us, we were debating whether to raise an A-round or take one of the several buyout offers on the table. Our high-caliber team and intimate knowledge of distributed systems had drawn “acqui-hire” interest from several large industry players.
Akamai was a different case: they were a pilot customer and saw our technology fitting perfectly with their own strategic vision for a foray into IoT. They are an expert in distributed systems themselves and run the largest Content Delivery Network (CDN) service in the world; they wanted to leverage the CDN infrastructure platform and build an IoT data platform on top, and Concord could be a key piece of the puzzle.
We were impressed by this vision; meanwhile, as mentioned earlier we concluded that Concord’s market timing was early, and it’d take tremendous time and efforts to continue educating the market and wait for it to catch up. We were eventually convinced that by partnering with Akamai, Concord could generate greater impact for the global enterprise customer base of Akamai.
From Concord to Select Star, what has changed in your approach to founding and running a business? What learnings are you applying?
As alluded to earlier, I’ve become more thoughtful about validating ideas and finding product-market fit. In addition, I’ve learnt to be much more selective with whom I raise capital from. I did not fully appreciate the first time how crucial it is to work with investors who truly buy into the founders’ vision, understand what they go through, and support them in the inevitable ups and downs along a startup journey. Too many investors would write checks hoping to win the lottery, but can’t deliver the promised partnership when founders need it the most.
What is it like for you to be a solo founder today versus a co-founder last time?
There are pros and cons to each. As the single founder of Select Star, I can make decisions much faster and have a single voice to align my team. On the flip side, it could be overwhelming at times having to deal with a myriad of decisions by myself.
I did not intend it this way. When the idea of Select Star came together, I was “founder-dating” [Author note: refers to screening co-founder candidates], but the business eventually moved too fast for it to pan out. A blessing and a curse!
Fortunately, I have an amazing team (my “founding team”) who are dedicated and hard-working to make Select Star great. My background also prepared me well for wearing multiple hats. I was a founder, an engineer, a product manager, and a salesperson all at once, when we simply needed things done to get Select Star off the ground. As we scale, I try to create operational leverage by developing a team that manages itself. I have both hired and promoted from within senior team members. Although I didn’t have a cofounder to start the company with, I’m making some along the way.
III. THE IMMIGRANT STORY
What brought your family from South Korea to North America? What were your early days here like?
When I was 13, my family uprooted and moved from South Korea to Calgary, Canada. The decision changed our lives, and my parents did not have it easy. Back in Korea, my dad had worked in pharmaceutical sales and my mom had run a large preschool as its vice principal. Despite those thriving careers, in a brand new environment with an unfamiliar language, they couldn’t find matching jobs in Canada. My parents had never lived outside Korea before leaving their birth country behind.
To lay down roots in the new land, where they believed their daughters would have a brighter future, my parents had to start from the bottom. In our first several years in Canada, they worked many gigs: laundromats, convenience stores, whatever they could do to make things work. Looking back, the transition must’ve been tough for them, and my sister and I are both grateful and admire them for how hard they’ve fought to earn a spot here – halfway across the globe from where they came, to give us a new home.
Has being a woman or Asian affected your founding experience?
Yes and no.
Take fundraising as an example, which is emblematic. Generally, regardless of who you are, you need to prove your idea’s worth to raise capital, often with tangible results. That said, investing is partly subjective. When people make judgments including on whether to back a founder, pattern matching can be an unconscious part of the thinking. This may disadvantage a minority founder, who on surface might not fit a familiar pattern for successful founders. Asian female founders are rare, especially in B2B enterprise and data infrastructure; as a result, she’ll need to go the extra mile to network and advocate for her idea.
But so what? A founder already needs to take on so many greater challenges. If you can’t change the environment today, you must find a way to work through it (or around it, and by doing so contribute to changing it over time). I ask less of why things are the way they are; I ask what I can do in spite of.
Keyi (Author): If you enjoyed this story, subscribe (for free) here to my interviews of immigrant founders, VCs, and operators, and you’ll get new stories in your inbox. Also — why not share it? :)
Find me on Twitter or LinkedIn, and leave any comment below!