Data science, the buzzword of this decade, came into the spotlight probably with the article “Data Scientist: The Sexiest Job of the 21st Century” by Dj Patil, originally published in HBR and then fueled by various companies utilizing the power of big data and analytics. The tech startup ecosystem of the west and other corners of the world adopted data analytics culture early.
Today, almost all big startups (of valuation around $100 million) have their in-house data science or analytics or business intelligence team. Whatever we call it, the core functionality of the team is to bring actionable insights from the data their users generate and help use that insight in building their products.
In Bangladesh, to my knowledge, Pathao was the first successful startup to build an in-house data science team (I am proud to have been a part of it). Other startups followed eventually and built data science teams to empower their day-to-day business operation.
If we consider the life-cycle of a startup, the operation starts with a seed or pre-seed round investment. The goal in the early stage is to raise the next rounds of investment while growing exponentially. However, it often happens so that people get confused about team building and structures despite having sufficient capital. It is needless to say that if a startup raises a round of investment, they must allocate the capital strategically. Optimal resource allocation is the ultimate challenge of building companies.
This is where the main thesis of this article comes in. I am writing today to suggest that building a data science or analytics team early in the life of startups can help a company to make better decisions at almost every stage of its operation. Data can optimize the operation and help make better decisions. My reasons to advocate the idea are clear:
1. Empower your day-to-day operation: From the day your operations start, users start generating and providing myriad types of data and footprints in your platform. This includes demographic, behavioral, and transactional data points. Consistently monitoring these data by the growth operators can provide an edge to your startup. For example, if the data says there is a demand-supply mismatch at certain hours of the day, you always should take some actions. The interesting part is that without looking into data understanding these types of mismatches is often hard.
Let's try another example. Suppose you see a drop in user engagement when there is a big cricket match happening, what actions would you take? Maybe you could send a notification saying something relevant to that match or you could partner with some company to offer a pizza with a discount from your app. Actions like these are likely to improve your customer engagement and retention eventually.
There is a myth that data science is only “science” but it is not. It is a practical thing and should be an integral part of your day-to-day operations. Data can help you with moving with trends and stay ahead in the market. This is why big startups like Uber, Airbnb, or Netflix have integrated analytics or business intelligence teams in every part of their operations.
2. Reduce your cost: Money has always been the center of growth for startups. With every ‘capital raise,’ there is always a target to reach with that money. Any business that has a transaction involved, generates enormous amounts of data from the signup to revenue funnel. These data points always imply a pattern of customer behavior and preferences.
Let's take an example of something more fundamental, suppose a company raises 1M USD for the next 18 months. They have a target of acquiring 250k customers within this period. Now, they are trying 4 different growth channels for the first 6 months. Let's analyze the data to see which is the optimal channel:
| Channels | CAC | Revenue (6 Months) Per User | Users | ARPU to CAC Ratio |
| A | $2.0 | 12 | 30,000 | $1.0 |
| B | $2.5 | 25 | 20,000 | $1.7 |
| C | $3.0 | 40 | 8,000 | $2.2 |
| D | $4.5 | 30 | 10,000 | $1.1 |
Clearly, channel C is better because it provides better user quality, which is evident from their ARPU to CAC ratio (USD 2.2 vs $1.1-USD 1.7). Although channels A, B, and C bring more users, quality is preserved better by channel C. Once you see this analysis, it is quite easy to make a decision.
The right kind of experiment and analysis can save money by helping you avoid less efficient channels and choose the optimal ones. If you run an experiment like this and in 6 months, can identify the right channel to grow efficiently, maximizing retention and engagement on your platform, you are more likely to raise the next rounds for sure. Such experimental setup and growth analysis for early-stage startups can save a lot of money - which can later be invested in more productive areas. We call this “optimized capital allocation”.
3. Create a culture of experimentation: My philosophy to get the right solution to any problem has always been through experimentation. A hypothesis-based culture can create a culture of experimentations in your organizations. From pricing to discount offers, everything can be fine-tuned through repeated experiments. It helps you to understand the “causal” behavior of your users.
Analysts who run the experiments are not always the ones who hypothesize it. As a founder, you should always be open to experimenting with new pricing models, new designs, new ways of showing your products or offers, etc. At the same time, you should be the key person to encourage others to experiment with new ideas.
4. Improve product and user experience: Constant monitoring and experiments will help you to understand the engagement, preference, and needs of customers better.
For example, do your users prefer cashless transactions? Do your users prefer an in-app messaging system to contact support? Do your users love to rate your product and do the ratings improve their repeat purchase rate or improve the likelihood of sales of that product? These questions are important to an analyst or a data scientist.
Data scientists who are also statisticians constantly design surveys and analyze them to gather these insights from longitudinal data. If these insights can be fed into the product life-cycle properly, the positive impact on user experience will be unimaginable.
5. Create new verticals of growth: When analysts say: “Data is the new oil” - they mean that like electricity revolutionized technological development, data will cause the next revolution in business and technology.
If your analysts keep analyzing data and operators keep monitoring it, there is always some interesting stuff about your user behavior that will pop up. For example: assume that you have an on-demand tourism startup. The average user base of your startup is urban and has a median age of ~25 years. Now, you run a survey and get the idea that there is a huge demand for tourism products in your user base. These data points suggest that you should cross-sell those products from your platform. Not only this will increase your GMV and ARPU but also will create higher engagement and stickiness.
Without continuous data analysis, you would not stumble upon these types of opportunities and probably some other player will utilize that opportunity. This is one example of how big startups keep launching new verticals, products, and partner with others using a data-driven strategy.
These are not the only reasons, there are more to articulate. For today, however, I think these are enough to convince the founders and their investors.
We always discuss in startups that what we are building today, we are building it for the future. It is evident that the future is all about this “data war”.
Many unicorn startups are now forming AI research teams, building in-house experimentation platforms, creating their own data pipelining tools, etc. to stay ahead of the competition. They all have only one core objective which is to create a platform that can infinitely hold a leading category position in their market and create a solid barrier to entry for anyone else even with better products and more capital.
Bangladesh is at a very early stage of VC-backed startups. With more capital injection comes more pressure to grow and build something that can be compared with startups from the West or India or Indonesia.
To meet these expectations, I would strongly suggest founders take data strategy seriously and invest in building a data team early, and think deeply about a coherent data strategy to make themselves untouchable by competition.
