Here is a quiz for you. A wizard appears on your front door and offers to cast a spell that will make any wish come true. The catch is that 9 out of 10 people who accept the deal are turned to ash. Would you take the risk?
Most people would say “of course not” and I would agree – the odds are just not in your favor. Interestingly, those are the same odds entrepreneurs take when they begin a new startup. You read that right, 300 million people worldwide play those odds each year.
There is a big difference between that wizard and a startup. An entrepreneur gets to make choices that may shift those odds in their favor. Unfortunately, most of those choices are made blindly or following intuitions. And while I won’t deny that gut feelings are important, they have to go hand in hand with rational thinking.
The core principles of data science have a lot to teach entrepreneurs who are about to take that leap of faith. And they don’t need to be programmers or have PhDs in Economics to do it either. For example, one could hire an offshore software development company to develop the tools needed for robust and simple data analysis.
Lessons Startups Should Take From Data Science
- Gather data, lots of data
Over 71% of new startups fail for 1 of 2 reasons, either there wasn’t a market need for their product or they ran out of money. I can immediately tell that in those cases there was neither market research nor cost evaluation.
How big is your potential consumer base? How many people can you reach? How many can you serve? How many times will they buy your product in a single year? How much does it cost to run your startup? How much do you make per sale?
That’s just some of the questions a startup needs to answer before opening for business, and that’s where data gathering comes in.
Follow trends on social media.
Find review sites and look at what people are saying about similar products.
This isn’t a once and done deal, data gathering never stops. Situations change all the time and if you aren’t careful you might end up making assumptions with obsolete data which is about as bad as using no data at all.
- Forget common sense, follow the data
We are terrible at keeping probabilities in mind when we make choices, and most of our decision making is based on heuristics, not logic. Humans aren’t as rational as we think we are.
If there is one lesson to take from data science is that what seems obvious might not be true at all. Always find a way to back up your assumptions with data.
For example, don’t just assume that people are buying from you because you undercut your competition. Take a look at your data. You might find out that people like your customer service, or your delivery times, or something else that you might decide to drop to keep costs low.
- Correlation does not equal causation
Yes, this is probably the most repeated phrase in the history of statistics, and still, it bears repeating again. Don’t just assume that because two things happen together that one naturally caused the other, no matter how sensical it may seem.
Just because one startup did something and at the same time their sales spiked doesn’t necessarily mean that you can do the same thing and get similar results. Two things might be related in such a way that a third unknown factor is the one creating the correlation, or, it might just be a spurious correlation.
Keep in mind that correlations can be useful when used correctly. As an example, Target figured out the buying patterns of pregnant customers and sent targeted coupons on products they figured their customers would likely buy.
- Prediction, prediction, prediction
In simple terms, predictions mean that by knowing specific data you can guess the outcome of something, like a number or category (within a margin of error).
Predictions are never a guarantee, but they are better than throwing caution to the wind and hoping for the best. If you’ve been doing your data gathering, with a little help from statistical software you can make predictions that will help you make better choices.
With deep learning and AIs, we are seeing more refined tools that are making predictions for pretty much anything, including stock market trends
- Automation is key
Last, but not least, if you thought to yourself “Wow, that sure seems like a lot of work” it’s because it is. In the last 20 years or so data scientists have been developing new technology to help them gather better data, and faster than any person could do on their own.
Think of every mechanical and repetitive task that you do on a day-to-day basis. Most often than not you can use a computer program to make them for you, which won’t just free up your time to work on other areas, but it will help you minimize human mistakes.q
Startups are a big risk, yes, but with the right attitude and good planning, you can stack the cards in your favor. Be smart, gather information, and make informed decisions that will help you protect your investment – and avoid getting turned to ash.