Use your Machine Learning Superpower to impact the Fundamental Business Equation.
The nascent Data Scientist focuses on mastering TensorFlow & statistical concepts. This is undoubtedly important. Yet too many forget about the third layer of becoming a quality Data Scientist. Put the finishing touches on your Data Science skill set by understanding the business perspective.
Photo by Samson Creative. on Unsplash
This post explores the Fundamental Business Equation that each Data Scientist should understand. The content aims at educating technical Data Scientists who wish to have a tangible impact through their work. Find out which role Machine Learning can play in your company and how to find opportunities.
Let the digression begin.
The Fundamental Business Equation
The goal of a business is to make money. As Dominik Haitz excellently states, Data Scientists are hired to help the company achieve this goal. How do you define “making money”?
In Data Science, accuracy is defined by the number of correctly classified examples out of the total number of examples. Similarly, profit is defined by the two variables revenue and costs.
Profit is the difference between revenue and costs. We have an equation — holla! Let’s explain each variable.
Revenue is the total amount of money that your company makes. Think of your high school lemonade stand. Every dollar customers spend to buy your delicious refresher counts towards the revenue of your lemonade stand.
Costs describe all the money your company spends. For the lemonade stand, that is the price of the lemons, the fee for renting the stand, and your fictional salary, among others.
Profit is the difference between the money your company makes and the money it spends. For your lemonade stand, it’s all the money that remains in your pocket after you pay back Mommy for the lemons.
What does all this have to do with your job as a Data Scientist? Now that we know the key variables in the Fundamental Business Equation, let’s talk about your role.
Optimizing towards Profit
So these guys in suits at the all-hands meeting keep on babbling about increasing profit, dangerous competitors, and new business models. As they continue to bore the audience, your mind wanders back to the super simple yet Fundamental Business Equation. Then you have the idea — Bazinga! Just like in Machine Learning, you’re presented with an optimization problem. Awesome! The key metric we’re optimizing for is profit.
You reckon that to maximize profit, you can increase revenue, decrease costs, or both if you’re a magician. Let’s talk about how your work with Machine Learning can influence either variable.
Superpower Machine Learning
As a Data Scientist, you’re blessed with the superpower of Machine Learning. You feast on data and turn it into superhuman predictions to power mind-bending products. Let’s explore how to use your power wisely, Luke.
Photo by Lopez Robin on Unsplash
Andrew Ng, Founder of Landing.ai and former Google and Baidu Brain Lead, states that Machine Learning generates value in three ways:
- Create new products
- Improve existing products
- Automate processes
Let’s go through examples for each category.
Create New Products
Machine Learning helps your company create entirely new products to increase revenue. An example is new mobility services powered by self-driving cars, also called Robo-Taxi. Without Machine Learning, this new product is hard to create. In this case, Machine Learning allows the company to develop an entirely new product to increase revenue.
The holy grail of Artificial Intelligence (“AI”)-powered products is a product that enters the Virtuous Circle of AI. It all starts with a product that is already good without AI but continually improves with more users and more data.
Tesla is a good example. A Tesla car is fun to drive without AI but constantly improves as more users collect data. This makes the Tesla Autopilot ever more enjoyable. Elon Musk, Tesla’s CEO, estimates that his company collects 99.5% of all data from vehicles globally. Ready, set, game for Tesla?
Source: https://landing.ai/ai-transformation-playbook/
When looking for potential new products, you can do four things:
- Evaluate competitor product offerings. Are they offering products that customers want and you could emulate through AI?
- Look to other industries. Can you learn an interesting approach from this Chatbot AI company?
- Brainstorm new ideas. Can you come up with a cool idea during a hackathon?
- Check your data. Can you use this data to create a new AI product?
Machine Learning increases profit by creating new products, thus increasing revenue. You can also improve existing products with your superpowers.
Improve existing products
Since your company exists, it has a product or service that it’s selling.
Popular item recommender from is24.de for new users
If you work in digital business, chances are that a recommender system exists. In the example above, you see a popular item recommender by Germany’s leading real estate site. It recommends houses in Sweden to me – it’s not relevant to me.
Many companies still use popular item recommenders. These sweet and silly algorithms show which products have been most frequently bought from your company. With Machine Learning, you can build a nifty algorithm that offers customized recommendations.
Talk to Product Managers directly to find out how to improve existing products. You have the knowledge about Machine Learning, they know their product inside out. They know about its challenges and roadmap ahead. Together, you will come up with a great way to improve an existing product.
You’ve seen how to increase revenue by creating new or improving existing products with Machine Learning. Yet Machine Learning can also drive down costs. Let’s find out how next.
Automate Processes
Machine Learning reduces costs by automating processes. Every company has costs to create products. Machine Learning tasks like computer vision or natural language understanding support processes perfectly. The larger the company grows, the higher the potential for repetitive tasks that can be automated. This is often referred to as Robotic Process Automation, RPA.
A good example of using RPA to decrease costs is automating visual inspection in manufacturing. Visual inspection is the process of inspecting produced goods with your eyes. Imagine a team of people checking if newly produced cars have any scratches on the exterior. Audi automated this process by using computer vision. Landing.ai is an entire company focusing on that task.
Visual inspection of a manufactured good.
To identify tasks for process automation, look for repetitive tasks which need limited cognitive capacity. Andrew Ng gives the rule of thumb that anything that takes a human less than a second to do, can probably be automated. This way you can decrease costs and impact the profitability of your company.
This is a great example of decreasing costs through automating processes. Now that we know how to impact profit through Machine Learning, let’s prioritize the search for your next project.
Applying the Fundamental Business Equation
Management Consultants focus on increasing revenue before decreasing costs. Decreasing costs can cause unintended consequences on the quality of existing products. This leads to lowered revenue again.
Revenue before Costs. — Harvard Business Review
Photo by Franki Chamaki on Unsplash
Plus, it’s more fun to come up with great new products. This idea holds for non-profit and governmental organizations as well. Thus, the general advice is to increase revenue before decreasing costs.
Key Takeaways
As a Data Scientist, you get paid to optimize your companies’ performance towards profit. You know how you can impact profit by increasing revenue or decreasing costs. You’ve gotten this far, now save the juice for work.
Photo by Roman Kraft on Unsplash
- Optimize towards profit
- Profit = Revenue — Costs
- Revenue before costs
- Increase revenue by creating new products or improve existing ones
- Decrease costs by automating tasks
Next, use the AI Project Canvas to pitch your idea with an impact on profit to management.
