How to Use Machine Learning to Reach Your Business Goals
How can machine learning help my business?
How can I get started with machine learning?
What tools can machine learning provide for my business and customers?
Imagine your lifelong friend Jin is going on vacation to the same spot you vacationed last year. He’s asked you to recommend some places for him to visit.
You know Jin loves swimming and tennis but hates arts & crafts and biking. Other friends with those interests have also vacationed here, and you’ve made recommendations to them and learned which ones were the best.
Given your knowledge of Jin’s interests, what are some places you should make sure you recommend for him to visit?
Just as you’d use what you know about Jin to inform your suggestions, Machine Learning uses data to make predictions to help your customers and business.
The difference, of course, is that Machine Learning, or ML, is a technology that uses software and algorithms to make these predictions. Luckily, to get your business using ML, you don’t need to know the nitty-gritty details of the stuff.
Instead, leave the software engineering to the software engineers while you focus on how to start using machine learning in your products and strategy to make a more streamlined experience for customers.
There are 3 basic levels for using ML in your business:
1. Improve an existing feature
For example, increasing the accuracy of a recommendation system
2. Enable new features
For instance, allowing for photo search by content instead of keyword or timestamp
3. Enable entirely new products
Think big, like driverless cars
Since machine learning is still a relatively new thing, the way that most businesses will incorporate it will fall into level 1, improving existing features.
To improve an existing feature through the application of machine learning, you should work closely with your engineering team. But think of this as an exciting learning opportunity for you, too.
Even though it’s just the first level, the skills you’ll use to implement machine learning at this stage are the same ones you’ll build off if you end up using ML in more advanced ways.
While you don’t need to know exactly how ML algorithms work, it’s good to know what general types of systems ML can create to make your work easier.
First, there’s recommender systems, which take a full collection of items (known as a “corpus”) and rank them. For instance, a streaming service might recommend movies from its collection based on a user’s viewing history.
Event or action prediction models try to predict the likelihood of an event or user action. For example, an online shop might predict if and when a user will close the page without checking out, and offer a coupon as an incentive to stay.
Classification models classify random objects into set groups. For instance, an email platform may categorize emails as “spam” or “not spam,” or a photo app could identify if an image contains a dog or not.
Generative models can take input you provide and generate the right output. For instance, a generative model trained to translate can look at the context of a text in one language and produce a correctly worded output in another.
Finally, clustering models can segment users into groups. For example, a flower company might cluster users into orchid lovers and rose lovers and target those groups with different promotional emails.
To understand how your business might use machine learning to improve an existing feature, let’s look at the real-world example of Google Forms.
Typically, to create a Google Form, a user manually names the form, adds questions, and chooses the question type – for instance, short answer, multiple choice, or checkboxes.
To improve their product, Google Forms came up with the idea to automate the question type selection. Not only can this improve users’ productivity, it would create a better product experience that users will be more likely to come back to.
With the help of machine learning, Google Forms can now predict the question type based on the text a user types into the “question” field, as opposed to having the user manually select the question type.
If a user types “Which foods do you like?”, the question type automatically switches to checkboxes since the question implies selecting more than one option. If the user enters “How happy are you?”, the type switches to a 1-5 linear scale.
LISTEN UP
Google Forms’ use of ML was pretty technical, so it was led by engineers, but there are plenty of places for non-engineers to get involved in brainstorming, figuring out how to best surface the experience (ex: suggesting vs auto-applying a suggestion), analyzing, or planning a launch strategy.
Of course, once you make improvements to a feature, you’ll want to make sure that it still works and does what it’s supposed to do.
Since ML models are trained on datasets from the real world and the real world can be messy, nonsense recommendations happen. Luckily, you can get an ML system to ignore a suggestion or correct an automated action.
In the case of Google Forms, during testing the team ensured a suggestion was only triggered when the ML model’s confidence was high. Basically, the feature doesn’t suggest a question use checkboxes unless it really thinks it should.
The Google Forms team also ensured that if a user explicitly changes a question type after the suggestion had been applied, that this was accounted for and the suggestion wouldn’t be made again.
DO THIS NOW
Now that you know more about machine learning – and that you don’t have to be a software engineer to start using it for your business – let’s make a plan for how you’ll use it.
If you’re participating in the course, go to the next section to access your self assessment.
KEY TAKEAWAYS
Machine Learning, or ML, is a technology that uses software and algorithms to make these predictions that can help your customers.
Since machine learning is still a relatively new thing, most businesses will incorporate it as a way of simply improving an existing feature.
There are plenty of places for non-engineers to get involved in using ML, such as through brainstorming, analyzing, or planning a launch strategy.