interactivated blog

magento & ecommerce

Understanding Machine Learning in Magento 2

Some people might think about artificial intelligence (AI) and machine learning as upcoming events that could mean an uncertain future for humankind. But the truth is they are not so much a thing of the future as they are of the present.

Artificial intelligence is already present in multiple areas of our daily lives, performing tasks for us in the backstage, even if we don't realize it. It is used in our smartphones' virtual assistants, for example. Siri, Google Assistant, Cortana, and Alexa all make use of machine learning to perform speech recognition and learn to recognize your voice better the more you talk to them.

AI is even starting to be used in other areas such as science, healthcare, and banking. Even video games have been using AI for a long time. In technology, besides smartphones, which we've already covered, AI is also used in smart home devices and smart cars, with companies such as Tesla making a huge investment in this field.

This also applies, of course, to eCommerce. Human eyes can't do much with the incredible amounts of data generated by consumers' searches and purchases, and yet this data is begging to be analyzed by machine learning algorithms to improve the user experience. Even Amazon's CEO Jeff Bezos has referred to AI as the "fuel" that drives the company's success.

And we can't talk about eCommerce platforms without mentioning probably the most popular one, which is Magento. In Alexa's top 1M eCommerce stores (Alexa Internet, not to be confused with Amazon Alexa), Magento was ranked #1 out of 384 e-commerce platforms, holding 19% of the market share.

Being a leader in eCommerce, Magento is obviously the perfect platform for developers to explore with AI-driven algorithms and functionalities. We'll now take a look at some of the functionalities which are already being used, or might be in the near future, to improve the experience of users in Magento web stores all across the Web.

1. Targeted User Experience
This is one of the main features of many web stores where the power of AI is revealed without us even noticing. But most eCommerce websites are already showing users messages at the right time targeting their individual preferences. For example, they might suggest you products related to the ones you've recently bought. This is possible thanks to machine learning algorithms that take big data as input and use it to target the end user with personalized experiences.

Some people might think about artificial intelligence (AI) and machine learning as upcoming events that could mean an uncertain future for humankind.

2. Context-Based Search
There are some good search bar extensions for Magento that can already save you some time. They do this by automatically showing some results in the search bar itself if there is an instant match between the keywords you are typing and the product database.

However, even these search engines may be limited if there is no match. A search that is only keyword-driven may show you no results, even if there are products in the database that don't fit the keywords but fit the intent of your search.

An AI-driven search will look beyond the keywords and try to understand the context of your search, which can generate more relevant search results. It is almost like going to a brick and mortar store and telling the assistant what type of product you are looking for.

3. Virtual Purchase Assistants
While this feature is not fully implemented already, there are some big things we can expect. Much like a physical store assistant, a virtual purchase assistant will be able to inform you of current or upcoming deals, make suggestions and even add products to your cart so you don't have to.

This can be combined with chatbots, which are basically programs designed to emulate a conversation with a human, to make the experience even more seamless. You can solve any problems you are having without having to resort to the phone or the e-mail.

In fact, e-mail newsletters may even be replaced with notifications from the virtual assistant. In the future it may even be able to make purchases on your behalf, should you instruct it to do so.

4. Visual Search
We're already witnessing applications of visual search with Google Assistant and Google Image search, for example. If you take a picture with your smartphone, it does a pretty good job in recognizing the object, animal or landmark in the photo.

Soon, the same may be possible with eCommerce stores. Perhaps you've seen a product you've liked and you don't know its name, or maybe you want to buy something similar to the one you've found online. No problem, someday you might be able to search a web store based on an image link or uploading your own image and get relevant search results right away.

5. Fraud Prevention and Protection
With the growth in online channels, it is much more difficult to control fraud and theft issues than in person. This is easy to understand, because online it is easy for an account to be taken over with neither the customer nor the dealer being aware of it.

For this reason, merchants have to rely on human reviews to have a more thorough look at suspicious orders. As they do not want to take risks, a reasonable percentage of customer orders is delayed, leading to some frustration.

Furthermore, reports have found that a considerable percentage of orders discarded due to assumed fraud was legitimate after all. These are called false positives, and are a growing concern among eCommerce owners. With AI and machine learning algorithms, behavior analysis could be done over large amounts of data that can help mitigate this, as well as other fraud-related issues.

Some people might think about artificial intelligence (AI) and machine learning as upcoming events that could mean an uncertain future for humankind.

Magento Extensions Making Use of Machine Learning

You can already find an array of extensions for Magento that make use of one or more of the features we've discussed here, so let’s take a look at some examples. Note, we're only mentioning Magento 2 extensions.

The personalized Products module by Steven Richardson uses PredictionIO, an open source Machine Learning Server, to recommend products to customers, much like the “Customers who viewed this product also viewed…” feature used by Amazon.

Another great extension is Chloe AI by Sundown AI. This one does the role of a virtual assistant, answering your requests just like if you were texting someone. You can do this from your computer or even from a mobile device.

Finally, we'll mention the Mollie module, by Mollie Payments, which is great to illustrate the topic of fraud prevention. Mollie uses a machine learning algorithm to do behavioral detection and reduce your false positive rates.

We've mentioned three AI-using extensions here, but there are several others. With the rising popularity of both Magento and machine learning, there are bound to be much more. We've already taken a peek at what the future holds in store, but who knows how many more great features you'll be able to include in your web store?

By interactivated • on December 3, 2017

Contact Us
Get in touch with a specialist
THE NETHERLANDS (HQ)
Herestraat 106, 9711LM,
Groningen (Netherlands)
+31 (0)50 711 9940
VAT: NL 852998521B01
Chamber of Commerce: 58348646
UKRAINE
Kostomarivs'ka 13,
61002, Kharkiv
SPAIN
Calle Jabea 18, 29631,
Benalmádena Costa (Malaga)

* Required Fields