I just needed a new pair of shoes… How did I end up buying all of these?!
When it comes to work, I have my standard routine: I play some music, prepare my coffee, visit some websites on a regular basis and then start actually working.
Those habits aren’t limited to my morning routine. I also have my free time and sleeping routine and they all include surfing the internet, visiting some websites that I find helpful, and some online stores that I usually end up buying from.
You may have different routines but these are actually my established patterns. And I used to hate receiving ads recommending some online goods that I end up buying, but now I understand why and it’s totally fine!
Did this happen to you either? Suddenly all the promotions you receive are exactly what you’re looking for!? And you ask yourself: How did this website know that I’m looking for such items?!
Well…Actually, You’ve been targeted!
Are you serious Amanda? You mean not everyone visiting this online store receives the same marketing messages?
No, not really! Don’t worry no one is watching you, but somehow you already provided information about what you’re looking for. Don’t you think? Let me show you how!
Last Monday you opened your Instagram and searched for ‘Hiking trails near me’ remember? This is how you ended up receiving ads recommending the hiking boots you’ve just ordered!
Mmm… Is that even possible? Can this online store understand from my search query that I’m going hiking and I may need new boots?
Exactly!
In simple words, some marketing messages can be classified as ‘one-size fits all’, similar to a one-size T-shirt. Other marketing messages are unique, or in other words, personalized, this is Personalized Marketing, sending the right message to the right person at the right time!
For you as a customer, the more you feel that someone understands your needs, the more you’re biased towards his store! This is simply done by collecting data from your browser to teach an artificial intelligence algorithm what you like or dislike based on your website visits and searches over time to predict your preferences and send you ads you may be interested in.
Before we proceed let’s see what we will cover in this blog:
- How is AI transforming the shopping experience?
- Stats and facts about the use of AI in e-commerce
- What are some AI applications in e-Commerce already in use?
- How is image search empowered by AI?
Trust me it’s simpler than you think, just be patient!
Can we rely on AI for a better shopping experience?
In e-Commerce, AI gives the ability to understand customers, offering them a better shopping experience, and a more feasible marketing strategy for retailers that minimizes their cost and reaches more customers, kicking off their profit margins!
Well, this sounds good but still not clear, right?
Simply speaking, machines can only learn from collected data. When a young lady searches for a black long dress and heels, recommending a clutch purse instead of a backpack is most probably all she’s looking for. But what if this purse matches the exact fabric of the dress? She won’t hesitate to order it! It’s the ‘Diderot Effect’ where the customer makes additional purchases to complete the overall picture.
This is what AI exactly does. By detecting a certain pattern in the search query, analyzing her clicks, how likely she makes additional purchases, her shopping basket, etc… The algorithm is able to anticipate her needs and therefore everyone is happy!
Still in doubt? Numbers don’t lie!
Do you know that Amazon started e-commerce in 1995 but started gaining profit in 2003 and is considered the biggest e-commerce store nowadays? According to Rejoiner, 35% of Amazon’s total revenue is driven by its recommendation engine.
As for ASOS, the famous fashion retailer, its yearly profit increased by 28% in 2018. The secret behind its success story is taking a step into AI-driven conversational interfaces. Customers in the UK can use Google Assistant to get directed to Enki, the ASOS shopping guide, by saying: ‘Hey Google, talk to ASOS’.
Similarly, implementing AI solutions such as robots to improve warehouse operations in JD, doubled the number of online orders in 2015 (compared to 2014) which reached 1.26 billion 85% of which were delivered within two days! And the list goes on…
A deep insight
Let’s dive a bit deeper…
Each click, view, like, and search query provides data to the recommendation engine. Here’s an example where the customer likes the three following images and how these interactions are actually understood.
Now the recommendation engine looks for brown heels with buckles and will recommend the perfect match! So the user ends up receiving something like this:
But how did it extract information? Now we’ve come to the juicy part!
Online activities can provide three types of information:
- User-Product Relationship
This relationship occurs when a customer tends to buy specific products. For example, an athlete might have a preference for Adidas products or healthy meals and therefore the website ends up building a relation such as Athlete→ Adidas.
2. Product-Product Relationship
Here items having similar features are clustered together such as ‘Italian dishes’, ‘Turtle Neck Tops’, ‘Wooden TV Units’…
3. User-User Relationship
This kind of relationship is when similar customers tend to have similar tastes about a certain product so the algorithm builds a relationship like: customers who have the same background (art for example) are more likely to buy neon colored tops.
With the help of AI algorithms, the system can correlate these relationships with other information such as user behavior data, demographic data, and product attributes to better understand the customer’s needs.
More applications? Keep reading…
Companies and retailers are relying more and more on AI by different means. Let’s see some of the AI implementations in eCommerce:
- Virtual Assistants: What’s better than having someone or actually something that can help you find what you’re looking for? The North Face brand uses Watson, a question-answering computer system that asks customers several questions and correlates answers with different data such as weather conditions to find their ideal jacket.
- Product tagging: Every product of an eCommerce store is unique and has its own description (color, texture, size, casual…). These tags not only help customers find specific items but also improve the store organization for example having ‘Accessory’ as a category and ‘Necklace’ as a subcategory.
- Trend Forecasting: Believe it or not, AI can actually predict the future of fashion! In fact, RushOrderTees, a technology custom apparel company, has been using AI not only to design new clothing but also to predict what people will be wearing in 2030!
A picture is worth a thousand words!
How many times can you picture something unique in your head but just cannot describe it accurately? I mean it’s not easy to describe a sunglass with a unique pattern! But what if e-Commerce websites upgrade their search strategies to support images?!
Wait a minute… you mean I can just upload an image of what I’m looking for? YES exactly!
Here’s how this works: the process is divided into two phases, an online phase, and an offline phase.
Offline phase: During this phase, the company catalog containing all product images is passed to a machine learning algorithm which will extract information representing the product-related features. So we end up with a model able to perform the feature extraction task and a list of features vectors.
Online phase: When the client uploads an image, the ML model extracts features from the image. What do we need to do now? Find a similar feature vector from the list we already have and therefore output similar images! See how simple this is!
Actually, the IKEA Place APP supports image search where customers can just point the camera to a piece of furniture and the engine tries to find similar items and not only this but also gives the ability to use AR (Augmented Reality) technology to see how it looks in their place!
Conclusion
Well, that’s all so let’s wrap it up!
In this blog, we saw how AI is shaping eCommerce aiming to better understand customers and improve their shopping experience by analyzing billions of interactions every day. AI algorithms can discover existing relationships between customers sharing similar interests, customers preferring products with specific features, products belonging to the same clusters, and much more.
And now that you understand the reason behind these personalized marketing messages, I’m sure you know you should check it out!
So be careful whenever you like or search for a product… someone is watching you 👀
Don’t forget to support with a clap!
You can join our efforts at Zaka and help democratize AI in your city! Reach out and let us know.
To discover Zaka, visit www.zaka.ai
Subscribe to our newsletter and follow us on our social media accounts to stay up to date with our news and activities:
