Empower Your Business With The Magic Of Machine Learning!
- +30 210 9701777
- Aristotelous 83, Athens 163 43
- contact@mobiplus.co
Empower Your Business With The Magic Of Machine Learning!
The goal of the Recommendation System is to predict where the user wants to go and help them get there with minimal friction.
So the user does not leave the site but comes back again and again !
I am sure you would like something similar for your business , your store , your e-commerce environment your e-shop , everywhere.
So read on !
YouTube is the leading online video content platform in the world.
It currently has 2.3 Billion users and is the most visited site in the world after Google.
For users who watch streaming videos it holds 94.5% of the market with Netflix second with 74.9%.
Users watch more than 1bn hours of video every day.
But how is this number so high.
66% of users in Germany watch YouTube to develop a new hobby.
94% of users in India watch YouTube to learn to do things on their own!
82% of users internationally watch YouTube to learn something.
Users watched 100billion hours of games in 2020.
72% of users watched YouTube to get fit in 2020.
So the strategy that YouTube is executing to catch the above numbers is:
It has the best recommendation engine in the world with more than 1 million lines of code.
YouTube is an amazing Knowledge Dissemination Engine for Maths , Physics , Engineering , Philosophy , Personal Development , Management , Do It Yourself , Entertainment and more!
The YouTube Recommendation Engine is a technological miracle!
So it doesn’t wait for the user to search to find what they want but guess what they want learn , see , have fun, exercise and recommend it.
It predicts where the user wants to go and takes him there.
But isn’t that what the user in your e-shop wants?
He wants to change something in his life and you need to help him get there.
He doesn’t buy clothes or shoes from your e-commerce environment because he has no clothes to wear.
He wants to become more attractive, to improve his home experience, to change something in his life, to go somewhere else!
This is what you have to predict when someone walks into your e-shop or your physical store and take them there.
To learn how the world’s top companies increase revenue up to 30% from existing customers using Recommendation Systems read the book below.
See below for some of the important features of YouTube’s Recommendation Engine .
O Christos Goodrow is VP of Engineering at Google and the Architect of the YouTube Recommendation Engine.
The Recommendation Engine has an important task to perform.
His job is to understand every day where we want to go and help us get there.
To learn our needs, which many times we do not know!
How many times has a customer entered your physical store and doesn’t know exactly what they want?
He knows where he wants to go but he doesn’t know how to get there!
Recommendation Systems is the area with the largest application of AI in the 21stth Century.
The first important point is how do you predict what product the user wants ?
The philosophy here is to show him something that is somewhat different from what he saw now, but not too different!
So you can help the user discover things that interest them!
The architecture here is the creation clusters (clusters ) of products and then seeing users who saw products in one cluster what other products from another cluster saw.
Example on YouTube have found that users who are watching Scientific videos also watch Jazz.
So to users who watch math videos then make recommend videos with Jazz.
These clusters are created using Collaborative Filtering.
And these are automatically generated by the algorithm based on what your users click on and buy within your your eshop.
Collator active filtering means which products the user sees together.
For example, this technology on YouTube creates automatic clusters of videos that have the same language.
Videos that talk about math , management , physics , yoga without being described in the video.
And then other clusters more specialized ones that combine more categories.
E.g. video management related to Innovation, Personal Development or Behavioral Economics.
Thousands of clusters based on user behavior that evolve daily.
So the algorithms are constantly looking and creating dynamic clusters every day based on user behavior within your e-shop or ERP!
The Machine Learning algorithms need customer purchase data from your e-shop or ERP.
They use category and video description initially (imagine your product)
When a consumer does a search on YouTube they use syntactic match to see if that word exists in the headline or category of the video.
Then they look at how much time the user sat on that video or left after a while , which means that the search word is not representative with the content of the video.
Then after you start watching some videos Recommender starts finding the clusters you belong to.
For how impressive the technology of Recommendation Systems is today, let’s look at the following example.
Someone doing research in America, for example in Physics, wants to see videos in English.
So if you search for something relevant you see video in English.
If you search how to make Baklava all the recommendations for cooking are videos in Greek.
So it predicts that you want the job in English and cooking in Greek!
It predicts that you want the job in English and cooking in Greek!
This is the Related Graph that is automatically created through collaborative filtering.
Also very important is for the Recommender to understand deeper the quality that the user perceives from the video (your product ) that he sees.
A first clue is to see which videos (think your product) you saw for starters.
But this is not enough.
The second thing to look at is how much time you spent on the video, did you see all of it ? half of it ? 10% ?
But even that is not enough.
Did you Like? Dislike?
Did you share it?
Did you write a comment ?
And that is not enough !
After a few days it asks you (Research) if you liked the video you saw. In order to make sure the algorithm is on the right clusters
.Has he subscribed to your e-shop ?
This says a little more !
Is he logging into your shop ?
This says little more!
Opens your application ?
Does it open your product?
Did you buy the product ?
How many stars did he put?
Does he want you to send him information about this product ?
Product, not your whole e-shop !
So it constantly improves clustering.
Plus they think about putting more than Like in a video you see..
Many times you put Like , but some times you don’t like it more than a lot.
Some other indication is needed e.g. I love !
You can see here the presentation of the YouTube Recommendation System by Christos Goodrow in an interview with Lax Friedman.
Contact us now to see how the mobiplus shopping recommendation platform can give you the above features and increase your revenue. 30%.
mobiplus member Elevate Greece
Confirmed Innovation.
Test out the Recommendation Engine for up to 20.000 euros in revenue from recommendations. No credit card required. Just enter your company information and we’ll contact you with all the details.
