Product Recommendation 
 Big Data Analytics

96% of users are not ready to buy when they first visit your webstore

When a visitors arrive to your site you have a few seconds of their attention before they move to another competitor

In order to convert potential customers into sales and increase revenue when need to show them the most relevant items/products quickly, and we need product recommendations analytic for that

Potential benefits of Recommendation Engines:

Improving with use, user retention. One of the core potential benefits of recommendation systems is their ability to continuously calibrate to the preferences of the user. This makes products that become more a more "sticky" in their customer retention as time goes on

Improving cart value. A company with an inventory of thousands and thousands of items would be hard pressed to hard-code product suggestions for all of it's products, and it's obvious that such static suggestions would quickly be out-of-date or irrelevant for many customers. 


Personalized recommendation

This is the use case more often comes to our minds when we think about product recommendation. It's based on analyzing the customer previous buying behavior and compare to users with the same likes or habits

Image

Personalized recommendation


Non personalized recommendation

But, what is this the first visit to our site? We don't know anything about their previous behavior. Does this mean we cannot make any kind of recommendation?

Sure we can, what it's called non-personalized recommendation. Perhaps we don't know this new user, but we have plenty of data about our current customers. Here we have 2 typical scenarios

First, offer items similar to the one the customer is viewing right now. And as he progresses in its current session we can narrow the results

Image

Similar items

Second, Customers who bought this item also bought this other items

Image

Subscribe

Demographics based

It's not right that we know nothing about a new visitor. We always have their IP address, and from there we can know it's country, state, city ... and in some scenarios event its zip code. This is a lot of information to feed our recommendation algorithm

For example, image our website is a travel agency. It's the colder days of the winter. Someone from Norway is looking for a travel to another country. Most probably he is not looking for a sky travel.

Image

Travel recommendation by origin

We can cross the information about the zip code with the information from USA Census. We cannot know the salary of the user, but we can know the average salary in his origin zip code. We can also know lots of other things, like probability to be married, with kids, home pricing, avg number of car per home ...

Now we can compare this user to users with the same demographics data and purchasing power looking for correlations and things in common.

Imagine for example that we are selling second hand cars. Someone willing to buy a car for his son or daughter will not choose the same type of car living in a regular neighborhood than living in one with higher rents

Image

Home Income by County


Seasson based

Do we buy the same things in each season?

There're items correlated with the seasson, like cloths, foods, or travels. During the year we do short trips 2-3 days while in summer we can afford a longer trip, 7-15 days

We must use the season as one more input variable when creating our recommendation model

Image

Food consumed by season


Time of day / day of week

For items that are used or consumed in a brief period of time since their purchase another import factor is the time of day or even the day of the week

If we're ordering food delivery to our favorite restaurant it's not the same noon time in a labor day, than a weekend and a party at home


Conclusions

We must leverage product recommendation not only for existing customers but also for new visitors.

Only those companies that can offer the product the user desires in the few time the user visits our site will increase their revenue and survive in the market


Next steps

With the Isthari Big Data SaaS you can reduce the time to market and budget while adopting Big Data Analytics and navigating the Digital Transformation of your company

If you want a personalized demo, with your own data, you can schedule a first meeting filling the form below

If you want your company to be ready for Big Data Analytics check our Big Data Academy offering

If you want to be informed of new tendencies in Big Data Analytics just subscribe to our news letter

Schedule a personalized demo