Machine Learning and Predictive APIs: Driving All Experiences Intelligent and Digital

The Experience Driven Economy: Understanding Your Customer’s Psyche

You’ve heard of digital transformation and how it is ‘transforming’ the customer experience (CX). Products now are focused on enhancing the ‘CX’ to drive bigger and better conversions. Take e-Commerce for example, it is not just about building a site to transact. Online shopping has become a lot more personalized and targeted. Marketplaces like Amazon know me and are tracking my ‘buying behavior’ and psychographics pretty diligently.

Based on all my previous clicks and buys, they ‘predict’ what I want and make the appropriate recommendations. They are tracking my abandoned cart behavior to see why I did not complete the cycle. They are making some fair predictions on how much I am willing to pay for something based on heuristics and not just the fair market price. This is something that Apple has mastered, all you iPhone7, 8, and 9 fanatics will agreeJ.

Taking it one step further, this experience has to be consistent across all your channels – mobile, web, wearables, TV, appliances, and other connected devices (think consumer IoT). This is what the omni-channel movement is all about. There’s a vast ecosystem of tools, platforms and ‘clouds’ that are trying to drive several of these moving parts forward and below is a view of that.

The Science of This Experience: Small and Big Data is Driving the Future

Now, let’s get into the HOW of this experience. Although it seems quite simple, there’s a huge science and approach behind it. And here’s where the industry crossed the chasmfrom natural to what we call artificial intelligence to predict human behavior. This journey however like in Maslow’s hierarchy begins with data. And, all of this is driven by Machine Learning.

In simple terms, Machine learning is the ability for the computer to program without any ‘explicit programming’. Take for instance the search criteria kids shoes. Every time I query kids shoes I get the same related search/recommendation. But if I programmed this a little differently to include or show results from top sellers in this category and limit it to only 2, I might get different results each time because the top sellers vary. I might also program to include related categories like kids socks or kids snow boots depending on the season.

Anyways, the global point here is this: this is an iterative process! First, all the data around customer behavior is collected and analyzed and there are a LOT of tools driving these analytics – KissMetrics, Google Analytics, Adobe Experience Cloud, Crazy Egg, RJ Metrics, Clicky and several others. Once the analytics are in, it is all about making the right hypotheses, creating the fuzzy codes and iterating to drive larger conversions. Hope you’re with me so far?

Machine Learning in Marketing: A Shift in Mindset

We’ve established that data is key for applying Machine Learning to marketing and especially e-commerce marketing. We’ve also figured how you can collect this data through various advanced tools.

But now, it is about creating the right models via Machine Learning and NLP (Natural Language Processing) to drive the top and bottom line for your business. Let’s investigate that a little bit. This exercise is also known as ‘data modeling’ and it is slightly complex when there’s a larger data set. It all begins with attributes and assigning scores to them, just like in the marketing automation system where we assign lead scores.

For example, we want to know if Padmini is interested is buying an iPhone7, for lack of a better product. We then evaluate people like Padmini who have made similar purchases to see if they have bought an iPhone7. We assign scores or points to each click that they make and then determine that if the score was above a certain threshold, then we conclude that Padmini would make an iPhone purchase most likely.

Now, once we determine that hypotheses, we target and recommend an iPhone7 to her on every page when she arrives by designing some predictive APIs. Although the model is explained quite simply, it is not. And that is why big data scientists are priced items in the market, to figure out the ‘nuances’.

Good luck out there and stay tuned to hear some more machine learning stories from me.

Have Fun!

Visit to learn more about me.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s