Amazon Echo-like AI to disrupt CRM, grow sales

AI Powered CRM Systems to Fuel Sales Growth

Small Businesses will grow sales using AI powered CRM systems that behave like the Echo.

The Amazon Echo uses Deep Learning Artificial Intelligence to understand and respond to human language. CustomBulkUSB is looking to AI enabled CRM and SEO solutions to help us improve customer service and grow sales at lower acquisition costs. Already, we are seeing initial impact on our sales of custom USB flash drives.

There’s a lot of buzz about Big Data, AI, Machine Learning and Deep Learning nowadays. But what is all of this mumble-jumble? And what can it do for small businesses, startups or a local mom and pop shop? The short answer is, A LOT! Probably much more than you think. Let me give you a quick example using AI powered CRM.

How CustomBulkUSB Using AI-Powered CRM to Prioritize Leads and Increase Sales

Example of machine learning model using MS Azure ML Studio

• First, we mashup all of our data, from our website Google Analytics information to Adwords, Webform Leads and CRM systems data
• Next, we pull our sales dollars from PayPal and other payment systems and mash up that data with all of the other
• Then we use some tools to transform and clean up the data before uploading it all into our Microsoft Azure Machine Learning Studio
• Once in MS Azure ML, we perform classification, clustering and regression analysis to create machine learning algorithms
• Results:
  • Better classification of customer keyword phrases with the highest sales conversion rates
  • Ability to score probability of a lead becoming a sale based on 8-10 key variables: we get a score of 0 to 1 for each lead; a 0 .87 means there is an 87% likelihood that we can convert this lead into a sale.

How AI Got It’s Groove Back

Flash-back to the start up of the Internet in the mid to late 1990s when nearly every business transaction (except phone or fax) was done face to face, in person — nothing was online. In a few short years that all changed. We are witnessing a similar change unfold before us: the Internet, software and many of our everyday physical products will start to behave more like humans. We are embedding these items them with a narrowly focused type of artificial intelligence called machine learning and deep learning.

After more than 50 years of trying AI, we suddenly saw major break-throughs take place in 2016.  And we predict that 2017 will be the Year of AI.  Amazon Echo, Salesforce.com’s AI Einstein, Google Translate, Self-Driving Cars are all examples of revolutionary changes now underway.  Why now? Primarily because of confluence of events, creating a perfect storm for AI:

• Computing power or Moore’s Law.  We have more computing power on our smartphone than NASA had when it sent a man to the moon. Indeed a person armed with one of our 64GB or 128GB OTG mobile flash drives and a mobile device has more computing power and storage than many desktops did a few years ago.
• Low Cost Data Storage.  Flash memory and other material science advances make storing data dirt-cheap.
• Big Data.  AI machine learning and deep learning algorithms love lots of data. And data is exploding in volume, velocity and variety.
• New Mindset about AI. The experts have learned that rather than trying to control AI with perfect if-then statements and software coding, it is better to feed the AI algorithm lots of data and let it find patterns (learn) on its own. Feedback or coaching / training can then help to refine how the algorithm will learn better and faster the next time.  For example, this is how the auto-complete function works on your cell phone.

“The last 60 years of computing, humans were adapting to the computer. The next 60 years the computer will adapt to us. It will be our voices that will lead the way; it will be a revolution.”  – Brian Roemmele, Forbes, Quora

I remember when my seven year old daughter was two years old and could not understand why she could not move her fingers across a computer screen or TV screen to tap, scroll or pinch in the same way she interacted with my smartphone. Earlier this week, I heard of a mother whose child tried to talk to the TV as if it was an Amazon Alexa speaker. As we now know, the Amazon Alexa is physical speaker with Artificial Intelligence software embedded inside of it.  This AI allows it to recognize human speech and respond back using natural language as if it were a real person. And that provides us peak into the future of computing!  But what is AI, really?
Quick Intro to Machine Learning and Deep Learning
Think about how you learned to speak English or whatever language is your native tongue. Now think about how we often try to teach you a new language in school or college today. Which way works better?

Well, for nearly 60 years we have tried to teach machines (computers) by using artificial intelligence in the same manner.  Think of how we teach students.  We make them sit in a classroom, we teach be rote memory, grammar, syntax — lots of rules, lectures, theories, concepts and endless mind-numbing tests. And few people learn to speak a new language even after three or four years. On the other hand, I lived in Brazil for a year and even after 15 years of living back here in the USA I can speak near-fluent Portuguese; much better than my Spanish — which I studied for more than four years in school and college.

What was different? While living in Brazil I kept daily journal notes to observe how I was learning the language. When we are newborn babies we cannot explain how we learn a language but as an adult in my early 20’s I was able to observe how my brain was learning. I learned the language by “context.” Based on the context (time of day, setting, location, type of people, nature of the conversation) I picked up on words and phrases that seemed to be repeated by lots of different people. I then used the phrase in conversation and observed how people reacted to what I said and learned to make adjustments based on that feedback; it was often non verbal feedback.

We have learned the same lesson with AI:  don’t try to teach an algorithm with endless rules and programming logic.  Rather, let the algorithm loose in the wild to simply observe large sets of data, huge volumes of images and lots of  voices and languages. AI will then start to identify patterns and in so doing learn to learn.

That new approach is leading to what seems like yet another major break-through in AI every day.

Natural Learning vs. Rules: Algorithms Love Big Data!

As noted above, we have found that rather than trying to define every rule using software logic (IF, THEN, ELSE statements), it is better to let algorithms discover patterns on their own from lots of data. This is the same way a child — or an adult — truly learns a language. By DOING, not pontificating. And the AI experts found that this works very well.

So this is where Big Data comes in. Experts say that we as a society have now  generated more data in the past two years than we have in the entire history of humankind! And this data explosion will continue to grow due to all of the data coming from industrial equipment, purchasing transactions in stores and online, machine sensors, our cell phones embedded with GPS and accelerometers, website visits and log data, social media, text, voice calls, healthcare data, quantified self and DNA info, videos, photos and other types of images created all around the world. While all of that data is overwhelming for us humans, algorithms simply love lots of data. Properly modeled, the more data an algorithm consumes, the smarter and faster the algorithm becomes. The big fear today is that the algorithms are surpassing human intelligence. We will save that topic for another day. Let’s see a simple example of how an algorithm works.

How Algorithms Work

Imagine if we fed an algorithm the height and weight of all of the males and females in the USA.  It would look at the information and notice a pattern that looks like the scatter plot below (Source:  Steven Buechler, University of Notre Dame).

Plot of Height vs Weight for Males, Females

In this example, we are only examining two variables or two dimensions: height and weight.  In the real world, there are often dozens, hundreds or even thousands of variables. On a two dimensional sheet of paper it is easy to see the relationship between these two variables (one on a horizontal axis, the other on a vertical axis), but when we have five, six or seven dimensions it is impossible for the human mind to visualize all of these interactions. However algorithms can do this very well. Perhaps you have met people who have excellent intuition or are considered very wise. These individuals are often seeing multiple dimensions through their other senses, much like the way good AI algorithms work. In addition to the height and weight data shown above, algorithms can find patterns in images and sounds as well and come up with equations that describe those patterns. Those equations then form the foundation for how the algorithm will then learn and adjust based on feedback.

AI vs Machine Learning and Deep Learning

To aid understanding, I will take the liberty of simplifying things a bit. For those seeking a more comprehensive understanding of this topic, take a look here.

We can think of AI as the overall category for how we teach machines intelligence. Modern AI has been around since the mid to late 1950s. Machine Learning is a subset of AI that uses things like classification, clustering and regression algorithms (for example, the line describing the plot of height vs weight above is a linear regression algorithm) to find patterns and help improve business decision making by reducing the cost of prediction.  Deep Learning (only been around in past 5-6 years) is a subset of machine learning and uses lots more data and more complex algorithms that mirror the way the mind works often using what is known as neural networks. Image recognition, language translation and self-driving car problems are often handled with deep learning algorithms vs machine learning.

How AI Will Impact Small Business Sales and Customer Relationships

So how will this technology impact every small business, startup or mom and pop? If you think about successful products, they have three elements working together very well: software, the infrastructure which runs the software (hardware, computer servers, routers, etc) and the users who make it all useful.

Consider Customer Relationship Management (CRM) software which nearly every business today uses. CRM systems like Zoho and Salesforce are like an online Rolodex on steroids. They are cloud based or online systems that keep track of all of your customers and the many touch-points you have with those customers. Your customer database is often one of the most valuable assets of your organization. Salesforce.com completely disrupted the CRM market (driving the market leader Siebel to join forces with Oracle or risk bankruptcy) by combining two of those elements: the software and the infrastructure into one. This meant that if you wanted a CRM database, you no longer had to pay thousands of dollars for the software and tens of thousand of dollars for all of the computer servers, routers and other hardware to set it all up. Instead, you simply logged on an online system such as Zoho.com or Salesforce.com for a free trial or by paying a low cost monthly subscription rate.  This led to the rapid rise of SaaS (or Software as a Service), meaning you simply rented it as a service — pay as you go / grow.  Salesforce.com was one of the first companies to prove that SaaS was a viable and profitable business model. Many others then followed.

Machine Intelligent CRM

We believe that the next big thing in CRM systems will be AI-powered CRM.  These new CRM systems will combine AI with software as a service (SaaS). Think about how your top sales people or customer reps interact with your clients.  That innate, human intelligence will be built into the CRM system. CRM will make your average employees smarter and faster and make your top performers superstars.

These superstars will likely be able to scale their intellect to a global level never before possible. Think of it as a professor who used to reach 200 students in a lecture hall on campus that now reaches millions by teaching online to students all over the globe. Indeed the best of the CRM sales superstars may go into the business of training algorithms to mimic their human intelligence. Over time, these AI powered CRM systems will develop a higher level of intelligence than the superstars. That’s when things will get most interesting. But that is still some years away.

Don’t Be Left Behind

CRM vendors who do not embed AI into their systems will be left behind. And small businesses, startups and mom and pop shops that do not adapt, will be outpaced by competitors and also left behind.

Here’s what you can do:

• Understand and appreciate the value of AI powered CRM for YOU
• Take a look as the Salesforce Einstein interview from CEO, demo
• Start using a CRM system if you have not already
• Link all of your marketing and sales channels and mashup your data (social media, CRM, Google Analytics, MailChimp, Google Adwords, Paypal, Quickbooks or other payment / accounting systems)
• Start simple by using Excel or Google spreadsheets (use “=CORREL” function to develop simple regression algorithms to better predict sales, top performing keywords)
• Have a consultant or Masters or PhD student in statistics from a local college in your area conduct a pilot AI machine learning project — a low-cost, proof of concept at a cost of a few hundred dollars can help you can see the value and possibilities of AI powered CRM
• Get up to speed on basics of AI and machine learning to start (deep learning if you are truly ambitious)

Conclusion

AI is bigger than the Internet, mobile and social media. And AI will dramatically change the nature of customer service, sales and marketing for all organizations, from small biz and startups to multinational corporations. Consequently, our CRM systems will be most impacted by these changes. You can start by extracting greater value from your CRM software. This can be achieved by using machine learning algorithms or leveraging a vendor’s system that already provides this capability.

Most important, take action today. Things are moving faster than you think and you do not want to be left behind. In fact, you may be able to gain a first mover advantage in our chosen market by becoming an early adopter.

Update on 08 March 2017:

It can’t be emphasized enough how much AI. is affecting us right now.  Recently Zoho has released it’s A.I. component to it’s CRM system known as Zia.  It allows businesses to build a model to help businesses decide which leads are the most likely to convert into sales.  In a highly competitive global market, differentiating yourself from the rest of the competition is key.  The video shown below can show how to quickly find your niche and reach the right customers. Take a look:

In this video we took a look at several of the most recent leads and analyzed which leads are the most likely to continue down the sales pipeline and make a purchase.  We chose for this particular video to analyze the leads data using the Microsoft Azure Machine Learning Studio to determine the probability that each lead would become a sale.

Even with a preliminary model this is a very powerful tool.  This is more proof that artificial intelligence is not only just extraordinary, but necessary to thrive in 2017.