Why 95% of Salespeople Will be Replaced by AI Within 20 years and Why Microsoft Will Beat Salesforce to It - Part 3 of 3 of the Changing Face of CRM

"If you want to know where to make money over the next two decades, look for companies that are finding ways to automate jobs that are currently being done by humans...that you wouldn't have thought previously could be done by a machine. Truck drivers are one thing and Google as well as Tesla have a great head-start in disrupting that market, but lawyers, doctors, teachers, customer service and sales reps – there are companies that are turning these professions into lines of code, and they're going to make a lot of money."

-Keith Rabois of Khosla Ventures (early investor in PayPal, LinkedIn, Square, YouTube, Yammer, Palantir, Lyft, and Airbnb among others) [note: Khosla Ventures is the most active investor in artificial intelligence startups]

Customer Relationship Management (CRM) software is a roughly $25bn a year market today and Gartner projects that it will be the fastest growing enterprise SaaS segment over the next few years, reaching over $40bn in annual spend in 2019. The importance of this market is being underscored by the all-out war between tech titans Microsoft, Salesforce, and Oracle who have already spent close to $40bn in the past two months on CRM-related acquisitions including LinkedIn ($26.2bn cash), NetSuite ($9.3bn cash), and Demandware ($2.8bn stock). Companies today are striving to leverage what is rapidly approaching the zettabyte scale data loads that customers are uploading to the cloud every year, and most CEOs understand that getting a better customer 360 will be a key driver of their firms' success. Ten years from now looking back, 2016 will be recorded in the annals of technology history as a year of transformation in the CRM industry, and it behooves us as participants in this market and users of CRM systems to understand how what's going on today will affect the future of the customer journey and our roles as sellers.

This is the last of a three part series aiming to provide insights into the current state and the trajectory going forward in the CRM industry. Apple, Google, Microsoft, Amazon, Facebook, IBM, and Salesforce are all betting that artificial intelligence, including smart digital assistants, natural language processing interfaces like chatbots and voice command prompts, and predictive analytics are the future of just about everything. As it pertains to the CRM market, we are witnessing a war for talent and software solutions that promises to produce the next generation of sales, marketing, and customer service automation. We now live in a world where machine learning programs like Google's AlphaGo are beating humans in a game of such complexity that most humans cannot grasp it. The pace of development in the underlying technology powering intelligent software is occurring at a rate that makes me confident saying that within 20 years 95% of salespeople will be replaced with lines of code.

That's a bold claim so let me give some perspective. Uber Technologies was founded 7 years ago and Google started its autonomous vehicle project in earnest only in 2011/2012 despite the first self-driving cars having been produced in the 1980s. Now it's 2016 and Uber just announced that it will be rolling out 100 fully autonomous Volvo XC90s by the end of the year. I remember when I first started using Uber and I used to ask my drivers, "What do you think about the fact that Uber is going to replace you with a computer eventually?" Most of them didn't think that the notion was very credible and would dismiss the question. Just a couple years later it's already starting to happen, and now I find myself in a profession where the same is likely going to occur for my line of work. Uber hit 1 million drivers in 2015 and the Bureau of Labor Statistics estimates that there are around 14.5 million people in the US working in sales (note this includes everything from retail to enterprise SaaS) currently. Cloud computing has driven the cost of compute power down to a marginal level, and the pace of technological advancement is rapidly approaching the moment in time that Ray Kurzweil has dubbed "the Singularity".

I believe that artificial intelligence will replace salespeople even faster than sensors combined with algorithms will replace drivers. Why? How? It's economics. If you look at the regulatory barriers as well as the onerous capital investment required for new vehicles or retrofits in order for Uber to successfully replace all of its drivers and compare that to how cheap and politically easy it will be to replace salespeople (it is one of the most reviled professions after all), it's not hard to see why I believe that 95% of current sales roles will be completely automated by 2036, and probably a lot sooner than that. Uber is aiming to replace the cost of labor in their unit economics equation. No drivers = more profits. Transportation is an important segment of the economy, but looking at gross numbers the sales profession employs many more and as individuals many of them likely earn more than a driver. So, if I were an investor looking for a market to disrupt by automating a profession, sales looks like a much better target than drivers. And alas, investors and entrepreneurs are piling into this market as my last piece noted.

Unfortunately Uber is burning cash at a phenomenal rate because they have to finance vehicles, which is still going to be a major cost even if they transition to fully autonomous vehicles. Yes, Uber's profits will be greater in the long run, but the investments they will have to make in self-driving cars will be enormous and the costs will only be recuperated slowly over the operating life of the vehicles. Salespeople, on the other hand, are one of the most expendable parts of an organization and if a manager had a chance to replace a sales team with robots that cost half as much and were twice as productive, the benefits would be realized immediately with the only need for upfront investment being the purchase of the software itself. This insight also gives us a glimpse into the pricing of the software which will likely be somewhere in the ballpark of the delta between earnings produced and corresponding compensation for a human salesperson vs. the earnings produced and the cost of training and maintaining a sales bot.

You may object that we are very far from technology that will allow for the full automation of salespeople since sales is such a relationship driven profession. Let's return to the Google AlphaGo example - a machine just beat a human at one of the most complex games ever developed by humans. Most successful sales deals are built on fairly replicable and easy to automate patterns of interactions and relationship building. On top of that, more and more of the buying process is being automated as product services information, reviews, and trust networks like blockchain technology proliferate. Indeed, McKinsey estimates that by 2020 customers will manage 85% of their relationship with an enterprise without interacting with a human. In another research piece they estimate that in the retail industry, 53% of salespeople's activities are automatable. Most people I know, including myself, do not enjoy interacting with salespeople because most salespeople are bad at what they do and are clearly self-interested. My hypothesis is that machine learning algorithms and personalized digital assistants, particularly with the coming ubiquity of augmented reality, will replace almost all salespeople in a relatively short amount of time. If you're still having difficulty believing me, I present the following video of China's Dragon TV using a weather reporter powered by Microsoft's Xiaoice, a deep learning chatbot used by over 40 million people.

So who's going to build the winning platform in the next great war to revolutionize the customer journey? I badly want to go broad here and look at the four companies that I believe are inventing the future of sales, marketing, and customer service which are Google, Microsoft, Facebook, and Salesforce. However, that would lead to a certain digression from the focus of this series which has been the CRM market, so I want to keep this piece more narrow and look just at Microsoft and Salesforce since Google and Facebook are not players in the CRM market at this time. Sorry to disappoint since I truly admire and respect what Google and Facebook are doing, but as they will be duking it out for advertising revenue and I don't sell ads I must leave you with a promise to revisit their efforts in a future post.

Why AI is the Future of CRM

Dreamforce is right around the corner and Forbes just broke a story on what Benioff will be pushing as the future of the CRM industry: Salesforce Einstein, a layer of artificial intelligence woven into every aspect of interaction with customers and customer record keeping so your salespeople can say just the right thing at just the right time to make the most profitable sales possible and push revenues sky high. That's all well and good, but the thing is that Satya Nadella has been talking about this concept for over two years. Satya refers to this concept of embedding artificial intelligence into products as a kind of substrate platform as "ambient intelligence" and Microsoft already has products on the market, Xiaoice being a perfect example, that represent some of the most advanced implementations of artificial intelligence in the market today. Azure Cognitive Services is another example and if you want to be really impressed, and maybe a little scared about what's possible with the power of artificial intelligence, I encourage you to take a couple of minutes to watch the following video which tracks a day in the life of Saqib Sheikh, a Microsoft software engineer in London who lost his sight at age 7.

There's a good reason that both Satya and Benioff are painting the vision of how AI will revolutionize sales and customer service: most companies are stuck in the reality of the 80/20 rule and want to escape. That is, most sales organizations have eagles who represent about 20% of the workforce and produce 80% of the organization's revenue. The main logic behind the investment that companies make in CRM systems is to increase revenues by harvesting insights from customer data and more effectively managing their sales organization to continuing growing. Unfortunately, very few companies realize the return on their investments because most salespeople are lazy and usually the best salespeople don't rely on their CRM systems for maintaining their customer records, but rather use an Excel spreadsheet and their gut instincts. Salesforce is notorious for having abysmal adoption rates and it's because most salespeople view entering data into CRM as antithetical to what they feel their activities should be focused on: pitching, relationship building, and closing business.

So now both companies will offer up claims about how their platforms are leveraging the most cutting edge applications of deep learning to arrive at what I call "AI Sales Nirvana" - a mostly automated sales organization that harvests customer data through interactions with bots and other communications tools, runs that data through a magic algorithm that will guide near perfect customer outreach strategies, and create a virtuous feedback loop of ever growing sales with ever less effort. Note, I think that this is the end goal which is still a decade or two out. What I want to focus on here is what is actually real today and what the roadmap for the next 3-5 years looks like.

Salesforce's Einstein Strategy: Weave Acquisitions into Core Clouds

First let's look at the capabilities and products built through Salesforce's recent acquisitions in artificial intelligence. Salesforce has purchased a total of six pure AI companies including RelateIQ, TempoAI, MinHash, PredictionIO, MetaMind, and Implisit Insights. RelateIQ was the first, largest, and most significant of these. Salesforce bought RelateIQ in July of 2014 for $390m. Over the past two years Benioff gradually gave more authority and power to RelateIQ founder Steve Laughlin and to date the product that Laughlin has been working on, now branded SalesforceIQ, is the only real tangible and customer ready application of AI that Salesforce offers. It is very telling to note that Laughlin is one of a bevy of recent executive departures from Salesforce including Jim Steele, former Chief Revenue Officer, Tod Nielsen, head of Platform, and Adam Bosworth, head of their Internet-of-Things cloud. Laughlin's exit was only announced last week with the release of the Forbes piece on Einstein, and Salesforce made a statement saying that Laughlin left in order to pursue an opportunity in venture capital and that he will be replaced with Richard Socher, the founder of another AI startup that Salesforce recently acquired called MetaMind. Socher has strong AI cred with a Ph.D. from Stanford in Computer Science focusing on deep learning, natural language processing, and computer vision and it's worth noting that among other awards, he was granted a Microsoft Research Ph.D. fellowship.

Based on these acquisitions it's hard to say how Benioff is going to cobble them together and what Einstein will look like. Most of these companies were acquired while still in their infancy, apart from RelateIQ. MinHash had developed an AI platform as well as a smart assistant called AILA to help marketers develop campaigns based on patterns its algorithms would recognize in order to make marketing spend more efficient. Tempo AI had developed a smart calendar tool, PredictionIO had built an open source machine learning database, Implisit Insights' product scans emails to ensure CRM data is accurate and help predict when buyers are ready to close a deal. MetaMind, the most impressive of all of them in my opinion, and apparently in Benioff's as well, had much greater ambitions of creating a deep learning program that could answer questions related to a selection of text and images in a manner that closely approximated a human response using multiple layers of logic and inference.

My bet is that Salesforce Einstein will be an attempt to weave all of these disparate solutions together in order to back up all core sales, marketing, and customer service activities with a layer of, to borrow a term from Satya Nadella, "ambient intelligence". This will likely include tools to do everything from analyzing customer sentiment in email and phone interactions so salespeople and customer service agents can know how their clients are feeling and how they react to certain questions or prompts. Marketers will likely get more insights and tools for making campaigns effective by targeting people at the right time with the right message based on that user's unique preferences and habits. I'd bet that sales reps will get tools for smart scheduling as well as a personal assistant to help them log data into CRM, remind them to follow up with leads, and feed them scripts and other best practices to help move sales processes along. I look forward to seeing what they will release and with the team that they've built I don't doubt that Salesforce will offer some compelling products that will lay out a solid vision for the future of true sales force automation.

That being said, I am confident that Salesforce will not release any amazing new capabilities that Microsoft can't match, and as is often the case with Salesforce and Benioff, there will be a lot of hype around a suite of features and capabilities that are overpriced for the incremental gains that they offer. Let me qualify that statement with a review of Microsoft's accomplishments in the field artificial intelligence, some background on recent acquisitions, and an overview of artificial intelligence features that are already available in Dynamics CRM and Azure.

Microsoft's Intelligent CRM Trump Card: Microsoft Research + Azure + a Treasure Trove of Data

I want to start this comparison by first giving you some perspective using scale as context: Microsoft is a 100,000+ person company with $93.6bn in annual revenue, with a net profit of $12.2bn. Salesforce is roughly 20,000 employees and with $6.67bn in revenue in 2015, Microsoft's net income was nearly twice as much as Salesforce's revenue. Microsoft spends $12bn per year on research and development, again also almost twice Salesforce's revenue. Salesforce's total spending on acquisitions and internal research and development in the deep learning field is likely under $1bn and their roster of talent is likely under a few hundred people. And to reiterate, apart from SalesforceIQ, they currently have not actually deployed any true deep learning technology in any of their core cloud products.

Microsoft started its deep learning research at the end of 2009 and Microsoft's Chief Scientist of AI, Dr. Li Deng, has a resume that make both Richard Socher and Steve Laughlin look like complete neophytes in the field of AI. According to his LinkedIn profile, Deng has 60 patents, has written 4 books on deep learning in addition to 400 technical papers, received the 2015 IEEE SPS Technical Achievement Award for "Outstanding Contributions to Deep Learning and to Automatic Speech Recognition," and a few related best paper awards in prior years. He was Editor-In-Chief of IEEE Signal Processing Magazine and of IEE/ACM Transactions on Audio, Speech, and Language Processing. You may not have an hour to spare, but if you do and you're passionate about deep learning, I highly encourage you to watch the talk below in which Dr. Deng gives an overview of Microsoft Research's work in the field.

Microsoft Research is one of the most impressive research institutions in the world, founded in 1991 and currently employing over 1,000 researchers in computer science, deep learning, computer vision, mathematics, physics, and engineering with multiple individuals who have various awards including MacArthur Genius grants, Turing Awards, Fields Medals, and the Dijkstra prize. Microsoft has already released 7 products to the general public that leverage work in artificial intelligence produced by Microsoft Research, four of which are already woven into Dynamics CRM. These seven products are: Opportunity Scoring for Dynamics CRM, Azure Machine Learning, Azure Cognitive Services, Cortana, Skype Translator, the Computational Network Toolkit, and Microsoft Pix.

These services are already powering increased revenue for customers of Dynamics CRM. Opportunity Scoring for Dynamics CRM uses a machine learning model to evaluate historical sales data and patterns and computes the probability that a given opportunity will close - you don't even have to wait a month until Dreamforce to have deep learning embedded in your CRM. Beyond that, Dynamics CRM also already has capabilities like machine learning driven cross- and up-sell recommendations as well as a voice controlled input and lead management commands with Cortana which you can check out in this video.

I repeat, you don't have to wait another month in order to take advantage of this. Microsoft has already developed this and they are extending their lead over Salesforce in applying artificial intelligence to CRM by the day. One of the most important things to note here is that machine learning algorithms are only as good as the data that you feed them. Salesforce simply does not compare to the amount of data that Microsoft has. Salesforce's data set is limited to what sits on their servers that their customers have entered on their customers. Microsoft has data from billions of devices running Windows, hundreds of millions of users of Bing Search, hundreds of billions of interactions with Chinese chatbot Xiaoice and soon to be many hundreds more chatbots, and most importantly for CRM, they now have data from over 450 million LinkedIn profiles. Beyond their internally developed artificial intelligence capabilities, Microsoft has also been acquiring technology companies in the space. This year alone Microsoft acquired SwiftKey, a maker of an AI powered keyboard that predicts what to type, as well as Wand Labs, a developer of AI powered chatbot  and customer service technologies, and Genee, an AI powered smart scheduling assistant.

With Microsoft's powerful AI capabilities, they have the ability to develop solutions that no one else in the market is able to provide currently. Apart from that, with Microsoft's mountain of cash as compared to Saleforce's almost completely barren coffers and highly overvalued stock, Microsoft stands in a position to continue scooping up the most promising AI companies for years to come. And they're just getting started. Even if I didn't work for Microsoft, my money would be on them for the long-run. The only thing that makes me nervous as a CRM seller at Microsoft is the potential acquisition of Salesforce by Amazon, which for all I know may end up happening in the not too distant future. That would make for a very exciting next decade in the CRM industry.

*Note: while I am an employee of Microsoft the views expressed in this LinkedIn Long-Form post are entirely my own and do not in any way represent the views of the company. If you enjoyed this article, despite how long it is, please give a like, leave a comment, and feel free to share. Also stay tuned for more posts on AI, sales, and technology broadly.