May the Salesforce be With You: A Rising Power in Machine Learning implementation

CRM & Salesforce Background

Over the past decade, customer relationship management (CRM) software has exploded as managers and sales teams have recognized the benefits of a dedicated CRM platform (vs. a tool such as excel) that assists sales teams in developing and managing prospect lists and potential customer pipelines [1]. For the past decade, Salesforce.com (Salesforce) has been the 800-pound gorilla in the CRM market with a current market capitalization of ~$100 billion [2], serving 90% of the Fortune 100 [3] and more market share than its next four competitors combined [4].

Machine Learning in CRM

CRM software stores a wide variety of data about interactions with current and potential customers including contact information, personal details, email communication, call logs, meeting notes, purchasing history, etc. CRM providers of scale can use machine learning (ML) to generate insights from the vast amounts of data stored across thousands of customers such as, who in an organization may be more receptive to a cold call, what additional public information is known about a contact at a potential client, how to prioritize sales leads based on the calculated likelihood of a sale, which customer service representative would be best to deal with a customer given his or her concerns, etc.

Given the opportunity of ML in the CRM space and how much data Salesforce has access to, it is therefore surprising that until recently, Salesforce did not have an ML offering and was technologically behind a number of new competitors such as Zoho and InsideSales. Competitors were starting to offer predictive analytics, powered by AI, to optimize efficiency in lead generation and conversion. Salesforce responded by investing nearly $1 billion to acquire a number of AI start-ups to bolster its technology capabilities and talent ranks and in late 2016 the company launched its first predictive analytics product – Einstein [5].

Einstein Overview

Einstein was launched with much fanfare but to mixed reviews as some customers came away unimpressed by its early features. Wall Street Analyst research predicted that it would take time to catch up to the more experienced competitors in the field [5]. Since launch, the algorithms supporting Einstein have improved from data across Salesforce’s 100,000+ customer base. Additionally, Salesforce recently partnered with IBM Watson to bring new insights, giving clients access to Watson’s existing information sources and its ability to analyze their data. For example, IBM weather predictions could allow a car insurance client to communicate with its customers that a hailstorm is likely in a given area.

Salesforce is currently working to add new capabilities to Einstein. Two functions currently in beta are Einstein Intent and Einstein Sentiment. They use natural language processing to classify whether the text of a message is emotionally positive or negative and determine what the intent of the message is, respectively. For example, a company could put to use Einstein Sentiment to classify the tone of their inbound customer emails and accordingly identify positive brand evangelists and escalate dissatisfied customer responses into service cases. Einstein Intent, on the other hand, can be used to augment Einstein Sentiment. It could classify each negative response and identify the source of customer dissatisfaction, like lost shipments and returned orders [4].

Currently, Einstein makes over 3 billion predictions and insights every day [6] and has emerged as an ML leader not only in the CRM space, but in the broader tech industry. A recent Deloitte survey of 1,100 IT business line professionals found that more companies are turning to Salesforce for ML implantation than Google or Amazon [7][8].

Recommendation & Conclusion

While Salesforce has become a machine learning market leader in the CRM software market, competitors are investing heavily in their own ML technology and at least two competitors have formed a data-sharing partnership, Microsoft and Adobe [4]. Salesforce may find it beneficial to find an additional partner with complementary data. A partner such as Facebook could help Salesforce drive more accurate insights into the purchasing behavior of certain types of individuals and further its lead in the amount of data it has access to.

In the arms race for data however, are their diminishing marginal returns? If so, at what point does that become manifest? If not, will competitors ever be able to catch up to Salesforce?

Word count: 708

 

  1. Marks, Gene. “On CRM: Can Einstein Help Salesforce Solve This Chronic CRM Problem?”. Forbes, 2018, https://www.forbes.com/sites/quickerbettertech/2018/10/09/on-crm-can-einstein-help-salesforce-compete-solve-this-chronic-crm-problem/#7d8be092333b. Accessed 14 Nov 2018.
  2. CapIQ accessed 11/12/2018
  3. Com, 2018, http://www.annualreports.com/HostedData/AnnualReports/PDF/NYSE_CRM_2017.pdf. Accessed 14 Nov 2018.
  4. Walker, Jon. “CRM Artificial Intelligence Trends Across Salesforce, Oracle, SAP, And More”. Techemergence, 2018, https://www.techemergence.com/crm-artificial-intelligence-trends-across-salesforce-oracle-sap/. Accessed 14 Nov 2018.
  5. Kim, Eugene. “Salesforce’s Big New Product ‘Einstein’ Receives Mixed Reviews Despite All The Hype”. Business Insider, 2018, https://www.businessinsider.com/salesforce-einstein-mixed-reviews-despite-hype-2016-10. Accessed 14 Nov 2018.
  6. 2018, https://seekingalpha.com/article/4202977-salesforce-com-inc-crm-ceo-marc-benioff-q2-2019-results-earnings-call-transcript?part=single. Accessed 14 Nov 2018.
  7. “A New Survey Suggests Salesforce And SAP Have An Early Lead Over Amazon And Google In The Next Frontier In Tech”. Business Insider, 2018, https://www.businessinsider.com/deloitte-survey-early-ai-adopters-aws-google-cloud-enterprise-software-2018-10. Accessed 14 Nov 2018.
  8. “State Of AI In The Enterprise, 2Nd Edition”. Deloitte Insights, 2018, https://www2.deloitte.com/insights/us/en/focus/cognitive-technologies/state-of-ai-and-intelligent-automation-in-business-survey.html?id=us:2el:3pr:4di4780:5awa:6di:MMDDYY:&pkid=1005631#endnote-sup-40. Accessed 14 Nov 2018.

Previous:

ETSY – “Keeping Commerce Human” through Machine Learning

Next:

“Thank You for the Music”: Spotify and algorithmic music curation

Student comments on May the Salesforce be With You: A Rising Power in Machine Learning implementation

  1. Very interesting post! Thank you for sharing. While this seems like a great application of machine learning, I think there are definitely diminishing returns, as you allude to in your last paragraph. One potential downfall of using Machine Learning to predict “who in an organization may be more receptive to a cold call, what additional public information is known about a contact at a potential client, how to prioritize sales leads based on the calculated likelihood of a sale, which customer service representative would be best to deal with a customer given his or her concerns, etc” is that you may overlook those who do not interact strongly with your data collection system. As a consumer of salesforce’s products, I would want to see hard data on the benefits of their Machine Learning offerings by industry.

  2. Nice article, Yoda. I would think there are definitely diminishing returns with more data. That said, I think we are still far from the point of diminishing returns. There’s an innovative new company called Gong [1] that is using ML to listen to sales rep calls and provide insight to sales reps about how to close the deal. By using the data of actual conversations to create insight, the company is really making an impact to improve sales force effectiveness. I think Salesforce has a way to go before they hit any point of diminishing returns, and the competitors may beat them by being more innovative rather than by having more data.

    [1] https://www.forbes.com/sites/gilpress/2016/06/21/artificial-intelligence-discovers-how-to-improve-sales-performance/#6abbcdb97093

  3. The fact that Einstein had negative reviews in the beginning brings about a big vulnerability and risk of AI and ML systems. These systems, in the nascent stage can throw out some unintended results, and these results can create a consumer sentiment that can have long lasting negative effects to the brand value of the product. On the other hand, AI / ML based systems get better with more data being fed to them, and rolling out to several clients enhances the capabilities. Salesforce should figure out this fine balance to ensure a solid brand perception for the product.

Leave a comment