How Salesforce is using machine learning to unlock more value for its customers (and their customers)

In recent years, machine learning has become a paramount technology for CRM products, and Salesforce is currently working on implementing an innovative machine learning strategy to fuel its future growth.

Salesforce is a software company that was founded by Marc Benioff in 1999. Salesforce’s core business focuses on Customer Relationship Management (CRM) software solutions. Companies, and salespeople in particular, typically rely on CRM tools to help them manage their entire the sales process, from customer sourcing to suggested next steps and pipeline management. In recent years, machine learning has become a paramount technology for CRM products, and Salesforce is currently working on implementing an innovative machine learning strategy to fuel its future growth.

Machine learning is critical to Salesforce’s product offering because it can help salespeople be better at their jobs. Machine learning can solve one of the most difficult questions that salespeople must answer: which potential customers are most likely to buy? Using the vast data sets that Salesforce’s CRM software collects on potential customers, machine learning can identify patterns of customer behavior which are too complex or obscure for a human salesperson to detect. Increasingly, customers will look to Salesforce not only to help them manage their sales process, but also to help them make decisions about which customers to invest in and which ones to de-prioritize. Machine learning can also help salespeople decide what the next step with a customer should be. Should we email the customer? Is a phone call more appropriate? Is an in-person meeting with the customer likely to yield a good outcome? These are decisions that salespeople currently make relying on their instincts, and statistics-based machine learning can significantly reduce the human inefficiencies in this process.

To address this issue, Salesforce’s management team has rolled out machine learning capabilities to their customers via a new analytics platform called “Einstein.” Their machine learning model uses “advanced machine learning, deep learning, predictive analytics, natural language processing and smart data discovery”[1] to help clients better identify good sales leads. Critically, Einstein is custom-tailored to each client’s unique needs and is constantly learning and self-improving. This means that, as time goes on, Einstein will get better at making predictions about the quality of a particular sales lead and recommending best next actions. To date, Salesforce has deployed Einstein to bolster their cloud offerings in Sales, Customer Service, Marketing, and Community Management.[2]

In the short term, Salesforce is focused on improving these offerings for customers and working directly with customers to tailor the Einstein platform to their needs (e.g. they have worked individually with Airbus[3] and Marriott[4] on such initiatives). In the medium term, Salesforce is going to heavily focus on acquisition of AI acquisitions for its future growth. Historically, Salesforce has always leaned more towards acquiring companies to fill gaps in its service offerings to clients, as opposed to developing software to fill those gaps from scratch. According to Keith Block, Salesforce’s recently appointed co-CEO, Salesforce will continue to employ his strategy to boost their capabilities in the machine learning and AI space[5]. In the last year alone, Salesforce acquired Datorama, an Israeli startup offering “an artificial intelligence service to marketers that was integrated into the Salesforce Marketing Cloud.”[6] In the age of big data analytics and machine learning, such acquisitions are particularly valuable because of the volume of high quality customer data that they bring in. By combining the data of newly acquired companies with their own, Salesforce can unlock important synergies and improve the quality of the predictions made by their machine learning algorithms.

As Salesforce looks to improve its machine learning capabilities, I would recommend for it to be careful not to overly emphasize a certain archetypical consumer’s past purchasing history as a predictor for future purchases. Often times, the reason why a potential customer may not decide to buy a product or service is not because they are a bad lead, but because the product does not solve their problem effectively. By overly indexing past purchasing behavior and ignoring customers that have not historically purchased a product, companies are guaranteed to miss out on opportunities for product improvement and innovation. For example, for a company that sells solar-panels, a great predictor of a purchasing customer may be whether the customer has a tilted roof which is amenable for solar-panel installation. However, by overly indexing the type of roof that each potential customer have, the company may miss out on the opportunity to improve their product to also work well on flat roofs.[7]

Open ended question:

  • How could a company like Salesforce ensure that its machine learning algorithms are not further feeding some of the biases that exist in society today? For example, a machine learning algorithm may rate customers of a certain background as “less valuable” or “less likely to purchase” because of their race or where they are from. How can Salesforce counter this issue?

(787 words)

[1] “Introducing Salesforce Einstein–AI For Everyone”. 2018. Salesforce Blog. https://www.salesforce.com/blog/2016/09/introducing-salesforce-einstein.html.

[2] “Introducing Salesforce Einstein–AI For Everyone”. 2018. Salesforce Blog. https://www.salesforce.com/blog/2016/09/introducing-salesforce-einstein.html.

[3] “Salesforce.Com CEO Marc Benioff: Next-Generation AI | Mad Money | CNBC”. 2018. Youtube. https://www.youtube.com/watch?v=TTcru4rzldU.

[4] “Marc Benioff Reveals Siri Integration With Salesforce Einstein, Lauds Apple’s Cook”. 2018. Appleinsider. https://appleinsider.com/articles/18/09/25/marc-benioff-reveals-siri-integration-with-salesforce-einstein-lauds-apples-cook.

[5] “Marc Benioff Reveals Siri Integration With Salesforce Einstein, Lauds Apple’s Cook”. 2018. Appleinsider. https://appleinsider.com/articles/18/09/25/marc-benioff-reveals-siri-integration-with-salesforce-einstein-lauds-apples-cook.

[6] Kovar, Joseph. 2018. “Salesforce Is Counting On AI, Cloud Acquisitions For Future Growth”. CRN. https://www.crn.com/news/cloud/salesforce-is-counting-on-ai-cloud-acquisitions-for-future-growth.

[7] Rosenberg, Scott, Tom Simonite, Shaun Raviv, Clive Thompson, Jessi Hempel, Dean Kuipers, and Nicholas Thompson. 2018. “Inside Salesforce’S Quest To Bring AI To Everyone | Backchannel”. WIRED. https://www.wired.com/story/inside-salesforces-quest-to-bring-artificial-intelligence-to-everyone/.

Previous:

Can artificial intelligence help address the complexities of non-artificial mind?

Next:

Keeping It Fresh in Packaged Food: An Examination of Open Innovation at Mondelez International

1 thought on “How Salesforce is using machine learning to unlock more value for its customers (and their customers)

  1. Thanks for the interesting and insightful read! I fundamentally believe that AI is the next big thing in the CRM space, and there are several high-potential companies, other than Salesforce, that work in this direction like Chorus, Cogito and Conversica. From my perspective, another interesting question would be whether customers are ready for this change and the touted impact that these technologies are introducing to their business. To your question, I am confident that the solution needs to come from the software developer, Salesforce in this particular instance, by, first, building diversity into the datasets used to train the algorithm and, second, assembling an inclusive tech team that is more likely to take such parameters into account and produce algorithm that are more likely to be fit for a more diverse world.

Leave a comment