All the Ways Salesforce’s Einstein Creates Unexpected Value for CRM Clients

All the Ways Salesforce’s Einstein Creates Unexpected Value for CRM Clients

When we think about Deep Learning and AI models, we usually don’t think about customer relationship management (CRM), as it sounds squishy and marketing centric. However, Salesforce, one of the largest providers of CRM via its cloud-based software-as-a-service (SaaS) offering has found a way to integrate machine learning and AI into its CRM offering to businesses. Since the launch of its CRM AI tool, Salesforce Einstein, in 2016, Salesforce has owned ~20% of the CRM vendor market.[1]

What is Salesforce Einstein?

By integrating AI technology with its SaaS CRM, Salesforce uses the data it collects on every user to power predictive analytics, natural language processing (NLP) capabilities, and machine learning to its customers. By using massive amounts of data from its customers, Salesforce Einstein creates value in four broad areas:

  • Sales – Einstein can tip the odds of winning a customer by predicting which customers are more/less likely to purchase a product with forecasting and opportunity (i.e., lead ‘potential’) scoring.
  • Service – Einstein can automatically organize customer cases and route them to the proper departments to ensure customer’s issues are speedily resolved. Further, Einstein provides smart chatbots to help customers resolve issues without needing to speak to a human.
  • Marketing – Similar to Sales, Einstein improves conversion rates by predicting which customers are more likely to engage with which types of communication by assigning an engagement score.
  • Commerce – Similar to how Google Ads works, Einstein optimizes the relevancy of product assortment (both digital and traditional) to improve the odds a customer goes through with a purchase.

Einstein captures value across these business swim lanes via its four main technical offerings: Machine Learning, NLP, Computer Vision, and Automatic Speech Recognition.

Machine Learning

Einstein provides machine learning in three predictive analytic tools: Discovery, Prediction Builder, and Next Best Action. Discovery is effectively an automatic insight generator: it intakes CRM data from a milieu of sources (e.g., website, email campaigns) and produces customer insights for you, the manager, to track.

Perhaps more interesting however are Einstein’s Prediction Builder and Next Best Action offering. Prediction Builder lets the user define an object they would like to predict (e.g., Customer Contact) set a field they would like to track (e.g., Customer Churn), define the fields the CRM will draw from (e.g., account name; sales in the past year) and then assigns a predictive score, depending on what the user is trying to track. This is a powerful tool for efficiency because it democratizes an analytic that many companies would normally have to outsource, empowering the least technically savvy employees to track complex KPI’s.

Next Best Action goes a step further from letting employees track certain KPI’s by passively running prediction scores in the background, and then recommending courses of action. For example, if Next Best Action thinks a customer is likely to churn, it might tell an employee to reach out to that customer via their preferred contact method (e.g., email) at their preferred time of day; if the employee agrees with the recommendation, Next Best Action would then set up calendar reminders.

NLP

Salesforce uses NLP via two tools: Einstein Language and Bots. Einstein Language infers intent from communication by scanning for key words that connote either positive or negative sentiment (e.g., ecstatic; successful; disappointed) and then amalgamates the inferred sentiment to let the user know how a client might be feeling towards a given product or service. What makes this so impressive and reliable is the scale at which it operates: Einstein can parse email, contact cards, notes, or chatbot conversations to amass this score. Einstein Bots are smart chatbots that respond to frequently asked questions from customers and help keep organizations lean and productive by freeing up users’ time for other tasks.

Computer Vision

Combining capabilities across image classification, object detection, and text recognition, Einstein can assess which objects are in a shot, how many items are in the shot, and parse text within the shot. Practical uses of this are wide: retail organizations can determine inventory restocking cadence; rental businesses can determine the state at which their items are returned, and thus determine how much to charge customers; any organization can take a picture of a business card and have all the germane data uploaded to a CRM database.

Speech Recognition

Einstein Voice gives daily updates to team members on the fly; noteworthy features are alerts for at-risk customers, a quick overview of their daily schedule, and the ability to update a customer’s contact information while on-the-go. Perhaps most impressive, Einstein Voice can analyze information its fed while on-the-go, and then give team members recommended next steps based on the information provided.

Challenges

Even though Salesforce holds such a strong position in the CRM space, there exists the possibility for other tech companies to copy their strategy. For example, Amazon Web Services (AWS) already has a competing CRM software on the market, named Amazon Managed Services (AMS).[2] Like Salesforce, AMS provides cloud-based CRM services, but goes a step further by providing more hands-on service via cloud service delivery managers (CSDM). These CSDM conduct regular meetings to discuss service (e.g., operations, product innovations) and executive tracks (e.g., satisfaction measures, changes in business needs). Furthermore, Amazon is one of the few companies capable of giving Salesforce a good challenge because of its scale and ability to move quickly into adjacent sectors. Despite the threat Amazon poses, Salesforce can maintain its stronghold by focusing on a simplified product: AWS products have a reputation for being effective, but also for being unwieldy for the less technically-savvy.[3] The beauty of Einstein is how hands-off the service team can be with the product because of how simple it is for Salesforce’s customers to use; leaning into this strength would only enhance its competitive advantage.


[1] https://www.gartner.com/en/newsroom/press-releases/2019-06-17-gartner-says-worldwide-customer-experience-and-relati

[2] https://docs.aws.amazon.com/managedservices/latest/userguide/apx-crm.html

[3] https://nandovillalba.medium.com/why-i-think-gcp-is-better-than-aws-ea78f9975bda

Previous:

AI Wars: Lockheed Martin’s Efforts to Weaponize AI

Next:

Plantix: Bringing Crop Science Machine Learning to Millions

Student comments on All the Ways Salesforce’s Einstein Creates Unexpected Value for CRM Clients

  1. Thanks for the very interesting post Jonathan!

    Einstein tool can clearly create a ton of value for companies and sales teams. I find particularly interesting how this kind of tool can fundamentally change the skillset required for account managers and sales persons. Instead of having to hire data scientists who may not be as familiar with sales processes or customer relationships to analyze sales and customer data, Einstein empowers account managers to derive insights and optimize their customer relationships and improve key metrics such as churn rates and customer satisfaction.

  2. Hi Jonathan, very interesting post!

    I also read about Salesforce and I think you touch on very relevant points, it’s impressive all the tools that are leveraged to make companies have better communication with their customers. Regarding your comment about competition, I believe another relevant player here to consider is Microsoft Dynamics, which also makes use of AI/ML to automate marketing tasks, create mkt metrics, and improve customer service. One relevant player Salesforce should keep on the top of its mind.

  3. Such an interesting use case and ML application! Sales seems an excellent use case for behavior prediction—I’m surprised more firms don’t use this (althoughI not Paulina’s reference of Microsoft Dynamics). Another interesting element to this seems like it will be related to adoption – will the “art” of sales accept the data driven, automated approach of AI. I think the more evidence there is to support AI’s successful driving of business results, the more the answer will be yes.

Leave a comment