The News
Last week Salesforce announced Einstein Search, an enhancement of the search mechanisms that are already available in its applications. As usual you can read the announcement online or below.
Salesforce wants to release three main issues with Einstein Search:
- The diverse interests and objective of users of enterprise search make it hard to be as good as a consumer search as delivered by Google or Bing, or the other consumer search engines, especially if in an ecommerce environment. In an enterprise setting, objectives can vary between closing a deal or solving a case, or creating new campaigns. This creates hidden complexities. There are no safe assumptions.
- Data is residing in different silos and frequently not linked. Further, there is no one size fits it all as Salesforce as an application normally is customized to suit an individual customer’s needs
- Third, the data simply does not belong to Salesforce, with the consequence that Salesforce cannot look into the data, even not with the objective of improving search. This makes it impossible to use traditional machine learning approaches.
As per now Einstein Search is in a private beta stadium with only a few customers using it. General availability is planned for 2020 but limited to customers on Unlimited, Enterprise, or Performance Edition plans with 150 or more active licenses for the Sales or Service Cloud.
So far the implementation of Einstein Search covers the top 5 searched objects: accounts, opportunities, contacts, cases and leads, but is intended to support further objects.
According to Will Breetz, VP of product management for Einstein Search at Salesforce, Einstein Search is a superset of Salesforce’s available search mechanism. Being implemented using Salesforce’s AI Einstein it delivers personalized result sets based upon the first three of the objects listed above. I also covers search in natural language for all five objects that are in scope although filtering can happen only around ownership, status, location plus a ‘handful of other filters’. According to Will Breetz time is not yet one of them. The search ability shall be improved by GA.
If the confidence about the result is high enough Einstein Search already provides detailed information about the result. This is driven by the underlying prediction model. This model works on the query terms, query independent terms like update recency and personalization signals that are derived from user behaviour.
With the addition of customizable ‘next best actions’ Salesforce claims a reduction of 50 per cent of clicks and page loads, which increases user efficiency and the user experience.
If you want to continue reading the announcement here, read on, if you prefer to go on with my take on it, just scroll down.
If you’re one of the 4.5 billion people connected to the internet today, you use a search engine to find, purchase, or learn about pretty much anything that comes to mind. Consumer search engines provide a seamless way for us to make sense of our complex world. And consumers are used to a search experience that is fast, accurate, and constantly improving. But when those same people try to search within their CRM at work, the experience is painfully underwhelming: too many clicks to find what you’re looking for and an interface that confuses more than it helps. This shouldn’t be the case today. Search should be intelligent and help you quickly find critical information, be more productive, and resolve customer issues faster. That’s why I am so excited to announce the arrival of Einstein Search, which brings the incredible power of intelligent search to CRM by making it personal, natural, and actionable.
The complexities of enterprise search
Enterprise search has lagged behind consumer search for a few key reasons. The first is a diverse user base with a diverse set of goals. When a consumer uses a search bar on an ecommerce site, the intention is universal: they are looking for something to buy. But in an enterprise setting, users have divergent goals that can range from salespeople trying to close deals to service agents solving customer cases and email marketers creating new campaigns.
The second challenge is siloed and dissimilar data. For instance, when customers buy CRM platforms like Salesforce to customize it, they are creating an environment that is completely unique to their business, from the fields they use to the custom objects they create. This means that a search model that might work for one customer will not work for another. At Salesforce, our customers’ data belongs to them, not us. That’s one of our core tenets and why so many companies trust us to run their businesses. This presents challenges for search because we don’t look at customer CRM data, meaning we can’t rely on traditional machine learning techniques.
Announcing Einstein Search
Einstein Search addresses these issues. In building this feature, we had an opportunity to completely rethink search for CRM. It’s already one of the most widely used features in Salesforce with more than a billion searches a month. And with our analysis showing an up to 50% productivity lift, we had an opportunity to fundamentally accelerate customer success at scale for our customers.
Personalized results for every user
Salespeople and service agents rely on Salesforce as their single source of truth for customer information. This is why we made sure that Einstein Search had the ability to return personalized results for every user. Each search result is tailored to what matters at your company and how you work as an individual. For example, if a sales representative covers accounts in the Northeast in the Financial Services vertical, Einstein Search will learn that and show them more of what matters to them. Under the hood, Einstein Search leverages innovative data mining and machine learning techniques to personalize search results, all while keeping specific user information anonymized.
Relevant results from natural language queries
When we type the words “where can I find the nearest coffee shop” into Google, we expect the system to render a list of the coffee shops closest to our current location. Enterprise users expect the same seamless search experience when they use their applications. Einstein Search understands natural language, specifically as it applies to Salesforce. For example, if a sales rep types in “my open opportunities in New York,” Einstein Search interprets that query like a human would. It’s a faster, simpler way to retrieve whole sets of information (like every opportunity with your top account).
An actionable search bar for quicker time to value
Sales and service teams use Salesforce to get work done. Einstein Search increases productivity by not only displaying the most relevant information for each user, but also serves up customizable actions within the search results. For example, instead of searching for a contact, clicking into their record, and then manually attaching the contact to an opportunity, you can take these same actions just by using the enhanced Einstein Search bar. Using Einstein Search can result in an up to 50% reduction in clicks and page loads for the most frequently-used tasks, such as editing sales records.
Customers are finding value right away with Einstein Search
Einstein Search is currently in pilot, and is already providing value for our customers. Brands including iHeartMedia and MightyHive are using Einstein Search to spend less time sifting through data and more time building customer relationships. “MightyHive is a global digital media consultancy with over 300 employees using Salesforce, and we are excited about the new Einstein Search capabilities,” said Laurent Farci, Director of Global CRM & Enterprise Solutions. “It tailors results to individual users, and significantly reduces the number of clicks to provide immediate access to needed information.”
Einstein Search will be generally available next year. This feature will be available to orgs with an Unlimited, Enterprise, or Performance Edition with 150+ active licenses for Sales or Service Cloud.
Interested customers can get early access to the Beta release this winter. Sign up for the pilot here.
The bigger Picture
Enterprise Search is notoriously challenging and so far has been a domain of third party vendors that specialize in linking data. On top of the difficulties that are explicitly mentioned in the press release there are also the topics of authentication and credentials. Not every user is allowed to use or even see the same data. Enterprise Search also supports data sources as diverse as applications and file systems, covering structured as well as unstructured data. These data sources stretch across all business applications and content/document management systems. As such it is important to have enterprise search and the data sources connected to a corporate directory service like e.g. Active Directory.
My PoV and Analysis
With this foray into enterprise search Salesforce broadens the footprint of Einstein. The early references are testament to Salesforce offering a valuable addition to its solutions although it is currently limited to Salesforce applications.
Personally, I am not so sure that the approach of showing results by objects is the right one but this may be a matter of personal preference. On the other hand, nothing prevents me from searching for opportunities with ACME to limit the result set to opportunities. This can be offered via the search box and/or a more old-style drop down that supports the search, preferably the search box, though.
Utilizing user behaviour for personalization is the right approach. It is what we are used to and it is what matters most, even in an enterprise setting, where one could use organizational information in addition. Organizational data, however, would help with people changing roles, and kick in before search behaviour changes because of a role change, be it the result of a promotion or a reorganization of the sales force, or whatever. Still, Salesforce having its core on the side of customer facing interactions I understand that this data might live in another system that Salesforce cannot access.
Given the limitations of its positioning – which does not cover the complete value chain of a company – Salesforce with Einstein Search delivers pretty much the best possible solution. It should deliver better results than the other built-in searches that I have seen so far.
I like what I have seen.
Still, I have two recommendations.
Salesforce should not only cover additional objects but also their relations, to be able to answer questions like ‘ Who do I need to talk to with a question on how to best pursue opportunity xyz at ACME?’ that needs knowledge about contact persons at ACME, own employees with their relationships to these contact persons and also external knowledge. To my understanding the framework to cover at least the company internal data is there with Einstein.
The second suggestion is to look into how the technology that was acquired with Mulesoft can help. Enterprise search is about crossing data silos to combine data to get new and additional insight. Mulesoft can help here.
I am really looking forward to hearing more about Einstein Search.