Salesforce Einstein… What?

Einstein in the Salesforce world has been a thing for a while now and my feeling is the concept has been hard to grasp. There are talks about Einstein in Sales Cloud, Service Cloud, Marketing Cloud and then there is Einstein Analytics. But what does it all mean? Lately, I am addressed and asked questions about “Einstein” a lot more than just 6 months ago. The questions I get is not always focused on Einstein Analytics and I have to specify that there is a difference. I have been curious about Einstein as a concept, which I am sure I am not alone in. So with this blog post, my aim is to de-mystify the Einstein Concept based on my own findings.

Einstein as a concept

Okay, so Einstein is really just a concept. It’s an umbrella that covers a lot of different products and features. When Einstein was launched it comprised of both new and old products and features. For instance, the Einstein lead score was new, but Einstein Predictions in Marketing Cloud had been around for a while.

What products fall into this category? Well, the name of the product is often a good indication, but in reality, it has to do with intelligent products and features that allow you the customer to be smart and informed.

Einstein Hooks

Within all the product there are Einstein elements or we can call them “hooks”. It is features that leverage intelligence and brings it up in the UI. An example is the new lead score – not to be confused with the Pardot score – which looks at all the converted leads in the org, finds trends on what makes a lead convert. Of course, the system is still stupid in the sense that it will look at fields completed and not look at a lead in a bigger context. This is also how the recommendations are generated based on trends in the fields and their values. These “hooks” are leveraging intelligence and pushing it to your users or customers within your existing Salesforce products.

Einstein Analytics

This is a product and not a feature, though one could argue that there are Einstein “hooks” on this platform as well. Yes, Einstein Analytics is a platform with a powerful engine that runs all your analytics queries. Anyway, Einstein Analytics is the Salesforce go-to platform for anything analytics and reporting, which allow your user to quickly navigate data in an easy way to gain knowledge and explore data further. Einstein Analytics is well integrated with the core Salesforce platform as you can easily access data from core Salesforce, embed dashboard on record pages, tabs, and community. It is even possible to interact with Salesforce data directly from the dashboard by enabling actions and bulk actions – allowing you to create tasks, add to campaign or whatever else your creativity may lead you to set up.

Einstein Discovery

This product is also a powerful engine! Take a huge file (preferably more than 10,000 rows) of raw data put it through the engine and Einstein Discovery will crunch the numbers using different statistical calculations and find trends you may or may not be aware of. The cool thing about Einstein Discovery is that all the findings or stories as they are called are narrative explanations, so all stores are described with visuals as well as text, highlighting the interesting bits. You can explore the stories by asking different types of questions like “What happened?”, “What changed over time?”, “Why it happened?”, “What could happen?”, and “How can I improve it?”.

Time to be smart like Einstein?

Hopefully, by reading this you find that the Einstein talk has been de-mystified just a little bit. And maybe you even want to try it out yourself. I will say that those products and features that rely on machine learning do demand a good volume of data, the more data the better since the statistic calculations behind will be better equipped to find accurate trends.

If you want to learn more I can recommend to head over to Trailhead and try out these three modules:

4 thoughts on “Salesforce Einstein… What?”

  • 1
    Steven R. Brown on January 17, 2018 Reply

    Great post Rikke, and great to meet you at Dreamforce. I think it is useful to think of Salesforce Einstein as the full spectrum of AI for CRM, starting with PACKAGED AI (Sales Cloud Einstein, Service Cloud Einstein, Marketing Cloud Einstein, etc.), then CUSTOM AI (Einstein Prediction Builder, Einstein Discovery, Einstein Analytics Platform), and finally PROGRAMMATIC AI (Einstein Platform Services = Einstein Vision, Einstein Language).

    The Einstein spectrum is nicely captured in one slide:

    Specifically the role of Einstein Analytics in this spectrum is two-fold:
    (1) visualization of aggregate Einstein (and non-Einstein) data for embedded interactive exploration (key for operationalizing AI), and
    (2) data prep for Einstein Discovery pulling in Salesforce + non-Salesforce data, combining via recipes, and exporting via the new Export to Einstein Discovery node in a Dataflow (

    Anyone can further demystify Einstein for themselves on Trailhead, and here are 2 handy Trailmixes to get started:

    • 2
      Rikke on January 17, 2018 Reply

      This is really amazing feedback! Thank you for contributing! I am really no expert on all things Einstein, but I hear a lot of confusion about the concept, so your further explanation is gold. I might try to update my blog with some of your comments, I love your view on packaged AI and custom AI it is a great way to look at it.

  • 3
    Roma on May 16, 2018 Reply

    Hi Rikke,

    This is actually an amazing blog and very informative.I have a small question , I am trying to export dataset from Einstein Analytics to Einstein Discovery .
    I was successfully able to to that when I have to make export in same org.
    What if I want to export Dataset of Analytics to Discovery from different org. E.g. from Production to Sandbox?
    How can I do this? Is this achievable? or is it out of scope as of now?

    • 4
      Rikke on May 19, 2018 Reply

      Thanks Roma.
      You can export data from Einstein Analytics to Einstein Discovery using the export node in the Einstein Analytics data flow. You can deploy that data flow using changeset or metadata, however, the actual data will not be deployed.

Leave a Reply

Your email address will not be published. Required fields are marked *