85. Einstein Analytics Dashboard | Carl Brundage

Carl Brundage, who joins to talk about Einstein Analytics Dashboards, is a Salesforce CTA, 30 x certified, and Data & Analytics expert.

Main points

In this conversation, Carl talked about the differences between Salesforce built-in dashboards and Einstein Analytics Dashboards (recently renamed to Tableau CRM), the various phases of utilizing data, and how to implement a successful data analytics solution.

Links

• Carl’s Twitter
• Carl’s LinkedIn
• Carl’s personal blog

Transcript

AI-generated, unedited.

[00:11] Hey everybody! This is Xi Xiao. This is yet a new episode of Salesforce Way podcast. Today I'm sitting with uh again a wonderful guest. He has made remarkable achievement in his career journey. So he's a Salesforce MVP. He's thirty Salesforce certified, and he's a Salesforce certified technical architect, the CTA. So by only mentioning these three achievements, I think there's only one person in the ecosystem did that. His name is Carl Brundage. Hello, Carl. Hi. Thanks for having me today. Appreciate it. Yeah, really honored to have you on the show. Thanks for your time. My pleasure. So in the past, we had a conversation with David Diction on fifty nine episode, if I remember, with Andrew Davis, talking about the Einstein Analytics in general, like the overall picture. Why it's exciting? How was the journey? The history? So today I want you to teach me how to do a little bit more hands on, how to build the dashboard and what the data is really about, how the data is flowing in Salesforce. How does that sound? Sure, it'd be my pleasure too. I've. Spent a lot of time working with analytics and data, especially even before I worked with Salesforce. It was one of the the first things I did after I got my degree from college was working with data warehouses and business intelligence tools, building them

[01:42] from scratch. So what I think I can share are are good tips for thinking about when you're building a dashboard to give to a person. How do you make that be a meaningful dashboard? Yeah, it's pretty easy with Einstein and other tools to throw together a bunch of charts and put something out there. But are you really achieving the goal? How's that sound? Yeah, it's perfect. The thing is that a lot of us, a lot of our listeners, they are technical people, just like me, and we love to play around with tools, play around with the data, with the programming languages. But sometimes we forget about the human side, right? Which is super important. So to build a successful solution. You need to involve people. Absolutely, the aspect of who you're building for and what they're trying to achieve, you're going to see that that's actually the the first step that we think about before we jump in and get our hands dirty or do anything inside the tool. We need to think about our audience before we really dive into the details. Can I ask you some silly basic questions? Sure. Most of us have that in our mind. So in Salesforce, we know we have these dashboards. We have these reports, which is a built-in feature in Salesforce. Everybody is allowed to use in that. So, what's the difference between that and the Einstein Analytics? Sure. So, if we think about it, there's a a big kind of umbrella called Salesforce Analytics. That includes a couple of different product offerings. That's where you'll find

[03:12] the traditional reports and dashboards that we're used to using right inside of Salesforce. Then it has the Einstein Analytics and Einstein Discovery umbrellas. So those are the products that are designed either from an Einstein Analytics perspective to give you apps, and the apps are really targeted towards specific use cases. So you can think just like you have a sales cloud app that you can use to do sales or service cloud. You have Einstein Analytics apps that'll let you look at your sales performance or your service performance or your marketing performance. On top of that, there is a platform. So Einstein Analytics Growth. Which you can think of like the Lightning platform, right? It's basically empty. You can build whatever you want on it. Same thing there. You don't have any pre-built templates or dashboards or anything to get you started. It's a platform that you can use to get data from Salesforce. You can use to get data from external sources. Then you can put it all together and deliver the insights that you want to. On top of that, sorry. And then the one last piece is the Einstein Discovery. That's where you're adding those AI powered insights on top of it. So it can use data from Salesforce, it can use data from external places, but it's going to give you the predictions so that you know, hey, this is what I should do based off of what a case is trending. How can you improve the estimated case satisfaction? So you mentioned there's a platform and there's Einstein Analytics, and there's Einstein Analytics Plus, which is the AI powered features. Yes.

[04:42] Okay, because at the beginning when I heard about Einstein Analytics, I always thought it's the AI, the machine learning behind. So that's only part of the whole platform. Yep, that's part of it—the upper end of the platform, if you will. So if you think about it, the apps are a great place to get started. You. Click on a few questions, give the answers, and it will build you something that may answer a lot of your questions. And then you can tailor it. The platform is that empty piece. Build whatever you want on it, and then the the plus includes the discovery piece that will do the predictions on top of it. So kind of different flavors for different audiences. Wherever you are on your data journey, if you're ready for predictions, great. If you're really just getting started and figuring out, well, how can we answer questions that people are having, then you can start with the apps and grow from there. And there are different license tiers, right? If I want to get the plus, I need to pay a bit more. Yes. So the apps are your least expensive option because they're targeted for something, and then the plus is your most expensive because it has AI as well as much higher limits. You can put 10 billion rows of data. I'll say it again: 10 billion rows of data inside. So lots and lots of data in the platform on the plus side. That's definitely the the volume of data the built in dashboard and the reports cannot achieve at all. Yeah, so if you start thinking about well, if I'm going to use reports and dashboards versus Einstein Analytics, where does that fit in? Like, what's my decision criteria to decide?

[06:13] Hey, I need to jump into Einstein Analytics. And you you hit on one ping, which is which is scale, right? So if you have millions, hundreds, millions of records, your Salesforce report's going to timeout. Like it's just not going to run. We can, from an architect perspective, put all the skinny tables and custom indexes you want in the world, but eventually you're just going to have too much data. So that speed and scale, it definitely works from an Einstein Analytics perspective. The other piece that comes into play is you can bring in data from external sources and combine it with your Salesforce data with Einstein Analytics. So it's pretty easy to digest data from AWS, from an Oracle database that you have on RDS, from Google, all these different places. Could you do it in Salesforce? Well. Kind of. You could probably do external objects, and you could bring it in, and you can report on external objects. But if you wanted to take, you know, three different sources and put it together so that you're looking at a customer, you're looking at their opportunities, you're looking at their orders, and you're looking at their, you know, their satisfaction from a commerce system where they've rated products. I wouldn't really do that in Salesforce, right? I would tell you do it in Einstein Analytics. I see. I see. And still, one more question is about the platform because you mentioned. I try to understand because we have different clouds. Can they all use Einstein Analytics? So I would say most of them. So if we think about the common ones, sales,

[07:43] service, yes, communities, yes. There are marketing apps. Whether you're doing B two B or B two C commerce, there's items for philanthropy, cloud. Where I'm not sure is Commerce Cloud. So B to B commerce, yes, it's on Platform. B to C commerce, I would have to look. I'm not familiar off the top of my head if there's specific apps that will help you ingest the data because that's not stored on Platform, right? B to B commerce is stored somewhere else, so you have to get the data from that cloud into Einstein Analytics in order to be. Because why I'm asking this is I try to understand technically the Einstein platform is it like an isolated platform it fetches the data from the different clouds or is it like sitting on top of the different clouds. So there is a a data ingestion later inside of Einstein Analytics. So you have to take the data from these different places and and bring them in and put them in a format. That Einstein is used to using, so you can think of it as a you know very very flat structure. So you you've removed some of the relational aspects of it, and we're having repeated values, really designed to handle the the big data aspects of processing. So, if you think about how does it get data from Salesforce, well, it's going to use the Bulk API and it's going to ingest the information into the Lightning platform. There are capabilities now where you can do it's called direct query, so you could actually query an object

[09:13] as it sits in Salesforce. Without ingesting it in, but you're not going to have that full feature set yet of all the things that you can do when your data is optimized for reporting purposes. So it's it's coming. I think you'll see. This is just me kind of trying to predict the future. I don't know anything particular, but I think you'll see more and more aspects of Einstein being able to query data where it lives. Whether it be in Salesforce and external systems, instead of having to bring it in, because I think that's the way the world's moving, which is don't take all of your data and replicate it into all these different places. Be able to keep data where it is, process it in in memory, and then give you the results so that you're actually querying and reporting from those external systems without moving data to another place. Indeed, makes great sense to do that. Um. When we talk about obtain or acquire data into Einstein Analytics, I always try to think the ETL tools. Right, that's the pattern I know. So I acquire data, somehow process and manipulate the data, and then somehow feed in to the Salesforce platform so that it can build dashboards, reports for me. Is it similar working way for? Sure, it it is from the standpoint of inside of EA, there is a data manager, and there's actually two flavors ways that you can manage data, and then a new tool that's coming with with the winter release here. So you had data flows.

[10:43] Which is more of a, I'll say, the coding way of doing it. It started without any type of interface, and you were writing JSON to say, "Hey, give me this data from Salesforce," and then take two objects in Salesforce and put them together, and then compute some other fields. So very, very much kind of the procedural language. If you're used to using a tool, and whether it be SQL Server Integration Services, Informatica, Delphimi, that that kind of aspect. Then there came recipes. Which are similar, but a little bit more click-focused, where you could look at your data as you're working with it, get all the common values in a column, see how many times a particular state or country appears, and then do things to clean it up. Well, those experiences are merging with the new data prep tool that comes out in winter, where it's going to let you have the best of both worlds, where it can put it together and you can then take that data with the process flows, like data flows, but then also have the visual aspects to use the recipes in it. So we're kind of excited to see that coming out. So that's one way where if you're taking it from Salesforce, you can bring it in. But there's another way to get data, and I'll talk about in a minute. If you don't have another question, I don't. Please go on. So, if you want to kind of go back to your traditional ETL tool routes, there is a external data API. So there's an API where you can push data into Einstein Analytics. So that's where you could take a ETL tool. I actually did a talk at TDX a few years ago where

[12:14] I was downloading data from these APIs. So these REST services with SQL Server integration services. Building out a set of files and then using a Java program to actually split them up and upload them through the external data API. And the cool part was, I put them on an AWS server that I just spun up when I needed to do my ETL and then turned it off. So it was a very low-cost solution for a particular problem that a customer was having, rather than telling them that they needed to spend a hundred thousand dollars on an ETL tool. So it was a way that yes, you can push data into Einstein Analytics if you aren't wanting to use the built-in tools or the data connectors that they have that go to common sources like Amazon, Google, and so. What is the data connector you just mentioned? So a data connector, you can think of it as a way to get data from common sources that are not Salesforce in without needing to go to that ETL tool. So I'm used to using AWS their their RDS service. So if you have a database in there, you can put in the connection information, and then on schedule, you can have it bring data into Einstein Analytics. Another place is with Azure. So if you're using the Azure SQL, you can put in your connection information. So I have a a customer that. They have they have a data warehouse that they populate every night, and then they want to see certain pieces of that inside of their Einstein Analytics reports. So I didn't have to write anything additional or do any additional ETL.

[13:44] I basically put in a connector to that data source, and then every night it brings in the tables from that data warehouse, and then I can use them inside of Einstein Analytics. So the external data API allows you to push data into Salesforce, and the connector is the other way around. You can pull out from other clouds, other data storages. It's but still into Salesforce. Okay, those connectors are they created by those third party? Vendors or Salesforce or who who are creating Salesforce Salesforce provides them, so it's part of your license to Einstein Analytics, especially if you are in the platform. Perspective: You can pull in external data, so you have access to those connectors, and you can bring in your information. So there's not an additional cost or additional third party. You do have to think about: Well, if it's Azure and I have my SQL, have I opened up connectivity so that the Salesforce IPs can pull in that information? So you know you have a little bit of the networking aspect of it, but you don't have additional tools that you have to purchase. It's also not one of those situations where you're installing a local agent because I'm familiar with that pattern of ETL tools, right? If you want to get to this database running on premise, then we need to run this little local agent that that does the. There's not that type of install. It's really going through exposed ports and IP addresses that you have to Salesforce. So we talked about how to get the data from different data sources.

[15:14] But there's still a long journey going down to visualize the data to the end user. What are the other steps, if you may, brief us? So I think we're if we think about kind of a pyramid of information, right? We're at the bottom right now. We've gotten data. Data is great, but data is terrible because there's so much of it. There's so much noise. You need to derive some information out of it, right? You need to start getting signal to know what you need to need to do, and then you build on it from there, right? You want it once you have information. You're thinking about well, what knowledge can I pull out of this? Like, okay, I know that my customers are happy, and then ultimate the pinnacle is wisdom. I know what to do about it. I know that if I want to have successful customers, I need to respond in cases in ten minutes, and then you can start working from that. So once we have the data in, that's where we can use the different pieces to put it together into data sets. So data sets inside of Einstein Analytics has taken that raw information and you've formulated it some way. Maybe you've taken all of your opportunities, you've added the accounts to it, you've added the contacts, and you've added the data of their last shipment from the ERP system. So you put that together in a dataset. Now you have the information that you want to start using because it's not scattered everywhere. It's ready for you to start looking at. Okay, it's funneled in the way. Absolutely, absolutely. So from there, what you'll do is you'll start. It used to be called lenses. Now it's called queries. You'll start writing queries. And when I say writing, there's a lot of point and click tools where you can go in

[16:44] and you can say, well, let's look at country. Let's look at the average sales size by country and compare it to how it looked like last quarter. So you can start creating those type of insights into answering specific questions in these queries, and this is where you can control the whole aspect of I want a bar chart, I want a timeline, I want a I want a grid of data. However, you want to visualize that information. Because I know SACL query language, right? Is it similar or is it different? It is similar in name. It's called SACL. So there's an A instead of an O. So analytics query language. But it really takes you from I'll say relational queries. I know SOQL is not SQL, but it's it feels very similar to writing SOQL is writing big data type queries, so it's more Apache Pig like from a language perspective that you're thinking about. How do I work with datasets and how do I do groupings and projections? So there's a little bit of a learning curve, for me it was tricky at first because I came from such a SQL or excuse me SQL background. Like I knew SQL in and out like. SQL Server was my my pride and joy. Like I wrote lots and lots of crazy queries and was really good at taking things that maybe took five minutes to run and getting it to run in seconds with with tricking everything out. So I wasn't thinking about sets. I had to start thinking about big data. So it was a little bit awkward at first, but you get the hang of it. There's a lot of good information

[18:15] out there in the community, blog posts, video series that will help you learn it. And the nice thing is you don't have to jump right to So you can use the visual tools. You can build out an analysis, and then if you're in the learning mode, you can switch to the SOQL tab, and you could see the SOQL that the system generates for you automatically. So it's helpful that maybe you don't need to build something right away in SOQL. You can see, all right, if I wanted to do groupings of state and the average amount of revenue, all right, this is what it's going to look like. It's going to load the dataset. It's going to do a filter. It's going to co-group, and then it's going to do a projection and then a filter. All right, now I see what the language looks like. Based off something that I already see the output of. Is it a industry standard, and then this is Saco language follows that, or is it created by Salesforce? It's similar, I would say. It's influenced by. It's not exactly the same. It doesn't always support the same things. I know I've gone and have done some research. Sometimes I'm like, oh, I really want to do this, and I'll find the Apache Pig version of doing it. I'm like, oh no, that didn't work in ZACL. It's not supported yet. And obviously, you know, there's different functions for converting between things that have their own syntax, but. If you refer back between the developer's guide, it's it's fairly easy to get a lot of the things that you want out of it. It's similar to SOQL and the SOQL language, right? So that it's different, but it's from the same same family. Tree. There, there is the same aspect of

[19:45] you're working with data and trying to get information out. You're not going to write a where clause. You're going to use a filter statement. It's very procedural. So we're S A C Q L or S Q L. You're used to kind of writing one statement that does a lot of things. Subqueries, you have where clauses. You know, maybe you go crazy and you write a stored procedure and it it does you know multiple things and it has cursor. In SOQL, you're really kind of going line by line. Like you're going to have four or five lines to do even the basic thing, where you load a data set and you do a grouping, and and that part is, you have to think that okay, this is performing an operation and it's giving me a result, and then I'm using that result in the next line, the next line. So it's very kind of procedural like opposed to. Being set based at the end, you get you get out the visualization that you want. So the query we just discussed—it's about the data manipulation, right? You got the data, you manipulate the data as the. User wants, and we need to still display it to the render it to the user, right? Yes. So we're going to take a lot of those queries, and we're going to put them together, and that's where we're going to build our dashboard, right? Our dashboard is going to be the thing that provides wisdom and insight into our user base. But before we jump in and start just throwing things randomly on that dashboard, that's where I think you you need to think about who are you building this dashboard for. You know, what's the person inside of the organization? Are

[21:15] they an executive that's going to have very high-level goals? They're going to look at overall revenue and then maybe a breakdown by region, or is it that regional manager that needs to think about how all of their districts are doing, or is it a is it a frontline employee that has very tactical goals based off of the opportunities that are sitting in front of them? Because the information and what you're going to measure is really different at that executive level versus the middle versus you know the the front line. So that's where knowing the persona of who you're building, what their goals are, what what may frustrate them, right? If we think about that executive, a lot of times with business intelligence or data warehousing, they'll see a report and it says, "Hey, your sales were down seven percent last quarter." Well, why? You know, there's there's a second question. There's a third question, and in traditional solutions, that often involves a lag. It goes back and forth, right? Oh, well, let me take it to the analyst to answer and give you a breakdown by region because you had that question. Or now we're gonna give it by state. Oh, it's this product in this this country that was the problem last month. So that could be a frustration. A lot of times, if you're not using a solution like Einstein Analytics, the multiple versions of the truth problem starts popping up. Everybody has their own spreadsheet, right? That says, "Oh, this is what I did last quarter," and they put a formula, and this formula is different than the other person's formula, and and everybody's arguing over. Well, no, it was seven, it was versus eight. Well, having it inside of something like EA, you eliminate that single version

[22:45] of the truth. And then the last piece is. A lot of times you see a report and it says, "Hey, sales are down seven percent." Well, so what? What what do I do? How do I take action? So making sure that you've embedded the ability to use that an actionable way, whether it's, "Hey, you need to find more opportunities," or you need to put tasks on here to follow up with your accounts. Whatever it is, you need to be able to take action in order to make these dashboards really important. That sounds kind of really interesting. So, because I I'm not a data engineer, I don't know what was the last generation tools about the data and what's really the frustration. I can sense. So the EA is more intelligent. It can drills down into the details. They ask why multiple times and then help the executives really to. Get the answers that they want. Yep. So if we think about it, we talked about speed. We're now giving you the analytics first approach of having the visualizations there. You can interact with them. And then you can consume it where you where you are. So you're not this other system, which is here's an email of all of your opportunities that's emailed to you, and now you have to go from your email and open up a spreadsheet and then look at it and filter, and then go to your CRM. Right? It's inside of Salesforce, so the actions are embedded right there that you can jump from a list of results into opening up that opportunity or creating a note on that account

[24:16] or creating a task for somebody to. To make a phone call or follow up with an event, so those are I'd say three of the reasons why Einstein Analytics really makes a lot of sense and and where it's a useful use case for it. It reminded me how Salesforce took off. Right at the early days, people store the data and store the CRM systems local. There were multiple copies, and then Salesforce come, and then cloud is the only source. So now I, you know, the EA is also doing the similar stories over there. Yeah. So the the version of if it's not in Salesforce, it doesn't exist. So if it's not in EA, it didn't happen. Yeah. I think I think you're spot on with that insight that it's a very similar pattern of how how you grow and and take. Take over a segment of the market. One more related question is for learning EA. Do we have a dev edition environment for people to register and you know play with all these features? Yeah, absolutely. So you can have access to EA from different dev environments. If you go on Trailhead, there's actually an org that you can start that is pre-populated with some data, as well as has some data flows already set up that you can build in, and you could look at the sales analytics. So you can kind of start from blank if you want, or you can use Trailhead to. To get started a little bit faster, which is really helpful, where you're not trying to figure out well what states looks good because if you start

[25:46] from an empty dev org, you might put in ten records, right? It's more helpful from analytics perspective to have hundreds or thousands of records to kind of see what a pattern looks like or see how things have gone over time. So that's usually where I point people when they are asking about how to get started. You have been so many years in a data engineer path. I still consider it's a data engineer in general. And do you see what's the future on this career journey? Because we have some young listeners there that just you know graduated from school. So do they want to be a programmer? Do they more want be dealing with data? So what what's your experience sharing with them? I think if we look at companies. Each company is at a different point in its data maturity. So there's there's still many many organizations out there that are just working at the fundamental or or base level of, hey, I need these operational reports. I need to know what happened so I can understand how my business is and try and think forward. So there's always going to be a need for kind of the, in my opinion, the core aspects of providing meaningful dashboards. And working with data and doing it. If you kind of go on the other end of the spectrum, there's a lot of companies that are viewing data as a really strategic asset. That when they put the artificial intelligence on it, so the the really hardcore data engineering of doing predictions and creating models and and letting the data guide you into. This is what you need to do to have a better

[27:16] outcome. That's kind of the the the far end where companies are, and I think you're going to see more and more companies have that as a must-have instead of a nice-to-have, and then it starts trickling down the market where more and more companies adopt it. So that's where the the future growth is definitely going to be. So it's a bright future. Well, if you think about it,

[27:38] you know, think about what is Salesforce. So Salesforce, you're using it to interact or know your customers better. Right. I mean, at its core, that's that's why you're doing all these things, and the way that you know customers, it's really through their data, right? It's what they tell you. It's hey, I'm interested in this product. I clicked on this particular part of the website. I downloaded this brochure. I have you know this wearable device that's reporting it. It's all tied into data. So data is kind of the core of how you know a consumer, what that customer experience is like. So that's why I think there's definitely a bright future in the data arena. Thanks, Carl. Before let you go, do you have some other things you still want to mention? I think the the thing I tell people is you know just try it out. You know, think about what you want to do. What KPIs are important for people, and then start building out dashboards that align along with what they want to do. If they're looking at sales, make sure you have something that sales focused and helps them answer those questions. And from there, you know you're going to have a lot of feedback. You're going to have a lot of iterative aspects of working with data, which is really important, right? The thing that you build to start, that's just a starting point. People are going to see it, they're going to react to it, they're going to have feedback. You're going to make revisions. You're going to put something out, and guess what? Well, what people thought they were thinking when you're developing it, it's changed now that it's live and you're working with real data, and you're going to need to make revisions. But that's okay. That's part of it. Is. Being able to iterate quickly and continually to improve—that's how you deliver meaningful analytics to to folks. So

[29:08] I would encourage people to think about that. That you know you're not ever necessarily done. There can always be a next release or a little improvement that adds additional insights for folks. These tips are not just for data engineers, right? For All the IT projects as well. It it can be a little more difficult, I think, in some projects and some things that you're building to iterate as quickly with Einstein Analytics. But it's pretty easy to say, hey, let's grab some opportunities from Salesforce. Let's put together a data set. Let's look at some pie charts. Oh, you want that as a bar chart? A couple clicks and it and it. It changes, right? If you're building an integration between two systems and you're using, you know, a common information model, like the amount of time it takes to respond in that type of project to do something differently, it's a little bit longer. So that's one of the benefits with EA is you can go pretty quick, which I would use that speed to your advantage. For people who want to learn more about EA after listening to our sessions, I know that you have a play by play with Don Robbins, right? Is that a good resource for people to you know to watch? So that's a. There's a lot of good resources out there. So our play by play, we did kind of an overview of Ice Analytics and focusing on. Hey, an executive says comes in and says, "I need to know these things. How do you, as a Salesforce professional, think through what a solution would be in EA, and how do you build out the different different pieces?" If you're looking for topical knowledge, there's a lot

[30:38] of great blogs by community members, Einstein leads champions that are out there on different topics. Whether it's working with with maps, whether it's doing bindings, doing SOQL, you can do all those things. There is an Einstein EdX learning journey map that walks you through different pieces and what you need to know. There's a sample app that you can install that has a lot of different bindings and SOQL in it that you can look at. Trailhead's good, so there's there's no shortage now of places that you could go. You know, when I started, it was literally like. Go to this session in person. We'll have somebody teach you. And oh yeah, by the way, here's the secret page. You have to go and paste JSON if you want to change your dashboard. So when I reflect back on that, like the the progress that's been made in the past four or five years is truly phenomenal. Like there's a lot to get people going very quickly inside of it. So I will make sure we touch some of the most useful links in our show notes. And also, I will put your personal Twitter and the LinkedIn information options in case our listeners want to contact you. Thanks, Carl. Excellent session. Yeah, thank you. Bye, bye.

[31:47] Hi, I'm constantly looking for good guests. If you have any guest recommendation, please reach me out. I'll make sure they're joining to the show to share their knowledge. Otherwise, thanks for listening to the show. I'll see you next Thursday.