CloudJungle's Steve Mursell talks through using Einstein Analytics for both business & fun by demonstrating how to find the best European cities to dine in.
In this informative and fun video, Steve Mursell, owner of CloudJungle, walks viewers through the process of using Einstein Analytics to find the best places in Europe to eat. Viewers will learn how this system can quickly sift through TripAdvisor reviews of 30,000 European restaurants to extract important insights based on factors like city, food type, and price. They’ll then see how Einstein Analytics provides rankings of restaurants based on specific details that might be of interest to them.
Viewers can watch how Mursell uploads the dataset to the system, and then listen as he discusses which factors seem to have the most impact on the outcome. Viewers should feel free to then make their own predictions about which restaurants will be recommended based on what they’ve learned so far so they can make sure they understand how it works.
Of course, finding the best restaurants in Europe isn't the only way to use Einstein Analytics. Mursell will provide examples of how else professionals can use this system. Those who might be interested in this video range from Salesforce marketers and salespeople to customer service agents and business owners in any field who are looking for a simple way to make use of data.
[00:00:00] Hello, and welcome to another Xforce Data Summit presentation. This one's a real interesting one. It's by Steve Mursell, who’s the CEO of CloudJungle. And basically, it's about using Einstein Analytics to discover the best food in Europe, which is something we'll all want to do after this quarantine is over.
[00:00:37] So here's Steve. Thank you then. It's, yeah, so great introduction. Thank you very much. So, yeah, so if you're listening to this, I guess it's as a prerecording and so something must've gone drastically wrong on the day, but I've just got my Twitter handle here @junglehq. So, I'll make sure that I'm around at the time of the recording.
[00:00:57]and so any questions, you could just tweet @junglehq. So, a little [00:01:01] bit more about me. So yeah, I'm a CEO/MD—as we call them in the UK—of CloudJungle. And we're Salesforce consulting partners. And my journey with Salesforce began in 2006 as a customer. I started a business doing Blackberry rollouts, if you remember them.
[00:01:19] and I used Salesforce for my sales service at the time, and it completely transformed our business. So, we were a cloud-first business in 2006. It was pretty successful and was sold in 2014. At that point I was then unemployed, thinking what do I do next? And a few of my customers said help us out with Salesforce.
[00:01:44] So we did. And then, yeah, so then I went off and got my certs when I realized that I didn't know too much about Salesforce, even though I'd used it with my previous business. And I wanted to do some contracting, and believe it or not, use Salesforce to generate applications to run power stations or things.
[00:02:02]and then before that, I decided to start my consultancy firm with a telecom [00:02:06] specialty that I had previously. And in terms of Salesforce products, we're also developing a product specialty in Einstein Analytics here today. I'm a member of the Salesforce Einstein Analytics Champions program.
[00:02:24] I think there's about 110 or 120 of us globally. And, yeah, we're all advocates for the platform. Oh, and as I said before, any questions you can tweet me @junglehq. So, I promise not too many more slides, and we're going to spend more time actually diving into Einstein Analytics rather than looking at PowerPoint.
[00:02:48] And so what I've done is, what we're going to do right now, is we're going to look at an Excel spreadsheet, and we're going to prepare it. And we're going to upload that into Salesforce. And that spreadsheet is 30,000 of the top restaurants in Europe by number of reviews. And we're going to analyze that to see where we
[00:03:09] spend our next vacation, whenever that may [00:03:11] be. So what we're gonna do is when we upload that, it's going to give us some insights into the data and can help us find where to go. And also the factors and the variables that influence the success of a, of a restaurant’s, score on God, my mind’s gone blank.
[00:03:29] What, what is it we're looking at? TripAdvisor scores, that's it. So whatever influences the TripAdvisor score, then, we can look at the variables that go into that. So, we'll also look at how you can actually get your hands on some Einstein Analytics with a demo org. And then obviously you can plan your next European food break, as well.
[00:03:51] So a little bit about our dataset. 30,000 of them are of the top reviews. Okay. So we want to know where to go. Okay. Which city? What our budget is…that’s another variable. And what type of food we might like. Okay. And our goal is to try and find somewhere with the best TripAdvisor scores.
[00:04:11] So that is our outcome when we look at this dataset. So then the variables are [00:04:16] city, price bracket, food type, and number of reviews. Okay. And then we basically give that to Einstein Analytics and they can then be our team of data scientists from there on. Okay. So without further ado, let's jump into the dataset.
[00:04:33] Okay. And what we're going to go through now is a little bit of prep of the data. And so we can see what we get. So this is the raw data downloaded from TripAdvisor. And I'll talk to you a little bit about that in a minute. So, so first of all, we've got the, the name of the restaurants, and there's 30,000 of these.
[00:04:51] We don't need those. Okay. Because we're not going down to specific restaurants, but what we do want to know is the city. Okay. We've got the number of reviews that the restaurants have had. Okay. The cuisine style. So what type of food it’s going to be, and what I've done with this, so rather than just upload this, I've gone and
[00:05:13] looked at, just a little formula to see if that contains, American, then true. If not, false basically. But what I'm going to do [00:05:21] is, and also show you the rating. It's very important. This is the rating of the restaurant. Okay. The ranking over here. So this is like the, the TripAdvisor ranking in the city for that restaurant.
[00:05:34] So this restaurant was ranked number 60. Okay. Now, we're actually going to load this ranking data in there. It's going to be full stated, but I'm going to show you how analytics actually handles that, because obviously if it's ranked very high in the city, it's going to have a good rating.
[00:05:51] So these two could be very much correlated. But we'll see what Einstein does with that. Then we've got the price bracket. So again, this is raw download from TripAdvisor. And so what we've done with that is I've changed it around. My columns were a bit wide.
[00:06:13] Okay. I've changed this to high, medium or low, because it makes a bit more sense than having the dollars in there. Okay. Now I've actually done a little bit of run through on this once before. So I know that some of this data is not going to give us very significant information. So what [00:06:29] I'm going to do is I'm going to prep this data somewhere, delete other than American.
[00:06:34] And I'm going to delete, French seafood and Italian, but I'm gonna leave vegan in there. So it may be a bit of a clue to the outcome. The reviews, this is interesting. Actually, I'm gonna delete this. However, in a future version of Einstein analytics, I think later this year, it's also gonna look at sentiments of, of comments.
[00:06:55] And so it will be able to look at reviews and look at some of the language that's being used, and generate a sentiment score. That feature is not available yet. So I'm gonna delete my, my, reviews. I've got my price range as high, medium and low rather than this dollar thing. So we can take out that.
[00:07:12] And I'm going to take out cuisine styles because I just used that to generate the other, the other information. The reason for sort of leaving that in was to sort of say that this is often an iterative process in terms of analyzing your data. You, as, as, as business people will have a lot of information and you know your business very well.
[00:07:32] So you should be extrapolating what information you want to give [00:07:34] to Einstein, because obviously it's only going to be as good as the information you give it. And this is sort of demonstrates the fact that we've been through a few iterations. We’ve created a few discovery stories, and we found out sort of the, the things that really, really matter.
[00:07:49] Okay, so I'm going to save this now. So we're going to save this file as a Xforce Eurofood Live, we call it. So just to say that this is in real time. Okay. And then continue. Okay. So that's, that's my…now my dataset prepped and saved. So what I now need to do is stick this into Einstein Analytics.
[00:08:14] Okay. So, so this is, Analytics. So for those of you that don't know, it's a, it's a platform that is different from Salesforce. it was an acquisition of Salesforce a couple of years ago, but it very much is embedded right in the heart of it. Now they've already integrated it very well. And then, so then to get a dataset into Salesforce,
[00:08:37] I go to my, my [00:08:38] data manager. And basically here, I'll just share on this connector here, we've actually connected to our local Salesforce data. We could actually connect to any other external system as well. And I think, Xplenty knows all about these various integrations. And I'm now going to create a dataset, which I'm going to upload something.
[00:08:55] I select my file, which is the Xforce Eurofood Live file, and then start the upload process. Okay. So during this process, it's looking at what columns we want to import, and basically I'm just going to accept the defaults and upload that file. So now here we go. Einstein is basically sucking all the data, the 30,000 rows.
[00:09:18] And, and he's going to start doing his analysis in the background. Okay. It'll take him about a minute. So I'm just gonna flick to back to my slides. And, I'm just gonna talk a little bit more about the, about the model. So, if you remember, what we want to do is we want to find the best place to go and eat.
[00:09:38:] So, so we need to define our outcome, right? So that, so that is going to be the TripAdvisor score. Yeah. And in order to [00:09:48] do that, we've got some variables that we need to model. Okay. and so, so it's very much to define our outcome and then define our variables. So, that's what we've just done when we've been sort of, selecting, selecting the columns to plan as they say, what, what are we going to model?
[00:10:05] So let’s flip back to Analytics to see if he's done his thing, and great. So now I've got my dataset loaded in, and now I'm going to create my story. Okay, so this is the analysis bit that's coming up and it's very easy, it’s a simple wizard. So this is the outcome variable that I want to measure. And it's not ranking.
[00:10:26] It is rating. Yeah. Remember ranking is, is how that restaurant performs in its various city. So I want to maximize the rating. I might expect some numbers greater than zero. Yes, I am. Hopefully, give it a name and then, and then select what type of story we can do a quick, dirty story, but we're not gonna do that.
[00:10:44] We'll go do our complete analysis here. Okay. So, so, but we are going to take the [00:10:00] automated version, and by the complete analysis, it's gonna help us. So down the line, we will be able to deploy this model to Salesforce. And so if you've got a record which you've modeled the data to predict outcome, then it'll give you your potential score for that outcome.
[00:11:09] So let's say for example, you're a subscription business and you've got a customer. And you want to see what the chance of them canceling their subscription is, or their attrition rate, you could actually do that by modeling customers that have and haven't attrited previously, and look all those variables that influenced that.
[00:11:30] and then that will give you a predictive score. So Einstein's crunching the numbers now. We’ll leave him to it for another minute or two and go back to our slides. And this is what we hope to see when he's finished his work. So, so we're gonna have a descriptive insight and this is like what has happened.
[00:11:49] So it's basically doing that BI basic analysis of our data. You know, what, what, what can we see from that data? Right. And you can pretty much [00:11:00] do this with any, any BI tool out there, right? Then we're going to have a descriptive analysis key. What were the variables that influenced the outcome? Yeah. What the positive factors and what were the negative factors for us to come to that conclusion?
[00:12:12] Then we could do some diagnostics so we can actually compare things. So for example, we could compare two different cities and see what the differences are between those cities. So in terms of the outcome, this score. Why is one city better than another, for example? And then we could do some predictions, and this was this predictive bit that we set up whereby we can actually just decide on like, yeah, I want to go to Athens, and I've got a medium budget and I
[00:12:37] want to go to a vegan restaurant. And then that will give us a predictive outcome score for going to that, to that restaurant. So spin back now to Einstein. He's done it. Fantastic. So, right. So, what has happened? So, what it does here, it's at the very top of the screen, there's a whole bunch of [00:12:57] analysis down here.
[00:12:59] It's picked up the biggest factor. And it says that ranking of the restaurant explains the biggest variation. So the highest scoring restaurants are the highest ranked ones. That is not surprising. Is it? So what we've done is we put a whole bunch of data into our system, which is pretty meaningless because ranking and rating are very, very similar.
[00:13:23] So what Einstein should have done is, he's making some recommendations for us to look at as well. And he said, hey look, well, ranking is our strongest predictor. And so what we're gonna do is we're gonna ignore ranking and see what else it comes up with. But unfortunately, what we have to do now is we have to run that story.
[00:13:41] Okay. So, so now we're running the story and he's crunching his numbers again now, but removing the ranking. Whilst he's doing its work, we can flick back to our slides. Right. Okay. I'm missing a slide. All right. Okay. What I'm going to [00:13:58] do is I'm going to talk a little bit about.
[00:14:05] Alright, I'll find it, but I don't know what was on there. So, I'm going to talk about, the datasets and where do we find this dataset of 30,000 TripAdvisor reviews? Cause what we want to do is from this is realize how simple it is to analyze data. So there's plenty of websites where you can download datasets.
[00:14:25] One of the big ones is called Kaggle. And so I went onto Kaggle and you can review all sorts of different datasets, whether it be the weather or your football scores or soccer scores in the States and the history. And you can import these, and, and analyze them because I like my food. I chose TripAdvisor, European restaurants.
[00:14:45] Okay. So. That's very timely. So now we got rid of ranking. And now Einstein's looking at this and going, wow. The most, the thing that has the biggest influence here on our, on our outcomes, so our TripAdvisor scores for restaurants in Europe, is if [00:14:00] it's vegan or not. So if vegan is true, the score is significantly higher than if it wasn't true.
[00:15:09] Okay. Now this doesn't mean that it is a vegan restaurant. Vegan was in the description. So the restaurant has a vegan offering. So typically if a restaurant in Europe has a vegan offering, it's likely to score much better than a restaurant that doesn't have a vegan offering. Okay. So this was the biggest factor.
[00:15:29] Okay. I can even scroll down now. And look at some other factors and there's also like the number of reviews. So, well, if a, if a restaurant's got a lot of reviews, then it's, significantly and vegan is false, it does better. Okay. Which is a bit of a strange one. So if you've got 860 to 16,000 reviews, could be vegan is false, does better.
[00:15:45] So maybe a bit of an anomaly. Nope, scroll down again. And actually in Rome, vegan false does better as well. So actually, if you go to a restaurant in Rome and they haven't got vegan on the menu, [00:16:05] it doesn't matter. So again these insights are something that we can, we'd never normally pick up just by looking at the data itself.
[00:16:12] Okay. So you go to Barcelona, vegan performs very, very strongly. Okay. So, what could do that is actually, let's look at these number of reviews, as a, an example here, and I want to analyze that a little bit more so I can put in a number of reviews there, and then we can see actually there's quite a strong correlation with score
[00:16:32] and number of reviews. So just to explain this here, where we've got some gray bars, the gray bars are not statistically significant, whereas the blue bars are obviously significantly low scores and the, and likewise up here with it, with the high number of reviews. So what this is telling us is the more reviews a restaurant has got,
[00:16:51] the better it scores. And obviously if it's got vegan on the menu, that's a good thing as well, unless you’re going to Rome. Okay. But what we really want to know is about where to go really. So I can sort of look and see what's happened to my data by looking at city. Yeah. So I'm just going to look at city on its own.
[00:17:07] And here we go. And here’s all of our, all of our cities. Now we can see that our average TripAdvisor score is four. Okay. So Brussels down here at 3.9, Lyon at 3.9, statistically lower than the average. The gray bars are kind of average and the blue bars above the orange line are very good. So, and this is ordered left to right in terms of the number of, reviews as well.
[00:17:36] So London has had, it's got the bulk number of restaurants, right. It’s got the most restaurants in this TripAdvisor dataset. We’re in London, Paris. Right. And so all the big European cities up here and then. Maybe some of the smaller ones down here. What it does show you is actually there's some, I was going to try and say Ljubljana, is a features pretty well.
[00:17:57] It's a, it's a good place to go. And as does Athens. Athens is probably the top one there. So what I can actually then do is I can even then drill down into Athens a little bit more. It's like, okay, I want to go to Athens. What does that look like? Okay, well, vegan is better. That's not [00:18:12] surprising.
[00:18:13] Okay. And actually also we can see that when the price point is low. or medium, it does better. So actually, if you're on a budget, Athens has not only got your best TripAdvisor scores going, but, but you can do on a budget. Actually you know, go to the higher priced restaurants in Athens, they’re kind of not so significantly different from the average.
[00:18:36] Okay. And then you can scroll a little bit more about Athens if that's a, if that's the city that you want to go to. So let's just exit that now. The next thing I'm going to do, so this is what's happening in the data, but now we're going to look at the reasons behind everything. Okay. So we can see here vegan false, was, describing a big chunk of the, the variance.
[00:18:58] So you can see here, this is our average rating. The average rating, rather, not vegan is true, average over the whole dataset. And then all these things here are negative impacts, which would drive the score down if vegan is false. So vegan is false on its own [00:19:16] is, is a big thing.
[00:19:18] Okay. There's a little bit of a positive one here. So what's this one? Yeah, the oldest is this number of reviews. And again, if it's like between 860 and 16,000 reviews and vegan is false, it does a bit better, a bit of an anomaly. But hey, it's still statistically significant. Paris, vegan is false, bad.
[00:19:36] If price is low and vegan is false, very bad. So get, just go somewhere cheap and they haven't got any vegan on the menu, not good. This is a big one here, by the looks of it. If the price is medium and vegan is false, ok, you get to a medium price restaurant. You might expect some choice in the menu.
[00:19:53] And then there's a whole bunch of small terms that, that influence the outcome. So overall vegan is false. It's driving down the average right down to 3.9. Okay, so this, so Walt's full chart. You could really drill down into the specifics right now. This is all very well and good, but probably it's a good idea to think about how this, may play out in real life.
[00:20:16] You know, you're not, we're not just choosing a restaurant. Now, the actual reason why I created this dataset in the very first place was [00:20:22] to pitch to a client, a couple of months ago. They are an airport lounge business, a global airport lounge business. And they've got the challenge of maintaining customer satisfaction scores.
[00:20:35] Okay. Now the busier they get, naturally the lower the score, the customer satisfaction. If you remember going into an airport business lounge, and it's really, really busy, you might be a little bit miffed that you might pay to go to a lounge or it might be part of your, you know, the expensive business class ticket that you bought and you get into the lounge and it's packed.
[00:20:53] And then you're asked to fill out a customer survey and yes, you're not really happy cause it's busy. Well, so, so, so being a busy airport lounge is going to be a negative factor on CSAT scores. But there's other things, other insights that you get, which improve the score. It might be that you got lots of nice scallops on the menu or the, or, you know, different types of food or that the magazines are organized or whatever it may be.
[00:21:17] If you're capturing that data, you'll be then to be able to look into all these factors and what this airport lounge business is going to do is take it [00:21:26] predictions on how busy the airport is going to be. So they know then to roll out that the real nice things to help their CSAT score, if they're going to be particularly busy. If they're not going to be busy, then perhaps they don't need to have those expensive scallops, because generally the customers could be happy going into a nice relaxed lounge in the first place.
[00:21:44] So it helps them manage their sort of customer satisfaction. Obviously, this was bad timing for me to go and pitch this to them because their business is just pretty much stopped overnight. So hopefully, hopefully it'll come back. Okay. So the next thing is I want to look at, what's the difference?
[00:22:03] Okay. So we're going to get some more of these waterfall charts. So let's just pick a couple of cities. So Budapest, Bratislava. So some of the sort of smaller cities, and if you want to compare these two cities, you could look at the average rating here. So, where Bratislava outperforms Budapest.
[00:22:24] So at the top here, this is the average when the city is [00:22:27] Bratislava. So let's get your 4.1 full, pretty much score. And, so Budapest, sorry up here. And Bratislava is down the bottom here. So it's a couple of points below. Now. These are all the factors, that drive the rating of the difference.
[00:22:46] So, so basically. Down here, we've got Bratislava and the fact is vegan is false is a real bad thing. And Bratislava has it’s, vegan is true, basically. Why basically, obviously that means that Budapest, outperforms Bratislava, regardless of whether vegan is true or false. And then we've got, what else here?
[00:23:10] In terms of rating.
[00:23:13] Oh, the rating. Not really sure what that one is. Okay. Price is medium and vegan is false. So various factors from our various outcomes, which is driving the score down, but we can see there’s one big factor here, which is helping the score in Bratislava. And in fact, that's just small terms.
[00:23:31] That's 62 various other things which are insignificant on their own, but they're combined [00:23:37] to be fairly significant. So what that's telling us really is don’t go to Bratislava. Go to Budapest if you have a choice. Okay. So that’s gives you some difference. Now, once you’d had all that, actually you decided no, actually I might want to go to, I'm not sure where I want to go.
[00:23:54] I'm going to try some different things. So Einstein Analytics gives you the options to see what could happen. All right. So I can now go, but I know that vegan is true is a good thing to do. So I'm gonna go pick a restaurant where vegan is true. I can now pick my city. So where do I want to go? So let's say I wanna go to Lisbon.
[00:24:13] At random. And what's my budget? High, medium, low? I want a low budget. And, am I just going to try and find that secret restaurant without many reviews or find one with lots of reviews? So I'm going to pick one with lots of reviews. Wow. And it said to me, these things, I'm going to get a really high rating.
[00:24:36] You’d like to get five. You've got lots of reviews, low [00:23:38] price in Lisbon. That's really good. That was completely at random. Let's change the price point to see how that changes. Yeah. So let's choose a medium one. Einstein is saying, yeah, you're probably average rating. That is four. And, if you go high price in Lisbon, but again, you've got, you've got another four. And from then we can look at all the reasons behind that.
[00:25:01] Okay. So we can look at all the influences for our selection. Yeah, Sam's a vegan is true, big positives. And then we can look at some negatives, may have fact that the price is high and the city is Lisbon. Those two things are not, not a good combination. And then here, we've got the number of reviews in there as well as it being a bad, bad combination there.
[00:25:24] So just by changing this around and the price points.
[00:25:30] Change the shape of my waterfall graph and actually this selection here for Lisbon, which is a price of low and a lower number of reviews. That's a really good thing. Yeah. So there's some really good, positive things to [00:24:00] come out of that. So that's helping me influence my choice and giving me a better, a better overall ranking.
[00:25:48] That's just given me a 4.4. I think Einstein is just rounding this up to four. Actually, the result of this would be a 4.4 rating. Okay, so that gives you some sort of insight into, and, and hopefully, we'll just spin back to my slides here. I think what we showed you is you don't have to be a data scientist to do that.
[00:26:13] All I did was I prepped some data, and, uploaded it into Einstein Analytics. I would have iterated a few times, so I could go through this process a few times. And as you saw right at the beginning, I took out some columns, which I knew would not have influenced. I took them out to help the upload go a little bit more quickly.
[00:26:34] And so, yeah, pretty, pretty simple place to go. So in terms of, how you get started with sort of Einstein Analytics, there's a really strong Einstein Analytics Salesforce community. So, if you've, if you're familiar with Salesforce, you're almost [00:26:50] certainly familiar with the
[00:26:52] Salesforce community and all of the groups that are involved with that. And so, so search out for the, the Einstein Analytics community, and also Trailhead. Trailhead is Salesforce’s learning platform and there are a whole bunch of trails to get you started with Einstein Analytics. And this is where you get your free Einstein Analytics trial org.
[00:27:15] Okay. So, so. So if you do these, these trails, you get access to Einstein Analytics. Okay. There are limits on the org. It's not like full-blown one and I've actually just been using a trial org for this. So it can handle 30,000 records to analyze. So you can still upload a decent amount of information in there.
[00:27:31] But there are sort of certain limits whereby, you know, you would use it in your business full time because you’d use the busting the limits, but, but for the example to play with the data, and get some analysis done, that's a great place to start. So, yeah, so, so go trailblazing. Don't be scared of getting your hands on.
[00:27:50] and it's something you can learn really, really quickly without being a data [00:27:56] scientist. So, that concludes the talk. So I want to thank Xplenty for inviting me along and it's been an absolute pleasure, and I look forward to seeing all the other presentations, at the summit. Thanks, Steve. I've got a couple of questions if you, if you don't mind.
[00:28:13] No problem. The first one was when you were doing this analysis, there was, you had the choice between getting insights only, and then the full-blown analysis you showed. What, what's the use case or why would you use one or the other? So, so this, so the, what could happen is the full blown one.
[00:28:31] Yeah. Which we uploaded. So, but so you can actually play with the numbers here and, and it does all this analysis for you. So it's doing your predictions on all the variables. So you can actually, rather than look at just what has happened, you can then chop and change it and pick your own variables to get a potential outcome, basically.
[00:28:51] So I'm putting this into real life. Let's say, I know you see your marketing and you're marketing your products across, across the USA. [00:28:59] you might have variables such as, city, products, a retail outlet, the type of outlets that is, and the type of promotion that you're doing, whether you're doing an install, promotion or a coupon, or whether you do a diet malformations.
[00:29:16] So you might want to look at all the different effects of their various campaigns. So effectively, if you've got a whole bunch of data from, from previous campaigns, you can set as a market. Yeah. Just play around with this, this feature here to see what might work best in one city might, might not be suited at another.
[00:29:34] For example, something happening in Florida. You might not want to sell hats and scarves in Florida in March, but they might go really well in, in Chicago. Cause it’s colder, for example. There's some obvious data there, but to give you the sort of the real specific insights, if you've got some stuff, which is less obvious, like, you know, which promotion type is performing better than others, then, yeah, then it's really good to insights for the marketing just to play around and get a feel for, for what's going on with the data and sorry, one more thing, and then [00:30:06] you can deploy this into Salesforce.
[00:30:09] So what I was saying before about like customer attrition for subscription businesses. And so the one that we're going to be looking at very shortly is a telecoms company who are doing broadband prescriptions. And they're going to want to see what the likelihood is of the customer stopping
[00:30:25] and, quitting their subscription and going to a competitor for example. But we'll know that from a lot of the previous interactions we've had with that customer and how long they've been with us, how much they’re paying, when we last renewed their modem, for example, could be a, could be a key factor. And then, so what that then does it gives your customer support agents when they find out, oh yeah, we're going to cancel our subscription.
[00:30:47] Your customer support agent will see that customer, that profile right there. And it will be able to make some immediate recommendations in terms of what to do to, to retain that customer. That kind of brings me to our next question. What, what we're generally, what I've generally gleaned from seeing Einstein Presentations and [00:31:10] talking to people who use it is that it's mainly useful
[00:31:14] if you already have a Salesforce org and you put Einstein Analytics on top of it. It's not quite as good if you just want to analyze whatever data you have. Do you agree with that? Or do you think there's a standalone use case for Einstein? There potentially is a standalone use case. So, it's interesting.
[00:31:30] I was on a call with Salesforce last week about this. Generally Einstein Analytics is very deeply integrated in Salesforce, right? So, but I think you can buy Analytics on its own because also you can connect it. Obviously we just pulled in a CSV dataset where you can connect to other platforms that you may have to do the analysis.
[00:31:50] But the key thing about this in terms of the Einstein discovery is the right back to Salesforce. So that customer support agent, for example, you know, gets that information from the analytics, right at the heart of where they need it most about that time, when that customer's about to cancel, they could see what to do to retain that customer.
[00:32:08] or it might be the other way round. They might want to get rid of the customer. [00:32:13] So you might, it might say, hey, let them go, cause they're costing us a fortune. So that's good that they’re going. But so, but to get back to your question. Salesforce has recently acquired Tableau. And so I believe that the way that they are positioning the products is if you do most of your work in Salesforce.
[00:32:35] then Einstein Analytics is definitely the right solution for you. if you are somebody that doesn't use Salesforce, and you're doing some data analysis, but you don't necessarily need all that customer data, then you could look at like, Tableau, it might be the answer. So, cause that's sort of very good at connecting to lots of different datasets and different platforms, etc.
[00:32:57] So that's the, the marketing spill from Salesforce. If you're not using Salesforce in your day to day work, then look at Tableau, and if Salesforce is deployed within your business and you're using Salesforce, then Einstein Analytics, which means they will then say, there's a case whereby you can have Tableau and Analytics from a [00:33:18] large organization.
[00:33:24] But we’re great advocates of a single source of truth. So we like to use Salesforce, like from like right at the beginning where you looking at acquiring customers to getting paid. So, so really for our clients, we would more align to Einstein Analytics. And unless Tableau has changed in the last couple of years, the last time I used it, it had zero predictive analytics.
[00:33:34] It was just simple analytics. I think they've got some AI stuff in there now. Don't quote me on it because it's a recent acquisition. And because of the way that we're positioned and what we do with our customers, we were not intending to go down the Tableau route at the moment, but it does look great.
[00:33:59] I think some of the visuals from that look really good. Yeah, you can do some amazing stuff with it. But we've not looked at any dashboards here. This is just a discovery part of Einstein Analytics. I mean, there's a whole big piece on the dashboard analysis side of things as well.
[00:34:15] And some of the dashboards that we're building at the moment for similar clients are stunning, you know, whereby we've [00:34:20] got, you know, millions of records and especially for the, for the board meetings now, whereby they’ve just got one dashboard. And they can drill down and see what they need to see in the business.
[00:34:30] whereas before they were like, yeah, we've all been in those businesses where at the end of the quarter, everyone scrambling around with Excel spreadsheets, trying to get the numbers in to their boss, that they can present to their boss. Their boss eventually gets on to the, to the board meeting. And by the time, the data gets there, it's probably meaningless anyway, because everybody's
[00:34:47] chopped and changed it to make it look good. Whereas this, with a really good dataset you can get a single source of truth, if you really know what's going on in your business. Yeah. We have a couple other presentations in the conference about Einstein Analytics that show dashboards. One is obvious, obvious.
[00:35:03] It's a guy from Austin who, who has a nice dashboard. One that's kind of hidden is a, a woman who is a Pardot Specialist with Pardot dashboards, which were really nice. Yeah. Yeah. You get some really good marketing insights from these dashboards. Yeah. [00:35:20] I think a lot of these now, some of the Salesforce premium products come with Einstein Analytics built in.
[00:35:27] So, and, and what, Salesforce will do is let's say like the basic Einstein Analytics dashboard. Oh, see, that's not a built-in feature with standard Salesforce license, but they've already created some great dashboards and some, some really good templates because obviously that's the standard Salesforce functionality of, you know, leads and counts opportunities and that sort of thing.
[00:35:50] And they built really good dashboards around, around that. So out of the box, you can get started pretty quickly. The future of Einstein Analytics looks really rosy. Earlier this week, we were talking about looking at sentiment as well. So taking tech fields and looking at sentiment, that's quite exciting, really.
[00:36:11] That's really exciting for discovery actually, especially when looking at reviews and things. And then also the data prep side of things. So, I guess the guys doing the, the dashboard, presentation. You may lose the fact that they're looking at data prep 3.0, which [00:36:27] is the latest version will be coming out.
[00:36:31] I think it goes better in the summer, but general release is probably winter. This year makes the data prep a lot easier. So actually for the building, building dashboards, bespoke dashboards, if you're not using certain templates, they can be a little bit challenging, especially if you're not super technically orientated to start with.
[00:36:49] but that's going to get really easy going forward. It’s gonna make it, make it a lot, lot easier. Well, that's fascinating. Well, I'm glad to see that Salesforce is still pouring money into Einstein Analytics, even though they, they acquired Tableau, and this presentation was awesome. Fascinating. And hopefully—when we get past this—useful.
[00:37:09] So thanks so much for your time, Steve. And we really appreciate it. Thank you.