Practice English Speaking&Listening with: Einstein Analytics – Release Readiness LIVE, Spring '20

Normal
(0)
Difficulty: 0

[MUSIC]

Elna Miller: Welcome to Release Readiness Live.

I'm Elna Miller, your Release Readiness champion.

We are here in the studio all this week to talk about new features from the Spring'20

release across all the major clouds.

I this broadcast we are talking with a team from Einstein Analytics.

Our hope is that whether this is your first Salesforce release or your 20th,

you'll learn something new that you can use.

Our product team is right here in the studio to tell you about some

new features.

Do some live demos and answer questions on the air.

Speaking of questions, let's take a quick moment to review how this works.

We will be monitoring both the Release Readiness Trailblazers Community and

Twitter.

And the most important thing to remember is to use the hashtag #salesforcelive.

I recommend posting your longer, more detailed questions in the community.

Where conversation can continue after the webcast.

Now, if we don't get to your question on the air, don't worry.

We'll be sure to follow up with you within a few days.

We are recording this broadcast so you can watch it back whenever you like.

Now, since we may be covering the roadmap,

we will be making some forward looking statements.

So please make sure you make any purchasing decisions

based on currently available functionality.

And now let's meet our product managers today, starting with Arthur, welcome.

Arthur Fabre: Hi guys, I'm Arthur, I'm the product manager for the Builder team.

This is first time for me so I'm pretty thrilled to

be sharing some of the awesome features for the experiences track.

Avni Wadhwa: Hi, my name is Avni, I'm the product manager for Einstein Discovery.

And I'm really excited to talk about the new features coming out in

the Spring'20 release.

Radha Vivek: Hi, my name is Radha, I'm a product manager as well.

I'm looking forward to showcase two new products that we are launching with

Spring'20.

Elna Miller: All right, thank you and welcome.

Radha Vivek: Thank you.

Elna Miller: Thanks, so Arthur, what's new with experience?

Arthur Fabre: All right, so let's dive right into it.

So as we were saying, there are three ladies here that will be presenting.

I'll be doing the experience one, Arni will be diving into the intelligence one.

And Radha will look at the Industry Analytics app.

So let's dive right into the experience summary.

And here there are three main themes that we're focusing on.

The first one being Fast Start, the second one being simple and

the last one being everywhere.

So I won't go into too much details here.

But just to highlight some of the key ones that we'll be talking a bit into more

depth.

So for Fast Start, Snowflake Direct and

watchlists are definitely features that you should watch out for.

And be sure to try them out, even though they're in beta for this release.

For the simple side, as you can see, there are quite some features here.

And I'll be doing a lot of demos about this.

But yeah, focusing on making it simpler for

you guys to build more dynamic dashboards.

And make them more creative, interactive and

all through the UI without having to dive in into too much code.

And last but not least, there's some awesome features coming out on

the engagement side with the email subscription panel.

And the ability to add new email recipients to notifications.

So let's dive right into the Builder side of things.

And let's have a look at the dynamic reference line UI.

So with this new UI,

you will allow you through a couple of clicks to be able to add a reference line.

With its dynamic label and value to your awesome charts.

The second thing that I wanted to talk about is the Advanced Interaction Editor.

And we're really excited to release this in the Spring'20 release.

You can now forget about common E, you can direct with this new tool.

You can now dive into your widgets and

queries without having to search for them and NLSE into the JSON.

In addition, this new tool will help you build and

preview your bindings before bringing them to life inside of your dashboards.

And last, we are releasing and enhancing targeted faceting,

which was previously doing both through code.

But we're now able to do it all through UI and without any JSON.

And it will allow you to select specific queries that you want to facet on this.

So without further ado, let's dive into some demos.

So that you guys can see how to use this tool and what they look like.

So, what you have here is a call center dashboard,

with some great KPIs about today's volume.

Also some metrics about the changes.

And some great dashboards showing you how many calls are allowed every two hours.

So what I want to do,

what I want to start doing here is add a dynamic preference line.

And the way to do this is we're going to go into edit mode and

we're going to look at their strength.

So, so far, so good.

The first thing we want to do is going to the Properties panel on the right

side here.

And the similar way your will add a static reference line.

We'll select the y axis and click on the reference line right here.

So preemptively,

I have created a query that will be the one driving the reference line.

And it's called average hours per day.

So what's new for you here that you can see here, is these two new icons,

here and there.

That will let you choose a value for the reference line based on an existing query.

So if I click on this,

it takes me to this new panel where I'm able to select the query.

So here I'm looking for the average number of hours per day

as well as the field, which is the average number.

And the last thing we want to determine here is whether it is a result or

selection binding type that we're choosing.

For this use case,

I'm just going to use result which is the default one and hit apply.

And as you can see, a reference line was added.

So, the next thing we want to do is have a nice label for it.

And here the same way we just did for the value.

We're going to click on here and go through the same exact steps.

So, select a measure and select a field.

And as you can see here, we have this,

kind of rich text editor that got populated with the field.

So what we want to do next is, let's say,

add some customization, so average is and then hit apply.

And now we have our dynamic reference line that was created with its nice and

awesome label.

So far you can tell me, this is just a normal static reference line look like.

But if, let's say, we go into preview mode and

now we apply some filter on the time right here and we change this.

Arthur Fabre: Hit Apply, as you can see, the reference sign value and

label were updated dynamically as well as the chart.

Pretty awesome, right?

And how easy was this?

It took me a couple of clicks, and it was all through our beautiful UI.

All right, so what I want to do next is demo targeted faceting.

And to do this, I'm going to go into the second page on my dashboard.

Where I have some different charts, some filters, some of these filters.

And also a new chart right here that I'm going to use to pass selections.

So just before diving into targeted faceting,

just to make sure everyone understands that.

So faceting is Consists of the selections that get passed

when you're clicking on different elements of your dashboards.

So here, if I go into preview mode, and I click on this chart,

you can see that those number widgets on the left side are getting updated, right?

So are the charts.

So let's say for some reason that here,

I always want this KPI right here, the 2.7K to stay like this,

I never want it to change based on the selections I'm making on this chart.

So let's say we have a user that is interested in looking at the numbers for

June or July, but I don't want those KPIs to change for the users.

So the way we do this usually, and so far, was going to the JSON and

have to add specific to modify the query so

that the facet incoming for this query is excluding this guy.

But with our new UI, you're now able to do that without any code.

So how we would go about it, we would look at this query first, since it's the one

that we're looking to exclude, and we're looking at what's its name, right?

And so here it's called hours per month.

So what we're going to do next is we're going to

go take a look at our number query.

And right here in the Property panel, we have the Query tab.

And if we look under the faceting sections that was existing before,

we now have the ability in this drop-down menu to select Includes or

Excludes, whereas previously it was All or None.

So let's say here we want to exclude the query running this chart.

The way we would do this is by selecting Excludes and then choosing the query.

And so once we hit this button, we pulled in this list of queries.

And here we are able to make multiple selections on the queries we're interested

in excluding.

So here, the one we want to exclude is the one named hours per month.

So let's do this.

And if we exclude this one, and we hit Apply, so

if ever you click the wrong one and when you want to go back,

you can simply hit Choose Queries, again, and it'll take you back to that model.

Arthur Fabre: But once this is done, we can go back into Preview mode.

And here, you'll be able to select any others, so make any of your selections.

And as you can tell, the value here is never changing, right?

The KPI is staying the same no matter what selections we're passing through it.

So the same way I did it with exclude, you're now able to do that with include

as well, depending on whatever suits you better.

Arthur Fabre: The last thing I want to get into is the Advanced Interaction Editor.

So what we're going to try and do now,

there are multiple aspects why this is a great tool.

The first one is something that I'm sure you all do, is building really complex

dashboards with complex binding and you're trying to make this more interactive.

And to do this, the only way is to go into the JSON editor through and

you land into this huge page with a lot of information.

And you end up having to search for your widgets or query for a while.

So what's great about this is, now, let's say I'm interested in this chart.

And I want to access the Advanced Interaction Editor.

There are two ways of doing so.

There's a button right here at the top of the right panel, on the Property panel.

And right here at the bottom in this blue panel.

So if I click on Advanced Editor, it takes me to this new model.

And as you can see right now, I'm in Widget property,

but I can also switch to the Query code.

So before going further, I want to kind of take some time to show you

the different elements and what each of those do.

So in the middle here, you have the code for only that widgets or that query.

Which will take away all the time searching for the query you want to modify

or make more interactive or change, which is its first grade asset.

The second thing is this last panel, which will let you build a binding.

And with the help of the bottom panel here,

which shows through the creative interaction as well as the result,

as you're building it, it will let you preview what you're building.

And you'll be able to change it or better understand what you're doing so

that you can modify it if it's not exactly what you want or what you're expecting.

So for the purpose of this demo, we're going to build a really simple binding,

and I actually have already built it, but l'll go through the process of doing it.

So here what we're trying to do is I created custom query that's powering this

widget, it's called, Arthur Fabre: Summer Month.

And it's a toggle widget, and

what I'm going to try to do is link this toggle widget to a text widget.

So it's fairly simple,

it's just to show the process of building the binding and how it works.

So to do so, we're going to go into the advanced editor for this text widget.

And here, this is what we're trying to get to, this one here.

So let's go through the process together.

So the first thing you're going to do is select a source query.

So here, we're looking for Summer Month, which is all the way at the bottom.

Another thing that I didn't mention and

that is worth definitely worth looking at as a feature that was

added in this release is this little eyeball icon right here that you see.

And what this does is it lets you preview queries.

It's a nice thing when you can't remember what all,

especially here in this dashboard you see there are a lot of really similar queries,

it will let you visualize really quickly without leaving this interface

what the query is and let you choose them.

So let's go back to Summer Month.

And now we're going to choose the data that we are interested in displaying and

binding it to.

So here what we want to do is select the column, because as you can see,

the only thing we have in this custom query is one simple column.

And we're going to click Display.

And so as you can see, as we're building it, it's getting populated,

the created interaction.

So once this is done, similarly to the dynamic preference line,

we're asked to chose the interaction type, so here, I want this to a selection.

In addition to this, under the More Options tab,

you can now set default value, which will generate a call as function,

as well as use data serialization functions.

So once this is done, Arthur Fabre: You have two options.

You can either copy it and paste it in, or go in to Edit mode if you want to

customize it further or add some modifications or tweaks to your bindings.

So for the purpose of this demo, I'm simply going to copy it and paste it here,

which is the exact same binding, Arthur Fabre: And hit Save.

Arthur Fabre: And if we go into preview right now, if I apply any selection on

June or July, so you can see it's getting populated within the text widget.

So this new tool will not only let you fill the binding but also preview them and

help you going through this process that we all know

is not the most fun one, right?

All right. So that was it for the builder part.

Now let's go back to the slides.

And let's dive into the Explorer Highlights.

So, on the Explorer side, you can now use Clicks-Not-Code for

pulling data, from multiple datasets, to build your insights.

And previously, this could be accomplished with custom SQL.

Now you're able to do it all through our beautiful UI.

So we understand that it happens quite often that you will

have data coming in from multiple data set.

And it can be quite tedious to combine them inside of

one single visualization so that it's much easier for your users to consume.

The next feature that I want to present here is Snowflake Direct.

With Snowflake Direct, it will enable live exploration and real time insights.

Arthur Fabre: These insights can be used against your E data so

that you can have multiple, multiple data set pulled in and have them compared.

So this the great tool, the great assets for Snowflake Direct is that it

will let you do some quick explorations on data that you haven't pulled in so

that you can manipulator and have some nice visualizations, but

it definitely has some drawbacks about on the performance side.

The next thing that I'm touching on here as the picture enhancements that

I'll go through real quickly,

as you will be able to see that measuring group pickers have new visuals.

So let's dive into the demo.

And the first thing I will be doing a demo here is the data blending UI.

So for this exercise,

I want you to put yourself in the shoes of original sales manager.

And what I'm interested in doing here, as having my sales compared to my target.

So the issue here is that my sales and my target are in two different data sets.

And they are also on different greens.

So for the target data set, we only have target and product subcategory.

As you can see here, if I'm going into my value stable for the target datasets.

Arthur Fabre: You'll only see product category,

product subcategory under target for each of those.

Whereas, if I'm looking at my sales dataset and

I go into the values table, you'll see that we have much more information.

Something some stuff about the customer names or

the shipping cost or other things.

So, what the question we are trying to answer here as for

each product subcategory, how am I doing compared to my targets.

And in order to solve this, what we first one is Inside of our sale dataset,

we're going to add a product subcategory.

Actually, sorry, I'm first going to add the dataset.

So I'm going to add the target dataset.

And so once this is done I'm going to add the products of Getager.

And as you can see here in this new model,

you are able to select it from both of the data sets.

So here, let's go ahead and select this for each of them, and click Add.

And so what this does is we now have.

Arthur Fabre: We now have the data coming from both of the dataset about the project

sub categories and their values.

What we're going to do next is

add a bar length because what we're missing here is our targets, right?

So let's go ahead and open the new measures.

And if we scroll down here, we can see that for

this new data set, we have new measures.

And so we select targets.

And now this looks pretty good.

We now have the values for

each of the product subcategories in terms of rows here and in terms of Sales here.

That's the next thing we want to modify as go here and instead of council rows,

we want the sum of sales.

Awesome, so now, let me just do a quick modifications for

the chart type and the format so that it's easier to consume.

So I'm going to change it to a column chart and

I'm simply going to put this in a single axis.

All right, awesome.

So this is what we're trying to get to.

We now have a single visualization,

which makes the data which takes the data from multiple data sets.

And is breaking down information across both of them by product gets subcategory,

and then able to compare my sum of sales with my targets.

So that's pretty awesome.

Now, let's switch to Salesforce, Snowflake Direct.

Arthur Fabre: So, in order to do this, I'm going to close my tabs.

Arthur Fabre: All right, and the way you do this is you go into the Data Manager.

And here as you can see, I'm going to the Connect tab and

I'm going to click on Connect to data.

So now in the data manager can control web connections are showed to the users and

those are the only one they can explore.

And I've already setup some Live Connections as you can see

in this new tab right here.

So those what you're seeing here is our

older connections that you're all very familiar with and that you can use today.

But we have this new tab right here in which

you can add connections that will be live.

So as you can see here that I have already added some new ones, some existing ones.

The way you would do it is click on Add and

select the Snowflakes Direct corrector and then you will go ahead and fill that up.

So very similar to what exists and you currently have for the other type of data.

So I'm not going to go through all of it.

I'm simply going to show you what this looks like for the end user.

So if I go back here in the analytics studio, and if I look up my data sets,

you now have a whole section here, and that's called Live Data.

And if we go ahead and select it, you'll see that

you can click on them and make a direct call to Snowflake against the schema,

and this tell me all the tables that are available for me to explore.

So in this case I'm going to go ahead and select orders, and

as you can see it's loading.

It's loaded up, all right awesome.

So, what this does, and it's allowing me to query,

against Salesforce Snowflake data, directly in my and

now I can do any exploration as I would be doing it with other data sets.

So for this demo we're going to do something very simple, we're simply

going to group this by order priority and

we'll make it look a bit nicer and change the chart to a donut chart.

All right, awesome.

So what's great about it is that now,

let's say we want to add this to our dashboard.

We're going to go ahead and put this to designer.

And let's say sales by priority, and go ahead.

Arthur Fabre: All right. So this looks good.

We now have a dashboard with a Live query to Snowflake.

And by the time the dashboard loads up,

it pulls the data from Snowflake directly in Live.

So I was able to drag that query.

And now if I go into a preview mode, you can explore the data directly.

Pretty awesome.

All right, so that was it for the Snowflake Direct.

Back to the slides.

And now I'm going to jump into the home highlights.

So here a couple things to highlight.

With the spring 20 release, Einstein Analytics dashboards,

users can easily manage the email subscription.

With the ability to edit widget subscriptions, titles, and

new email preview.

So as you can see on the left,

we now have a panel that will help you manage those all in one place.

The second feature that I want to talk about is the ability to now

add email recipients to notifications.

So previously, notifications were linked only to one email.

And it was quite restrictive, so we're now added the ability to

have multiple emails being linked to the notifications.

And last but not least, the watch list.

This is an awesome feature that's coming up in beta for this release.

And that will help you gather all the most important metrics into a single view.

In addition, watch list is the only feature that allows trending.

That will let you see the evolutions of those most important

metrics all in one place.

So very awesome features and we're all excited about.

And that was it for my section.

Elna Miller: All right, thank you.

That's a lot of great stuff, and

there's a lot of excitement, especially from Christie.

She loves the dynamic reference lines and the targeted faceting.

And she's really glad she doesn't have to go into SCALA and JSON.

She's just going to reserve that for

when she wants to do something fun, [LAUGH] rather than that type of thing.

So thank you for that initiative.

John also said that Einstein Analytics is really gearing more towards clicks and

no code.

And really great for

admins who just want to get things done quickly and easily.

So appreciation for that, and

I thought that Live Query to snowflake was really impressive.

I mean, it looked so easy and fast and you just don't have to it upload.

Arthur Fabre: Yeah, it's pretty straightforward and it goes well.

Elna Miller: Nice, so a couple of questions for you, before we move on.

So speaking of the Clicks Not Code Initiative,

what are some of the next widgets or capabilities you're working on?

Arthur Fabre: So that's an awesome question.

So we're currently working on the text/number widget.

So we're thinking of improving the ability to make that dynamic.

And we know everyone uses text and number.

And that's the demo I went through to show you the advanced interaction editor.

To kind of show you, how to make this dynamic.

So we know it's something that users definitely would like to see.

And that's what we're working on right now.

Elna Miller: Right, and next question is how are the data sets connected together

and what kind of blend is it?

Arthur Fabre: Yeah, so that's great question as well.

So for you to know when you're, by default when you're building.

When you're connecting the data sets together,

as connected through a lifetimes.

But you have the ability to choose whichever one you want.

You have right blends, inner blends.

So let's say like for the use case, the example we were looking at.

If you had some rows where you wouldn't have any target or sales number.

Using inner blend, you would be able to take those away.

Elna Miller: All right, great, now, about Snowflake Direct, so

I think it was pretty clear what the advantages of that were.

I mean, having the connector, a lot of that work is sort of done for

you, but what are the advantages of using that?

And why should I still import or when would I still need to import data?

Arthur Fabre: Sure,so the advantage of Snowflake Direct,

it's mostly as for quick exploration.

So it will allow you to get a sense of what your data set will look like and

kind of understand your data better.

But one of the bigger drawbacks for sure, and one of the reasons why we think

it's very still relevant to import your data, in terms of performance.

And there are a lot of things that will not be supported yet

with Snowflake Direct such as bindings and other elements.

So that there's definitely a huge attraction for

this feature because they will let you do great things.

But we still have to improve it in order to make it to the same level as well.

Elna Miller: So maybe you're looking at importing a subset when you want to do

a real deep analysis.

Arthur Fabre: Yeah. Elna Miller: And

kind of make a different associations.

Arthur Fabre: Definitely.

Elna Miller: Whereas it just want to take a quick look kind of get a sense

of the direction that the data is going.

And to kind of give you some ideas of what you might want to import.

Arthur Fabre: Yeah. Elna Miller: Then the direct query

would be useful for that, right?

Arthur Fabre: More relevant for sure.

Elna Miller: Right, got it.

That's really interesting.

So I think we're ready now to move on to intelligence.

And Avni, it's over to you.

Avni Wadhwa: Great, so I just wanted to start out with what the highlights for

this release are.

So we have a lot of really great features, but

the ones I really wanted to focus on were.

We have new correlation calculation coming out.

So we're all used to seeing,

how correlated are the fields in my data set with what I'm predicting for?

That's really useful information that you used to be able to get

after you run your first story.

Now when you're setting up your store, you can see that right off the bat.

So if you have something that you want to ignore,

that looks a little bit too correlated, you can do so.

Or if you see some that aren't correlated at all, you can also ignore those.

We have piloted tree based models.

So I'll talk a little bit more about this.

And you'll see it in the demo, but we used to just have linear models,

but now we are piloting GBM and XGBoost models.

We have what could happen, so

we now give users the ability to do some what if analysis to play with.

I have this model that I'm having it deployed.

What are possible predictions that could come from it?

We have the long awaited story versioning that just went live.

So you know, now you don't have to create new containers of the same story.

It's all logged within one story container.

And then lastly, scoring with data set joins, so

external data can now be used to.

Or more than one data set actually can be used to score on new data as it comes in.

Avni Wadhwa: So first I want to talk about tree based models.

So what you know we're you're pretty used to is seeing the linear models in

Einstein Discovery.

So it's just the linear model where you just do one of two types of predictions.

Now what we're going to do is pilot both GBM and XG boost.

So users will have the ability to turn this on.

And just play with what these new models look like,

what their performance looks like.

Or you can leave it up to discovery to pick what the best algorithm for

your data set is.

So we'll run all of them and give you the one that performs the best.

So that's really exciting but I want to clarify what all is included in the pilot.

So right now, these models are not deployable, you cannot go ahead and

deploy them and score them on new data.

It's really only for the purposes of playing with the performance.

So seeing what type of accuracy lift you got.

And then some of the expandability functionality is not yet available either.

So in that why it happened,

the second-order analysis might not be totally functional.

Yeah, but we'll get those definitely cleared out before it goes to GA.

Avni Wadhwa: Next, what could happen?

So you'll see this in the demo that I give,

but really love to play with some what-if analysis so

you can get some live predictions on a row of data that you create on the fly.

And see even the improvements and the top predictors.

Avni Wadhwa: Next, story versioning.

So get that lineage of the story that you created.

So you're taking those really great recommended updates to generate

new versions of the same story, and you don't want fully new story containers.

So this is a really great place to see the lineage of the entire data of the entire

story and see what you changed to change the performance.

Avni Wadhwa: And in score with dataset joins.

So, you're only able today to use one Salesforce dataset to do scoring on.

Now, we give you the ability to do more than one dataset.

Add a supplemental dataset so that if you've done any feature engineering

before you've brought the data into discovery.

We take account of that before you score on new records.

Avni Wadhwa: And the last thing that I want to mention before I dive into

the demo is the end of life for classic discovery.

So as the Spring 20 release rolls out,

which will be some time in February, you'll see that if you're still on AWS,

now you're automatically ported over onto the new version of discovery.

So that were really end of life in the classic discovery.

If you still need to see your old data, it'll be available for three months.

Please log a support ticket to be able to revert back to that old version.

And the last thing that I want to mention here is that existing stories will not be

ported over, so you'll have to recreate them on the new version of discovery.

So with that, I'm going to go ahead and

dive into a demo of the features that we talked about.

So in analytic studio, we pick a dataset so

just as we normally do and create a story.

Avni Wadhwa: So in this use case,

what I want to do is maximize my employees not a trading.

So what can I do to keep my employees?

I'm going to do a model with insights so

both the live insights as well as the predictive model.

And then I want to manually take a look at my data before creating that story.

So the first thing I want to point out here is that

these correlations are automatically calculated on the fly.

So this column used to be able to see after the first run of your dataset, but

now I see right away what are the most important columns in this dataset.

And if you see any that are 0% or 0.2% correlated,

you can go ahead and drop them so that you know right off the bat

which are the important columns that you should be paying attention to.

The other thing that I want to point out here is this is how we

enable the GBM pilot.

So if you're interested in this, please reach out to your sales rep and

I will go ahead and turn that flag on for you so that you can go ahead and

try out this GBM pilot.

You can turn this on and it'll automatically create a GBM story for you

so you can compare it with what your old linear model story might look like and see

what kind of accuracy lift you might be getting from this new type of algorithm.

What's really exciting here and what we're working on before we GA is,

all of the explainability that you're used to in discovery and that you love

where you get to see exactly why the model made the decisions that it did.

Are going to carry over into three based models.

So the user will have exactly the same experience.

You won't see any change in the explain ability, the model metrics,

anything like that.

So that's really, really exciting for us.

So let's move to a story that's already finished.

So this is a screen that most discovery users are very used to.

The first thing that I want to point out here is that we've gone ahead and

cleaned up our tooltips a little bit.

So instead of seeing just a blatant number here, you get a percentage.

So this is a much more easy to read kind of view of the same insights that

you're used to.

So let's take a look at the version history.

So this is versioning live in action.

You get to see all the versions of the story and a timestamp around them.

So you can rename these versions, you can say final or

anything like that when it comes to sharing these stories.

But you get a full lineage of when the story was created and

you can compare multiple stories,

and you know that compare model metrics feature that came out last release.

So you can see how changes in data or how recommended updates that you took into

account, how they changed the performance of your story.

Avni Wadhwa: Let's go ahead and move to what could happen.

So let's create a row of data here and

see how it performs with a prediction on the fly.

So here, I'm going to say that I have an employee who works overtime and

who's a sales rep.

They've been working four to five years.

Let's give them that monthly income.

We'll say, they're 27 to 29 years old and they're married.

So here, right off the bat and really automatically, I'm getting a prediction.

So for this row of data, for this employee that I've constructed,

Einstein is telling me that they are going to a trip.

Down here, we see this is the cutoff.

So at this cutoff,

we say that an employee is going to a trip versus not going to a trip.

And we get the top predictive factors here as well.

So it's telling me that works overtime is the number one reason

that this person is going to a trip.

And that as maybe their boss, I have something that I have control over.

So I can mark that works overtime as actionable.

And now Einstein is telling me that I have a recommended improvement that

I could take to lessen this person's chances of a trading.

So if I change whether or not they work over time, let's see what happens.

So I click that and we see down here that now with the improved outcome,

the possible outcome that could exist whether if I took that improvement

into account is telling me that now this person is not going to a trip,

after I took that action.

So this is a really, really great place to play with a model that maybe you're not

fully ready to deploy but you want to play with and

see possible results and possible predictions that you could be getting.

So we're really, really excited about this feature.

Now, and one more thing that I want to

mention here is that this threshold is based on that maximum accuracy.

But in that model metrics, if there's anything else that you want to play with,

so say that F1 score is what I want to maximize.

I can click on that, and as this threshold changes,

the what-if is respectful of that, and so

you'll see that the threshold here changes as well.

So that's really, really exciting.

It's all dynamic, all really in tune.

So let's say that I'm happy with this model and I'm now ready to deploy it.

We have a little bit of a new deploy flow.

So, here, let's say I want to deploy as a new prediction.

So first thing I want to point out here is, before you used to have to connect it

with a Salesforce object, that's no longer true.

You can now deploy this model without connecting to a Salesforce object.

And why would you want to do that?

So the first use case for

that is say that you want to surface these predictions outside of Salesforce.

So maybe in the use case that I just talked about

where I'm predicting whether or not an employee is going to attrit.

That might not be based on Salesforce data,

that might be like based on work data.

So I don't want it connected to a Salesforce object,

I want to get those predictions outside of Salesforce.

The second reason is, maybe this is a new data set that I want to score.

So its not really a new Salesforce object or anything that I want to connect.

It's just a new data set that I'd like to score on.

And that's when you would move forward without connecting to a Salesforce object.

So in this case, let's say employee is the object we're connecting it with.

And let's go ahead and map the fields.

So it's automatically mapping those fields back to that employee object.

But we see that monthly income is missing.

Because monthly income might be coming from a supplemental data set,

an outside data set.

So let's add a supplemental data set.

Say that that data is coming from all employees, that's another data set.

And I'm adding a second data set to kind of supplement what I already have and

really respect the feature engineering that was done before this data

set was used to build a story.

And then all else I have to do is specify the field to join on.

So in this case, employee number is going to be represented in both data sets.

And I can specify that, and now you'll see that I can choose monthly income.

So now all of my fields are mapped successfully and

I can move forward with deploying this data set,

scoring this model using more than one data set.

Last thing that I wanted to point out before passing it on to Radha is so

this is a the What Happened page that you're fairly used to.

You can now go ahead and hit Explore and explore that insight in a new lens.

So this is really, really exciting.

You no longer have to build this visualization,

you can go straight from the story and explore that card in a new lens.

So we're really, really excited about that.

With that, I'm going to go ahead and pass it on.

[CROSSTALK] Elna Miller: Why don't we take a couple of

questions.

Avni Wadhwa: Of course. Elna Miller: Thank you.

I love the multiple data sets.

It was so easy to just, just kind of knows what you

can now map once you've added that additional one, so very nice.

So in the tree-based model pilot,

could you just go over what is included in the pilot?

because sometimes it's partial functionality, or [CROSSTALK]

Avni Wadhwa: Right, definitely.

Sure, so in the pilot,

what we have today is the model metrics will all be available.

The what happened, so the descriptive analysis is available.

And so what the goal of the pilot really is is to have users see if they get

any sort of lift in improvement in the metrics.

So is there an accuracy lift that you're getting from this new type of algorithm?

Where the limitations lie are, so we're not going to have scoring, you can't yet

deploy these models.

Those will, of course, be available when we go GA, but that's a work in progress.

And then some of the explainability, so

the waterfall charts in the why it happened are not yet fully up to date.

So as we GA them, they'll be totally functional.

And you'll have the same experience that you were used to with the linear models,

but now with the tree-based models.

Elna Miller: Okay, and are there any special requirements or

is there a special process to get into the pilot?

Avni Wadhwa: Right, so

just reach out to your Salesforce sales person and they'll let me know.

I just need an org ID and I can go ahead and switch that on.

I'd really love to have, you know, new users to play with this pilot.

Elna Miller: Great next question is what data sets can I use for

the scoring with data set joints?

Avni Wadhwa: Sure, so any analytics data set, so again, those could be Salesforce

data sets, data sets that are just CSPs that they've uploaded, or

take advantage of the huge list of connectors that we now have.

And just bring those data sets into and create an analytics data set.

And any of those could be used as a supplemental data set to do scoring.

Elna Miller: Okay, great, now,

a lot of folks often wonder when they would use Einstein Discovery.

So can you just give a few of the most common use cases?

Avni Wadhwa: Sure so Discovery is more of a platform.

So it's definitely geared around CRM use cases, but

also since it is a platform, any external use cases could also be used.

So for example, lead scoring is a huge one that we see churn,

whether it's, customer churn or employee churn as we just demoed.

Things like that are great use cases.

So the two types of problems, just to be clear, that we use either 4GLM or

the new type of models that we're introducing are either predicting

a number, so you know, what is my C stack going to be or what is my sale?

What are my sales going to be?

Or a bucket.

So will my customer churn or not?

Or is this going to be a higher low effort use case?

Those are the two types of problems that Discovery today targets.

Elna Miller: Right, now one more question on the tree tree based model.

So it's in pilot now.

When do you expect it to go GA?

Avni Wadhwa: Sure, so we expect it to be GA in summer.

We're working hard on getting that done.

And so those two new types of models, GBM and extra boost,

we expect to be GA in the summer.

Elna Miller: All right, great.

So we have about 30 minutes left.

So we have one more speaker.

And I want to encourage everybody to ask as many questions as you want,

because we'll probably have time to do them on the air.

And with that, I'll had it over to Radha.

Radha Vivek: So while my two awesome colleagues talked about analytics

platform capabilities, enhanced features, and all that, I'm looking forward to talk

about two new products that we are launching in spring '20.

And both of them are purpose-built applications for, one for consumer goods.

And the other one is for customer lifecycle analytics.

Let me talk about consumer goods analytics application first.

So this is a purpose-built, pre-built analytics application

that runs on industry specific data modeling for consumer goods.

This is geared towards two primary target persona.

One is a salesperson, sales manager and

the second one is Field Sales Reps who are making store visits.

And using this pre-built analytics application,

they should be able to optimize their store wizards.

Promotion analysis, any store KPIs, compliance KPIs that

they care about in maximizing the sales, engaging with their customers and

planning, product, white spaces and things like that.

With that, let me dive into the application.

As I log in as a sales manager, within my homescreen,

I am presented with the top KPIs any sales manager cares about

with respect to field sales or any retail execution tasks.

For example, I can look at my territory revenue that I am responsible for

quarter to date.

And is it increasing or decreasing with respect to the prior period?

And also it gives me a big number that is giving me

the number of stores that need my attention.

And also gives a short list of stores where I need to pay attention.

And as a sales manager, if I have a team of sales reps that I'm managing,

it also gives me a leaderboard with respect to who's falling behind.

So that I can have quoting opportunities,

mentoring opportunities as I have one on one with the sales reps.

And last but not least, it also gives me what are the different sets of products

that I can sell in different stores and things like that.

Let's take a deeper look at one of the stores that need my attention.

Let's see this application in action so I'm going to click on More metrics and

it is going to launch a dashboard that is going to present

me with key KPIs with respect to all of the stores.

Such as revenue, unit volume and what is the compliance rate how

many visits are being held and then, are there any open cases?

Essentially augmenting all the CRM data with the consumer

goods cloud data with respect to how you can prioritize

some of the store visits, as well as sales strategies.

Let's dive into one of the stores that I want to focus on today.

For example, I'm going to look at this Ruivally store.

Right out of the bat I can see that the in store volume sales volume is declining,

so I definitely want to pay attention to the store.

So I can come in and then I can see what's going on with the store.

Radha Vivek: So I'm launching the retail store page right from

the analytics dashboard, right?

So right here, I can see the two dashboards have been embedded

without losing the context of the story that I want to analyze.

So as I come in here,

I want to focus on the last 30 days what's happening in the store.

And I can see that all the KPIs are being reflective of the ratings that I selected,

and typically today I'm going to focus on the unit volume.

Radha Vivek: So I'm even going to filter down to the quantity and

I can see that some of the products have been decreasing with

respect to average quantity that I'm selling in these stores.

So now that I know there are a couple of products that are not really selling well,

I want to see what I can do about to keeping those store sales intact.

Maybe I want to see what other products that I can position for this store.

So I click on the whitespace analysis and then maybe I want to look at

what are the top selling products in the store in the past.

So I'm going to look at maybe this quarter or maybe last 90 days,

and then I can see that there are a couple of products that I could position.

So let me see this product, so

obviously it looks like this is being sold pretty well.

So I'm probably going to position this product for this as well, and

maybe let's look at a different product.

So this seems to be up and down, maybe I'm not going to focus on this one.

So as I demoed you in the couple of clicks,

I was able to look at high level KPIs in my homepage.

And then I drill down to the store details and then I already have

a couple of sales strategies particular to this retail store.

Let me go back and then show you how I can use some of the prebuilt

content to measure and monitor the sales team performance.

So as I login as I clicked on sales team performance dashboard,

it is going to present you with a number of KPIs that are relevant for

each of those sales reps.

How you can have an engaging conversations,

let's say I'm going to look at Wendy today.

To see how she's performing with respect to her own sales targets,

unit sales volumes and the revenues and things like that.

It almost looks like she's falling behind with respect to unit sales volumes,

and it directly correlates with her sales revenue and the targets.

And I also notice some of the out of stock occurrences for her.

So I might engage with her to say, what is going on?

I can directly chatter with her to see out of stock

occurrences, let's follow up with us.

So as she logs into her own salesforce for instance, she's going to take

a look at some of these coaching opportunities that I had mentioned her.

So now, I talked about sales managers, now let's see how

a sales rep can take advantage of the prebuilt content that we have.

So for example, Wendy is planning her day to one of the stores.

So as she logs in, she can go ahead and she can click on the store insights.

As she logs in, for the sake of simplicity, I'm not switching

between devices, but this can be viewed on the mobile device as well.

So as you can see, it is going to show her within that store what is the total

revenue, and is it an increase or decrease compared to the prior period.

Which is giving a nice visual cue for her to have a better

conversation with the store managers as she plans her visit.

And she can also see all the compliance metrics that are relevant for

this particular store.

She can actually see that she can also see that there are a number of out of stock,

compliance metric has been low.

And then she can come down and

she can even see some of the cases that are out there.

She can actually quickly see that there are some of the cases related to late

delivery or the product recalls or discontinued product so

that she can have a meaningful conversation with the store manager.

So this is for the sales manager, Radha Vivek: Sorry sales sales rep so

she can have a better, so she can even collaborate with her manager to say.

What can we do about this product outages or

discontinued products, things like that.

So, this is all about consumer goods, back to slides, I want to talk

about the second application that we're launching with spring release.

Radha Vivek: So this is going to be based on customer life cycle analytics

essentially this is based on Salesforce survey data.

Wherein you can help CX managers and CX analysts to measure and

monitor the customer engagement, net promoter score and

how is the surveys or are you getting enough responses?

If you're not getting enough responses,

how can you manage their customer journeys and things like that.

With that, let me dive into the product.

Radha Vivek: So in here, we are primarily targeting two target personas.

One is a service agent, who is in front line helping customers

with their own questions about any product related questions, and things like that.

They care about customer satisfaction firsthand.

So in this card, you can see while they're helping

the customer on a specific customer case.

They can see the customer satisfaction,

rightfully embedded within the context, based on their survey responses.

Let's see how a CX manager, or

a CX analyst can take advantage of this prebuilt analytics application.

Once again, they are in their homepage.

They can quickly look at how is their net promoter score, and

specific to any customer journey that they want to zoom in on.

In this case, I selected one specific customer journey, and

it looks like the net promoter score is not so good.

And not only that, it is also giving me what is my customer base?

And who are those people who are responding to the surveys?

And at what stage of their customer journey do they help pinpoint?

So that as a customer experience manager,

I can focus on specifically to drill into that particular process,

whether I can optimize it, or maybe better engage with customers.

Let us see this application in detail, and

how they can get a deeper understanding.

So I'm going to look at the previous fiscal year, as we are in January.

So once again, you can see how which stage of the journey they can be improved.

And of course, it's giving you the nice breakdown of each of those stages.

And then how many customers do I have?

Who are responding?

And of the people who are responding, how is my score is going on?

This is one of the things, and

the other thing that I want to highlight is the response distribution itself.

For example, you send out a lot of service to customers.

You send out to a number of people, but you want to know what's working, and

what's not working.

And even within the surveys, what's been positive versus negative.

So in this case, you can see for all the services,

sorry, surveys that you're sending.

How is rate?

And how is the response rates?

And how many responses that you're getting.

How many notations have been sent out?

So basically, it's going to help you reach out your customers

in meaningful ways in appropriate touch points of their customer journey.

And you can also get to know the demographics of the people who

are responding to your surveys.

For example, you can see who are responding more, and

what's their age range.

And then what is the distribution by gender?

And then you can even look at the engagement ratio, and then you can even

look at the satisfaction ratio for all of the service that you're sending up.

So with that, these are the two products that I want to show a live demo.

And then I also want to talk about enhancements that we made for

the existing apps that we have.

So four other apps

that we already have with respect to different industry verticals for

financial services, insurance, HLS and manufacturing.

We have enhanced all these four products, even in spring release.

So specifically for wealth management, you can look at the configurability.

Meaning if a customer doesn't have financial transactions

captured in Salesforce, now they have the option to not

bring in those visualizations that required financial transactions.

And the second enhancement we made for FSE wealth management,

we deployed a managed package on appexchange.

Which allows customers to take bulk actions to add number

of clients to campaigns and create campaigns.

I've done the campaigns.

That's available on the manage package to download.

And then it can be used for any other use cases,

where you want to add a number of accounts, of clients, leads, or contacts.

So campaigns, and on the insurance side,

we have enhanced a lot of content that would include claims,

and also the distributor 360.

And on the [INAUDIBLE] side, we are including two predictive use cases that

are going to help for pre-authorization risk, and K plan adherence.

And we also enhanced some of the content updates for social determinants.

Basically, you can identify members who may

have social conditions that are preventing them from getting care.

And then take preventive actions, and

then making sure you can avoid any avoidable claims.

And for manufacturing, we have added configurability once again, to

provide customized app install, based on whether you are using forecasting or not.

And we also added product recommendations,

dashboards that you could take advantage of.

With that, thank you.

Elna Miller: All right, thank you very much, lots of great apps coming,

and what you're working on.

So I have a few questions, but first, just a little clarification.

Are these apps kind of a completely separate thing?

Or are they something that you would purchase on top of the standard?

Radha Vivek: Yes. Elna Miller: Einstein Analytics?

Radha Vivek: A great question.

All these industry products are very on top of the industry specific data models.

So for example, if you want to use consumer goods analytics application,

we would expect you to have that base sense for consumer goods and

retail execution data model readily available in your organization.

Elna Miller: So then this app would just basically save you time building the most

common things that are needed.

Radha Vivek: Absolutely, so all the features that my

colleagues have talked about, we are making it easy for you to consume, and

then accelerate your analytics journey with prebuilt content, curated datasets.

And you can even take these curated datasets and

run Einstein discovery stories on top of these curated datasets.

Elna Miller: Interesting, so you still have the full functionality doing your

own custom- Radha Vivek: Yes.

Elna Miller: Datasets and visualizations.

This sort of gives you just a big head start

with- Radha Vivek: Yes.

Elna Miller: What is typically needed for those industries.

Radha Vivek: Yes, on top of it, carefully thought out user experience.

Elna Miller: Right, thanks.

Now, what about bringing in external data, point of sale, for

example, for the consumer goods analytics?

Radha Vivek: That's a great question.

In fact, Einstein Analytics,

all these industry apps are powered by Einstein Analytics.

Platform.

So what that means is, all the capabilities is once again,

the data connectors, external data whether it is through

other Salesforce instance or connectors or CSV uploads.

They can happily bring in those data and then augment it with Salesforce data.

Elna Miller: Right.

And other than having the standard Einstein

licenses are there any other prerequisites to access the pre-built apps.

Radha Vivek: Yeah, I mean, like I said,

they should have industry clouds enabled in there orgs.

And of course we expect them to capture some data.

And then, with the Einstein analytics for

those specific industry products,

they can install this application via a templated install.

And then get to all the dashboards,

data sets, data flow, and use cases.

Elna Miller: Gotcha, and what all is included in these pre-built applications?

Radha Vivek: Yes, each of these applications comes with its own data flow.

I mean, we have made it easy for

you to make all the connections to these data models.

So that's one of the things that you can take advantage of.

And then of course, the data flow means we have already provided you plenty of

samples for doing rich transformations, aggregates and things like that.

And of course, pre-built dashboards that, they can either readily use or

they can use, they can customize and then build their own visualizations.

And of course they get curated data sets.

And some of these apps even come with the Einstein discovery stories embedded

within the pre-defined template, and also the story and the data set.

Elna Miller: Right, thank you.

And when you were showing the sales manager,

I think it was the sales manager experience?

With the homepage, and there was a dashboard there.

Is that a component that they can add to the homepage with

the cloud industry cloud setup or?

Radha Vivek: It's a analytics dashboard.

You can readily embed the dashboard in the homepage and

they can even embed some of the pre-built content into Salesforce record pages.

Because as I showed you, one of the dashboards embedded in the home and

one of the dashboards was embedded in the retail store page.

You can build your own content, as well.

Elna Miller: Great, thank you.

So I have a couple more questions and then I think we're about ready to wrap up.

Since Einstein Analytics does have quite a few different variations and

products and different license types?

So, maybe we could just go down the line.

And, for each of the areas that you talked about.

What type of license is required?

Which features are standard?

What requires an add-on?

Obviously everything we talked about today,

it requires some type of Einstein analytics license.

There's kind of the main base license, right?

But then, I'm assuming everybody would need the base license for

everything that you all talked about.

The main.

Radha Vivek: Yeah, I can speak for the industry analytics products.

So essentially they need to have the data that sorry, the cloud,

industry cloud, the specific industry cloud should be available.

And then on top of it, for example, Einstein Analytics for

Financial Services is an add-on to the FSC Cloud.

Einstein Analytics for

Consumer Goods Cloud is an add on to Salesforce Consumer Goods.

Elna Miller: Gotcha. Radha Vivek: So, something like that.

Elna Miller: Okay, and- Avni Wadhwa: Yeah, and

then Discovery is included in Einstein Analytics Plus, as well as

Einstein Prediction, so those are the two SKUs that include Discovery in them.

Elna Miller: All right, thank you.

Arthur Fabre: And for the features of that demo.

Most of them are available in the standard engine analytics one.

Elna Miller: Okay, perfect.

That was for Yogesh and anyone else who's curious.

So, back to the, I think this was about dashboard components in the builder.,

but maybe Christy can keep me honest here.

So, what about formatting, like making things bold or italic?

Arthur Fabre: So, that's something we definitely have in mind.

It's been asked multiple times and we know it's something our users want.

So, it's in our pipeline.

We want to get it done.

It will come down mostly to priorities, but

it's definitely something we're thinking of tackling.

And if it's not in this release, it's probably the next one.

Elna Miller: All right, great.

Well, thank you everybody, for presenting today.

That's about all we have.

And, I guess we'll just give you a few minutes back in your day.

Thank you for joining us.

We hope you enjoyed it as much as we did.

And if we didn't get to your question or if you have another question you think of

later, feel free to post it in the release readiness community.

We'll follow up with you in the next few days.

Also, this has been recorded so you can rewatch it any time or

share it with others.

So come back and join us again tomorrow at 9:30 Pacific for

The Platform Session, which also will be followed later in the day by Flow.

So we definitely can't miss that one.

So I can't wait to hear how you're using new features to innovate.

So, let's just keep the conversation going in the release ratings community,

where all trailblazers meet, to be release-ready.

The Description of Einstein Analytics – Release Readiness LIVE, Spring '20