Customer Analytics (AMGT350)

Yeah, everybody.

It's Ryan and welcome to our next lecture series This

week, we're gonna be talking about customer analytics.

Um last week we were talking about customer optimization.

Right? But and that's going to lead us into this

topic. But it's important to remember that both product optimization

and customer optimization, both of those things have roles to

play. Right in in making the company competitive in allowing

the company to get customers, keep customers and grow customers.

You can't just have great customer optimization.

You have to have good product optimization to right if

you're gonna do those things.

But in this interactive age that we live in now,

at least much more so than in the past, right.

Individual customer information is what dr what drives the central

engine of our competition and our ability to build a

competitive advantage.

Right? The more information that we have on our customer,

the better the advantage we have over our competition.

Um because really if we don't have reliable insights into

the value and the needs of an individual customer, then

really our strategy has no direction right?

Um We're guessing, we play the guessing game right?

And every once in a while we might come up

with a product.

That's great.

But it takes customer optimization and customer analytics to predict

what customers are going to want and gives us the

direction that we need to give them what they're going

to want.

Okay, so in this chapter we're going to start with

the definition of data mining and how the idea of

data mining, it's very similar to customer analytics.

Then we'll talk about like what analytics look like.

Well what what does the analysis part of customer analytics

then who benefits from customer analytics?

And then finally we'll talk about how we can change

the customer experience using customer analytics.

Let's go ahead and get started and let's talk about

data mining.

So data mining is actually a term that's been around

for quite a while.

Um And you got it probably has some negative connotations

if if if like we just did a footbridge and

said well what is you know what do you think

of when you think of data mining you probably think

of something fairly negative.

Right? But it doesn't have to be.

So it's the process of extracting and presenting new knowledge.

It was always there but it was previously undetectable and

we were finding it through databases and we're turning it

into um helpful decisions right?

It's we're digging through the old databases.

All that information that we've gathered were mining for.

I think that the I think the analogy that you

hear with data mining is like there's all these little

nuggets out there right?

And so data mining is this process of going through

all that data and just trying to look for those

little nuggets right, extracting that.

That really juicy stuff that really works well.

Um But I think it's important to understand that as

relationship revolution right?

Are this this relationship revolution that we're in right now

where customers are trying to establish relationships with us instead

of just um you know having this as a customer.

Um The data mining process itself has undergone an important

transformation. So we have to understand that data mining today

is not what data mining used to be, you know

in the pre interactive age.

Data mining.

You know these were like techniques that we just used

to uncover information about types of customers um and looking

at like well who you know um would be a

good way to say this.

So like yeah which particular offers should be made to

which people?

Okay. Like, so I guess the best way to put

that into a question and be like, so data mining

used to be used to say who is the next

most likely customer to buy this product?

We got this kick ass product over here.

Who is the next most likely customer to buy this

product? Okay.

And I'm not saying that's not important information.

It certainly is.

But what we're looking for today, the question asked by

companies that are that are engaged in managing an ongoing

interactive relationship with individual customers and treating those customers differently.

The question now is not who's next most likely to

buy this product?

The question now is what is the next most likely

product that this particular customer we want to buy?

Right? They've purchased from us before We've got some information.

And can we predict what the best what the next

most likely product that they're going to buy is.

Right? Because in the past we were we were so

concerned with production, were so concerned with producing things at

a cheap cost and mass production, right?

We talked about that a few weeks ago, mass production

versus now we're looking at mass customization and say a

man, if we can predict what um product this customer's

gonna want to buy next, we can make it okay,

we can provide that service, we can provide that product.

So that's how data mining has really changed in regards

to what question we're trying to answer.

Yes. Um No.

So this this lecture series is called customer analytics and

this is really just a preferred term nowadays for data

mining um in the interactive age.

Right? It customers demand and and and companies are required

to have real time analytics.

We can't we can't say, well let's run a sales

report and see what we should do next.

We need real time data.

And if we can do that, we can provide continuously

developing real time insights into individual customers.

Right as we're dealing with the customer, we can look

and see what is going on with them right now.

And what this does is it provides the missing link

to understanding the customer.

It provides the missing link that allows us to predict

what it is they're going to do next.

Those real time insights give us some really good um

information. Really good understanding of what the customer will probably

do next.

And there's a ton of value in that.

Okay, so with that being said, I can read to

you the goal of customer analytics.

It's along the same lines, it's a little bit more

technical and we're gonna get to these parts of these.

We're gonna break this down a little bit more the

future slides.

But the goal of customer analytics is to enable the

enterprise okay.

Help the company to classify to estimate, to predict to

cluster and to more accurately describe data about customers and

that's using mathematical models, algorithms, whatever it takes that will

helpful, hopefully in the end, simplify how we can view,

it's our customer base and how we behave toward individual

customers. That's what we're looking for in customer analytics.

Okay. Looking for that data to figure out and combining

that with with algorithms and math mathematical models to say,

can we just break this down to some more layman

terms here to see how are we doing when it

comes to behavior with our customer base?

Like, you know, are we doing a good job?

Are we providing what they want?

Are we providing what they want next?

Right. So.

Mhm. We're gonna get into these parts up here, the

classify estimate, predict, and that comes in the analysis process.

Okay. It's important to first understand, right?

The company's, most companies anyways produce a ton of data,

right, huge amounts of data, whether it's through customer related

business processes.

Uh Maybe it could just be through order entry, through

billing through taking reservations every time we take a complaint.

Um Every time we create a if it's business to

business, every time we create a product specification for that

customer on a new product.

Um Web interaction sales calls.

We just there's so much data there and for, you

know, a long time we didn't really know what to

do with it.

Right. All that data is being captured though.

At least we're doing a good job with that, right?

The data is being captured.

It's being fed into some big data warehouse where a

lot of it lies in some kind of like you

know just file here right?

Some kind of data tomb that's just it's there to

die. Um And it's forgotten about for years.

Okay. I mean even when a lot of times even

when a company has customer analytics resources to to look

at some of this value in the data they even

sometimes just come across like there's some problems and say

like oh my gosh look at this data.

We have oh it's expired it's irrelevant.

Doesn't really have anything to do with what we're looking

for. Its non sequential.

So it's like it's it's out of order so it

doesn't really help rights non sensible.

It's not gonna it's not it doesn't make sense.

There are ways to take that type of what they

call that dirty data and we can clean that data

right? Where we can say let's eliminate some things.

Let's update some things.

Let's look for some correlations.

Let's refine the data and we can clean it up.

But even that takes a lot a lot of work.

So it is important to understand that if you've got

a whole bunch of data and you don't you haven't

been using it you don't know what to do with

it. You've probably got a lot of dirty data that

needs cleaned up.

That's at least better than having no data.

Okay. But it's gonna take a lot a lot of

work. Um The various activities involved in in readying the

customer data for analysis and the analysis process itself is

what these steps are about.

Okay so when we've got all this data how do

we how do we get it together?

How do we put it together so that we can

analyze it and then analyze it using these things?

Okay, so first of all classification, uh assigning instances to

a group, then using the data to learn the patterns

or the of the traits that identify the group to

which each instant belongs.

Now this is almost like.

So when I teach the fundamentals of marketing class, we

talk about segmentation and I talked to you guys about

it real briefly talking about like splitting up the market

into different groups.

This is like the first pre segmentation step.

Right answer is a very broad question that says, you

know what, what is similar about these people and their

interaction with us?

Um Right.

What is instead of instead of segmenting people by age

or gender?

Um We could say, right, what is similar about these

people? what is similar about the way they interacted with

us? Um Like we just say like well customers that

have purchased from us multiple times, right?

Like let's put them in a class.

Let's classify them in a group that would be kind

of like segmentation and then look for other things that

maybe they have in common.

I'll mention it again later.

But like it is important to understand that not things

don't, well it kind of comes up later.

Like what we're looking for is is commonalities to say,

oh wow.

If if there are customers who purchased from us multiple

times and they also have this in common, then maybe

we got we've got something but doesn't necessarily correlate that

you have to be a little bit wary of that

sometimes. Yes, sometimes not the estimation process.

So now we're determining a value for some unknown continuous

variable. The examples you give here, maybe a credit card

balance, uh maybe income, right?

We can take a look at those types of things

and say, you know um you know, let's let's we

can um try matching that up with the classifications and

see an estimate.

Like is there any correlation there?

Um but um these are kind of things are a

little bit outside of the core business.

Right. Were we might have to take a little bit

of a guess on the estimation side of things.

Then there's regression, which is, which is kind of what

I'm talking about, It uses the existing values to forecast

what continuous values are likely to be.

Right? How strong is the link between income and whether

or not they did business with us more than once

if it's a really strong length and there's strong uh

you know, correlation there.

Gosh, this is a huge slide and I'm running up

against the clock.

So let me posit here and I'll continue with prediction

in volume two.

Customer Analytics (AMGT350) - vol. 1

Chapters

Audio TranscriptAuto-generated