Where is your Installed Base?
Is your Installed Base data a mess? How to get more visibility into your Installed Base Data. This session will cover: - Implementing Installed Base Visibility - Benefits of Installed Base Visibility - Enhancing Customer Relationships - Driving Revenue Growth
Let's get started. Let me quickly introduce our speaker for today.
Lullet, but Lullet is the CTO and title and has been at forefront of
designing both SAS and AI based solutions in the industrial
space for more than a decade.
He's one of those industry leaders who loves to get down in
trenches and has immense expertise in things related to
install base. And normally when he's not around his work desk,
you'll find them posting exciting stuff around data and AI on
LinkedIn. So with that, Lullet, I'm sure you're excited to get
started over to you.
Thanks, Sussan. Thanks for the introduction and welcome
everyone. Good morning, good evening to whichever part of the
world you are. So today as the point of the discussion is the
building how we can build the visibility of the install base
data. And why that is important will touch upon that a
little bit as well. But I want to start with this picture, which
when we talk to our customers, our prospects and generally in
the B2B world, everyone agrees to the fact that install
base is a gold mine. But that gold mine to mine that one that
that somewhat has becomes very, very difficult. And that's a
representation of the picture that we get when we start
talking to people that vary our data is and seems it's lying
all over the places. What kind of interaction the data points
has between them. It seems everyone is talking to everyone
even to try to figure out something very simple thing. So what
it leads to is that like the data is messy. You have different
kind of data points. Even the same data is represented with
different notion that they have a different categorization
and that is sitting in different kind of systems. And these
silos then lead to when we want to make them to talk to each
other leads to a lot of non scalable integrations. And fundamentally
what it leads to is that if anything useful has to be done. So
just for example, you might want to do some reporting or do
what to some adopt analysis or even want to figure out the
valid share. What it basically leads to is that like post
figuring out where the data is lying in different systems, then
start pulling that data, harmonizing them, putting it
together, unifying it. And the problem is that that whole
exercise takes a number of resources, a ton of time. And by
the time you are ready with the result, sometimes it's too
late to do anything with that kind of analysis. So with that
in mind, we'll just move ahead and why that happens. It's just
because I would say that actually that's how the industry
is evolved when we talk about space with the B2B
manufacturing world. Sometimes we talk about companies which
are there for like the gates and even centuries. And what has
happened is that during the time they go through multiple
technology transitions, they're still a part of the things that
are manual in spite of all the digital transformation. There
are big divisions. The sometimes it becomes more centralized,
sometimes it becomes more distributed, but all these leads to
different kind of systems that come into play. The system
come into the system, but never go away. So sometimes people
write internal tools, sometimes the acquired tools, but they
just remain there. And that just leads to a lot of this messiness
and lot of data silos that is sitting all across the
organization. And then we also know that I think what also is
happening is that this is like a lot of now tribal knowledge
that starts acquiring in the system. There are people who know
about everything, but even they don't know what they know
about everything kind of thing because you have to poke them
and figure out that okay do you know, then they will recollect
okay yeah, that is always sitting in my spreadsheet or in
some of the notebook and things like that. And that is what leads
to the whole message of the installed ways. So just before
move ahead, we can just do a small guess like for those who
are participating here today. And if you are dealing with
the installed base data, what state or the installed base data.
And then I think if you see that's a continuum basically. It's
mostly like either it's very manual within spreadsheets, PDFs
and those kind of things. There are organizations which are
I would say that still on a further in the curve that they
have systems in place, but there are multiple in picture. Then
there are organizations are further high in the curve who
usually have this master data management notion or
sing the source of unified data. But then there is this what
we call the Nirvana state where people have really using them
for the intelligence over unified data. So yeah, I think we
can move ahead on this one. Okay, so okay, looks like
multiple system that's what this problem that one. So and
that's not unusual that's what we have seen with most of the
organization and that's where I would say that the state of
the technology is also right here. We have like lot of this
ERP systems lot of CRM system and especially with bigger
organizations what we see is that these systems come
bound because of different mergers, equiditions kind of
thing. And just to basically look into some of the data
points what we have seen is that like this is a rough estimate
what we have seen among our customer base. Usually there
are 7 to 12 system that contain customer product and asset
information. And then this data quality because of that
usually the productivity losses by 20%. One of the basic
reason is that either the customer data is wrong or it is
incomplete which leads to a lot of work to figure out to
build the picture completely to build the rubric cube. So
that that can be actionable. And what it leads to is that
any simple workflow that has to be done minimum requires
4 to 5 people. We have to reach to them figure out those
data points collated together and then build an
X-rayable inside to them. So how we can solve this
problem. So it's like what happens it's like going to the
other end of a continuum saying that what happens if there
is a single source of truth is available to all stick
folder and it can automate the insights and people can
collaborate over that one. Would that be a Nirvana
state kind of thing which would help in mining the
install base the intelligence around it and drive the
growth through that one. And we also talk about a lot of data
driven decisions and that's where these all things start
coming into play. So we we talk about AI intelligence
and everything. But essentially at least what I have seen
this is very personal. I would say that 70 percent I
arrive just taking a number but it's in that range kind of
thing. The quality of analysis is a big function of the
datasets that it going inside those models that is going
inside those analytics. Otherwise it just becomes a
garbage in and that is out. So what we'll do is that now
we'll look into how we can do that journey. What is the
way to do that one. So the first and foremost thing is to
make sure that you get control of your install base data
quality. How that happens? You would have your data
sitting in all different systems. You need to take it and
take it through a process. Again I have put it like you
have to profile it clean it, map it and then when I say
map for example like in the install base world it's like
that whether it's an asset or it's a part those kind of
classifications come into picture. What is the mom
structures around that one. Then you have to enrich it
miss fill out the missing data points need to click it
them if I'm stitching. And this process is usually not
like one step after another. It's like a lot of it.
If things happen and then what happens is that once
you're able to do it you get your install base data
model which access to single source of truth. Believe me,
I think getting to this picture itself it's a journey,
but the journey worth taking because on the other side of
the thing now very clean data and what it opens up is the
possibility to a lot of different kind of analytics over a
clean set of data and the quality of analytics automatically
becomes much better just for one reason that the data is
very very clean now. So I'll just give another picture on
that one and actually just a small story out there because
entitled as you know the name means entitlement. So it's
like when we started almost 10 years back. We thought it
will just focus on the analytics AI part of the things. We
would be expect we were expecting that our customers which
is this B2B equipment manufacturer. We would get good
quality data and we will just focus on the AI model. And I
think that's where we wanted to focus on as well. But then
you quickly realize that I think to reach to that point we
will need to take care of this data story because the data
that you're getting was I think saying messy would be an
understatement. It's just like it comes from all over the
place kind of thing no relationship between them you get a
bunch of CXL files or bunch of dumps basically and start
making sense of that. And that's where we started doing work
on our data pipelines started making sure that if we need to
scale we need to get handle on this whole thing. And that's
where I'm trying to put a picture in place that's how we
look into that whole thing. We get the data on the left
inside of the things it could be coming in any format it
could be coming in CSV data could be served to API is
different kind of ERPCs. She has systems and even like
right now we see that a lot of people start putting the
data in data lakes. But the problem with data lakes is that
what people are doing is they just take the data and dump it
into data lakes. It's just like now the silos are sitting
inside data like okay you have all the data but doing
anything meaningful becomes very very difficult because the
data is not harmonized. A part at one place with a different
time and it is another data cello it's a different name.
How to make sure that we are talking about the same part. So
all those things start becoming a big problem. So what
happens is that then bring all this data into basically
harmonize it and then start pulling out a canonical
version of your install based data which would include a
clean data, clickers list of customers. They're different
locations. What all equipments they have got. What all parts
they have got. What are the service contracts which have
expired which are in force. What are the different
warranties. The complete transaction history. The
complete service history. Once that is in place, the
right hand side starts making sense. Otherwise they are
just models doing nothing. So how we can do that one. So I'm
just putting like this is typically and we have seen that
I think customers do go in that one even in our initial
days. We had a similar stack in place. You basically start
writing a lot of custom scripts. Take this data out.
Mishmash them together. Take them through different tools.
There are those ETL kind of tools. Then you'll take it
through analytics and this kind of things. The only problem
is that the stack becomes too complex very quickly. And the
bigger problem happens is that these are very generic tools.
So making doing to do anything purpose built for them, it takes
a lot of time to make sure that they are aligned with the
problem statements. And of course, one thing always that
comes into play is that the domain knowledge. That how
that easily how easily that can be captured as part of the
tool setups as part of the scripts or as part of the
recipes that they are put in. And usually that leads to a lot
of time resources at various skill sets that are needed.
And yeah, I'm not saying that we have seen this happening, but
what we also seen is that it usually turns out to be very
costly. And it becomes a effort, a normization in itself in
this up and that's what we did is that we actually brought out
this tool set in our part of our solution, which basically,
it is very purpose built. I would say that it's like this is
not like any generic tool, but we've just focused on building
this picture for our B2V customer base, building the picture
of the installed base data. And it again has the same set of
steps like data profiling, they're cleaning, deduping and
reaching unifying. And then what we have also got is that we
have put recipes in place. So what happens is that whenever we
get the data, say the initial set kind of thing. Of course,
like we had interaction with our customer because we have
to also understand the business domain, the way people work
that terminology they use, the categorization they use.
There might be two HV companies, but two XS companies still
are two very different companies in terms of a lot of
those smaller details. But what happens is that we captured them
as part of our recipes. And once it is captured, then then
as per the continuous process, whenever the new data comes in
that we just keep harmonizing it against the existing data
model that we have created. So I'll just put a small picture
in place, which gives the overall sense of how the whole
arc looks like. So again, like it's just again, like now you
would have been very comfortable with like how the whole
process is working. But on the left side of the thing, you have
like all the data silos from which the data would be picked up.
And then what happens is that it goes through this data quality
engine, which is called the install based data studio. And then
we do all the analytics and AI. And then it is served through
web and mobile. You can you now have your complete 360
install based visibility may not be today is not the right
day, we are not part of that one, but yes, at some point we'll
probably show our web and mobile interfaces are not weeks
each year. And then this is a complete open architecture. We
have like we can integrate it very clean. We can integrate
it very natively with the Salesforce and with open APIs and
all. It's very easy to pull the data into any other system.
Just a sneak peek of how the application looks like and the
how the web looks like. If you see you would be seeing a map
basically in which now you can actually visualize all the
all your install based. So if you see location here location is
basically a customer location. And you can go to the equipment
view and it will tell you equipment view and in fact, what we do
is that we actually pull all the possible bits and bytes of
that equipment and create a equipment 360. So you know that
to which I guess we should order it was purchased when it was
installed, those kind of thing, but also visiting it has been
done. What is the bomb associated to that one. And these
fields are many of these fields are editable like if someone
wants to track the hours there in the field. So with mobile
app and all this business very easy on the go when people are
visiting, they just update this one. But what it gives is
that it gives very clean and very powerful data points that
now can be feed into the AI models to figure out. So why we are
doing all this. We are doing all these things so we can figure
out the entire revenue opportunities. And that's where the
story comes to that's where the story starts are
lying in with the revenue goals basically. So we'll start
also figuring out which customers are at risk those kind of
things. But more important thing looking at those patterns of
conjunctions or parts and services. That system can now
predict that where we would see more opportunities now which
can be tapped. And of course, like there are other things like
propensity and also it will help out in prioritizing the
efforts as well. And we know that like Salesforce here is
probably one of the most big cases. See are them right now. So
what we have done is that we bring this so that so Salesforce
helps us in managing the whole workflow of the thing. But
just imagine if that whole install based data is available
of that particular customer as well. How powerful it can be
in terms of doing the conversation with the customer. And
that's what we have got where we serve this I be intelligence
data. So for example, just showing you a Q next communication.
But when you go here, there is this I be intelligence tab
which pulls the data from the install based data stores. But
you can now see all the overviews all the different
equipment parts services service contracts. Every day all
the data points are available a complete 360 view is available
to help the sales people doing a much better conversation
with the customer about their needs and predicting their
opportunities. And you will also see opportunities
tab. So we also have like different ways of creating
opportunities where there are even AI models that help
in creating opportunities. So yeah, I think with that I guess I
hope I am able to complete the picture on the I install
based visibility and how it helps in basically driving the
revenue further. I guess yes, and I think that's where we can
I don't know if there are any questions we can take that up.
Perfect. Thank you. Lullet. That was fantastic. Let me have a look.
I do see a couple of questions that have come in.
Let me take this one first. It's an interesting one looks at
the broader scope of the industry. It's what does future of
this industry looks like?
That's very interesting. We know that we see a couple of
trends out there. We definitely has seen a huge digital
transformation trend which is still going on. But what we
see is that there are two parts of the puzzle that we will
see a lot of things will need to happen. One is definitely
on the data side of the things to have a very clean data,
very unified, harmonized data so that the models can run
very efficiently. And then of course, the whole AI side of
the things. These two things are very interrelated to each
other, but that's where we see a lot of. Of course, like there
are workflows on top of them, but I think those workflows
would need a set of clean data and a set of very powerful
models so that these workflows can run very efficiently as and
that has to be like part of the playbook of any organization.
Right. That's true. As you always say, no data is like a
new titanium. It's all about using it wisely.
And moving on to the second question, this is a little
industry pertaining. This company works or manufactures
electrical engines. They sell it to OEMs. They do not really know
whether OEM sell that machine to is a any way we can locate
that install base. Okay. So yeah, that's the end customer
mapping problem. We see multiple times. Again, we have
done work even to achieve that when I'm not saying that it's
salvage straightforward, but there are ways to do that. We have
we have a couple of ways to handle those things. For
example, sometimes a service history has those information
sometimes there are regulations which allow which need the
companies to maintain the end customers and there are
different other databases which can be merged together to put
that picture in place. Yes, that's the problem that we deal
with that one and there are ways to solve that. 100% no, we
also talk. I would probably put a disclaimer that even we
have not solved with 100% but yes, with bits and pieces, we
have been able to do a fairly good job in piecing that picture
with that. Perfect. And that's to thanks to the question. Thank
you Lolli. I think that was the last question we have. Okay,
then thank you very much for your time and a big thank you to
everyone who joined in. Folks, be sure to check our next event
happening on July 24. They can scan the QR code to hop on to
our website. You can type on www.entitle.com. But that's it for
today guys. So thank you for tuning in. Take care and have a
nice one. Yeah, thank you. Thanks everyone for joining.