PWC Develops Analytics Platform to Improve Retail, Media, eCommerce Insights on Road to Deep Predictive Modeling
PWC Develops Analytics Platform to Improve Retail, Media,
eCommerce Insights on Road to Deep Predictive Modeling
Transcript of a BrieﬁngsDirect podcast on how HP tools are changing the face of data analytics
in the era of big data.
Listen to the podcast. Find it on iTunes. Sponsor: HP
Dana Gardner: Hello, and welcome to the next edition of the HP Discover Podcast Series. I’m
Dana Gardner, Principal Analyst at Interarbor Solutions, your host and moderator for this
ongoing sponsored discussion on how data is analyzed and used to advance the
way we all work and live.
Once again, we’re showcasing how thought leaders and innovative companies
worldwide are capturing myriad knowledge, gaining ever deeper analysis, and
rapidly and securely making those insights available to more people on their
Our next innovation case study focuses on the growing needs among business verticals for a
powerful, fast, and targeted data-analysis capability, and we will explore how one technology
solutions provider, PricewaterhouseCoopers LLC (PwC), is crafting just such a set of services.
We’ll now learn how PwC has connected the dots between data access,
transformation, integration, and analysis to provide rapid and recurring insights
for such industries as retail, media, healthcare, and e-commerce.
The goal is to develop deep predictive modeling capabilities for such business
activities as dynamic real-time pricing and responsive consumer services.
To learn more, about how high-performing and cost-effective big-data processing is coming to
market as a vertical industry solution, please join me now in welcoming our guest.
We’re here with Dhiraj Pathak, he is the Director of the CloudLab at PwC in New York.
Dhiraj Pathak: Hi, Dana. Good to be with you.
Gardner: Glad to have you with us. Let's start at a high level. How is big-data analysis changing
to such a degree that we can now make it into something that’s appropriate for vertical industry
solutions? How is it that things have changed so that we can now take what was basically a
technology and make it into a business service?
Pathak: Well, Dana, a couple of things are happening. As clients are becoming more-and-more
digital in nature and transacting with their users using digital channels, a lot of the business
performance around these interactions is tied to an ability to process the data that’s underlying
these interactions in meaningful ways, and in ways that can be turned into business insight, can
be used for decision-making.
We’re witnessing a shift from looking at big data as a technological or a technical capability to a
business need, which then has to be addressed and resolved.
Gardner: What's interesting to me is that we’re not just taking a one-size-ﬁts-all approach to
this data capability. We’re looking at the business requirements vertical-by-vertical, perhaps even
company-by-company and starting to adjust how we extract analysis in such a way that it's of
greater value to that vertical. How are we able to do that granular or niche approach to analysis?
Pathak: That’s a very critical observation, Dana. The core of that is that meaning from data
cannot be derived unless you put the right context to it. To be successful with
big-data analysis and turning that data into business insight, what's critical is
to combine it with enough context of the business area that’s focused on
honing in on aspects of the data and asking the right questions that allow for
that value to come out of that data.
Gardner: I suppose another big change nowadays is around velocity or
speed. We've been able to take big chunks of data, do batch processing, and
extract analysis and reports. But we’re now at the point where we can take
activity in real-time, like a customer in a retail environment or someone on a website doing e-
commerce, and deal with vast amounts of data and then react to them. How is this speed element
impacting the market?
Pathak: The speed element is essentially based having on this mobile device with you all the
time, as all of us are now adopting. That’s leading to a need for an analysis capability. But ofﬂine
analysis is not going away, because that's where the deep insights come from. That’s where the
predictive analytics happen.
What's also becoming extremely crucial is to serve up the results of those analyses and to apply
those analyses in real-time. As you said, imagine a shopper on a e-commerce site around
Christmas, making a last moment purchasing decision. They’re looking for insight. They don’t
necessarily know what they might be what they might become interested in.
To the extent that that e-com site is able, based on a knowledge of who that customer is and
based on the behavior that the customer begins to show on the site, if an insight or a
recommendation can be made in context, in real-time, that’s where the value is. That then leads
to a different set of requirements around how to process big data in order to be able to produce
those insights in the moment.
Gardner: How signiﬁcant is this likely to be, this velocity increase, the amount of data being
brought to bear on these businesses? Are we talking about incremental improvement, or is this
something that could be rather dramatic, something catalytic even?
Pathak: For a period of time it will have a fairly dramatic kind of an impact, before it's all
settled down and then that becomes the new normal and that’s how businesses interact with their
But for the next two to three years, I think that we’ll see dramatic shifts and dramatic examples
of how businesses are able to inject a lot of momentum into their interactions with their
customers. They’re able to process the underlying data factor and are able to turn that data into
insights that get presented in the course of a transaction or in the course of that interaction that
the customer is having with that business.
To the marketplace
Gardner: Well, you've certainly piqued my interest. Let's learn a bit more about PwC and how
it's working in the market to exploit these technologies to make them into discrete business
services. Tell us how we got to where we are in terms of you’re taking this out to the
Pathak: As I said before, one of the most dominant trends that we’re observing with our client
base is that all of our clients, or at least most of our clients, are turning into digital businesses. If
you look at our retail clients, more-and-more of the transactions that these businesses are having
with their users are happening on digital channels, on e-commerce sites, but it's not restricted to
just the retailers.
If we look at hospitals, banks, or media companies, anywhere you look, businesses are creating
capabilities and services that are digital, where people are interacting with these services using
their smart devices, their phones, tablets, etc. In fact, in many cases, there is a whole new
emerging area around variable technology, where people are monitoring themselves for different
If that’s the market opportunity that we've become squarely focused on, the requirement around
this is to be able to process the data that gets generated when customers and end users are
interacting in a digital way.
Increasingly, it's about how to turn that data into a couple of different things, and turn that into a
better shopper experience. Whoever is interacting with business is getting a sense that they are
being served as a category of one. To the extent that there is that personalization and a sense for a
white-glove treatment, that creates an enormous stickiness factor and an enormous competitive
differentiation for that business.
The other aspect of it is that all of this data that’s getting aggregated at an enormous rate and in
very large quantities is allowing for analytics of a very different kind to get applied to this data.
Those are the two requirements that many of our clients are having. What we've been doing is to
create platform capability to help accelerate the business’s ability to look at its data and convert it
into these kinds of responses that they’re looking to create.
Gardner: Is PwC showing them the way to build their own data capabilities or are you focusing
on a vertical industry, where you can bring your own data, analysis, and aggregation to bear? Are
you teaching them to ﬁsh or are you going to be doing some of this analysis for them as a
service, perhaps as a cloud service?
Pathak: The way we’re going into market around this platform capability is to do what makes
the best sense for our clients. In some cases, we’re working with clients who are viewing us as a
service bureau, where we take on their data and turn it around, combine that data that comes
from a particular enterprise with more data that’s available outside, and provide a result back as a
But in other cases, there are a number of clients we’re working with for whom their interaction
with PwC is more to bring us in to help create that capability more organically within their own
So we see the full spectrum of delivery methods around this kind of capability, but the common
factor is that the need is emerging very rapidly in all of the different sectors we’re working with
and with clients both big and small.
Hurdles and roadblocks
Gardner: Dhiraj, you mentioned earlier that there has to be a new level of technology
capability in order to accommodate the speed, the scale, and the responsiveness. Tell us about the
problem. What has been the hurdle or the roadblock to being able to do this data analysis at scale
and at velocity up until now?
Pathak: There are several problems. It is important to recognize that there has been a pivoting or
a transition in the last two to three years, where big data has suddenly really become a reality.
About 12 to 18 months ago, people said they were working with big data, but clearly they were
not truly working in big data. They just had large data sets, and traditional analytics technologies,
data management technologies, and infrastructure technologies were at least technically
sufﬁcient for that situation. However, there was still quite a bit of heavy lifting that was required
to turn that data into insight.
Now, truly big data, where we’re seeing businesses really getting into the petabyte range, the
problem is several fold. It's not that there is a ﬁrst problem and then there is a second problem,
but there is a whole bevy of problems here.
Some of the things that make up this complexity are ﬁrst getting a grip on how can you harness
the data -- the what question, before getting into the how question. Many of our clients are now
at the point where they can clearly see that they are creating this enormous data wealth, but
they’re not able to see what to do with it.
So there is that issue, and in many cases, we go in and we work very collaboratively with our
clients and help work through what the opportunity might be around the data that they’ve got
available to them.
After that issue has been addressed, the next thing is a ground reality, which is that the available
data, in most cases, is not ready to be mined. There is a big gap between the data as it exists and
needing to be converted or transformed into a shape and form that analysis can be done.
In many instances, if you look at the overall effort required from identifying what you want to do
with a big data set and then ending up with some value from that big data set, 60 to 80 percent of
the effort would go into just doing various things to pull all that data together and put it into a
form that you can do analytics on.
Then, when you come to that point, there is another an issue looming in front of you. In most
instances, it would require the assembly of many different kinds of analytics engines. If it's
structured data, then you have to approach it in a certain way. If it's unstructured data, you have
to approach it in another way.
Depending on the type of data, it could require a traditional data warehouse kind of an approach.
If it's more unstructured, it might require some of the newer distributed ﬁle systems kind of
approaches. All of that technology that’s required to process the data and convert it into value
becomes the third bottleneck.
What makes sense
Those are some of the issues that we’re helping to resolve for our clients by ﬁrst ﬁguring out
and coming forward with, in an industry speciﬁc way, a framework of what kind of questions
make sense in speciﬁc industry sectors.
Then, it’s following it up with a platform capability that does all of the heavy lifting around how
to stream data into an infrastructure and how to apply different kinds of ways of analyzing the
data. All of that tooling has gotten put together in a platform, and then that helps to accelerate the
time to value.
Gardner: Given this very complex interrelated set of requirements, you also need to create a
platform through which you can re-use that platform for a variety of different verticals with a lot
of different business issues and requirements. When you looked for the right platform to suit
your needs, where did you go and what did you ﬁnd?
Pathak: We had some very speciﬁc requirements. Having done big-data analysis work with our
clients across industry sectors for a while, we had started to converge on some core requirements
around the right technology building blocks.
One of these core requirements was the need for openness and a platform or an architecture that
would allow for a plug-and-play, where capabilities could be dropped in and taken out as needed.
The second major requirement that we had was a platform that could scale.
Big data, as I said before, is in that petabyte-plus range and that’s only the beginning. You have
to have an architecture that’s designed for scale and scale in terms of the data itself, but also in
terms of processing.
As we talked about before, a lot of the big-data value proposition is now tied to being able to do
it in real time and to effect, in the moment, context decision-making. So it's both about large-
scale data storage, large scale computation.
Then, the third requirement that we had established was an architecture that had a built-in
support for connecting to different systems and different sources of data.
If that support isn’t there, that typically translates into a lot of ad-hoc work that has to be done in
marshaling the data, putting it together, and having it ﬂow in the right-way. Then, even after
you've achieved that objective, you’re stuck with the fairly daunting task of maintaining all of
that connectivity to data over time.
Gardner: Getting back to the use of this in the market, are you in the proof-of-concept (POC)
phase? Do you have a steady set of examples? Can you give us some sense of when you do this
well, building it on the right platform bringing it into the business vertical in a context that’s
suitable to their real world needs, what paybacks we’re seeing and what efﬁciencies or
innovation are gained?
Early launch phase
Pathak: We’re in the early cycle of this, an early launch phase. We’re working with a number
of clients who have big-data problems that they want resolved. The early payback is our ability
to do a POC and move quickly to the point where we’re having a real meaningful interaction
with a client around their data and starting to show them what value is hidden in that data. We’re
ﬁnding that we’re able to now do that proving out of the value from data much faster.
In the last 60 days or so since we've launched this solution, we’re now in front of at least six or
eight clients who are looking at this capability. They’re seeing the value of having a platform that
they can connect into, where the discussion is around what your data is and what you want out of
It completely circumvents the discussion around what tools you need to buy, what infrastructure
you need to have, how would this all be integrated, and how long of a system’s integration kind
of an exercise would have to be undertaken before you can begin to see what your data can
actually tell you.
Early in the analytics lifecycle, that acceleration is very crucial and sets the stage for whatever
comes next. That’s where we’re seeing the early value. We’re also projecting that, and this is a
little bit more futuristic, based on real client feedback, the next logical step on this is to provide
this platform as a service (PaaS), where the data can simply be streamed into the platform and
end results be provided to the end customer.
Right now, the value that we are seeing in this is accelerating our client’s ability to establish a
value proposition around their big data sets and being able to provide them those early insights,
which are then shaping their strategy around what to do next.
Gardner: Terriﬁc. I'm afraid we’ll have to leave it there. It's been a very enlightening discussion.
We've been learning about how technology solutions provider PwC is crafting a powerful, fast,
and targeted data analysis capability for such industries as retail, media, healthcare, and e-
And we've learned how this will soon lead to deep predictive-modeling capabilities and a
reduction in the time to value, while in a sense masking a lot of the technological hurdles along
So I want to extend a huge thank you to our guest, Dhiraj Pathak, he is the Director of the
CloudLab at PwC in New York. Thanks so much, Dhiraj.
Pathak: Thank you, Dana. Good to have talked to you.
Gardner: And also, a huge thank you to our audience, for joining this HP Discover Podcast.
I'm Dana Gardner, Principal Analyst at Interarbor Solutions; your host for this ongoing series of
HP sponsored business innovation discussion. Thanks again for listening and do come back next
Listen to the podcast. Find it on iTunes. Sponsor: HP
Transcript of a BrieﬁngsDirect podcast on how HP tools are changing the face of data analytics
in the era of big data. Copyright Interarbor Solutions, LLC, 2005-2014. All rights reserved.
You may also be interested in:
Healthcare SaaS Provider PointClickCare Masters Quality and DevOps Using HP Service
Software Security Pays Off: How Heartland Payment Systems Gains Steep ROI Via
Software Assurance Tools and Methods
HP ART Documentation and Readiness Tools Bring Better User Experience to Nordic IT
Solutions Provider EVRY
Nimble Storage Leverages Big Data and Cloud to Produce Data Performance
Optimization on the Fly
MZI Healthcare Identiﬁes Big Data Patient Productivity Gems Using HP Vertica
Thought Leader Interview: HP's Global CISO Brett Wahlin on the future of Security and
Panel explains how CSC creates a tough cybersecurity posture against global threats
Risk and complexity: Businesses need to get a grip