How Big Data Will

Transform The Mobile

Ecosystem
Alex Cocotas | June 4, 2013

How Big Data Will Transform The

Mobile Ecosystem
Alex Cocotas | June 4, 2013

The world is awash in data.

CIBC, a Canadian bank, predicts that information-generation
growth will increase 50 times over the next decade.
IDC, a market research firm, similarly forecasts a 44-fold increase
in data volumes between 2009 and 2020. Mobile is playing a large
part in driving this explosion in data.
Apple upended the electronics business six years ago with the
release of the iPhone. The iPhone ushered in an era when design,
both of software and hardware, became the paramount concept in
the tech world.
Could data be the paradigm that anchors the next revolution?
Suhail Doshi, CEO and co-founder of Mixpanel, thinks so. "Data is
the next design. Products that don't consider data will flounder,"
he says.
"Big data" has become the catch-all buzzword to describe this
phenomenon. Unfortunately, like many buzzwords, it is advanced
by some as a cure-all, even as its exact meaning remains obscure.
Poor customer relationship management? Big data! Need better
advertising? Big data! Achy joints and trouble sleeping at night?
Big data!
Part of the problem is that the term's meaning is vague. It doesn't
really mean anything at all on its own. Big data is most commonly

Copyright 2013, Business Insider, Inc. All rights reserved.

defined as data sets that meet three attributes, known as the three
"Vs": volume, variety, and velocity. It's data that is generated
quickly, comes in all shapes and sizes, and in great quantities.
But, as we can see, the three Vs themselves are simply adjectives
and also somewhat vague.
But this is not to say that this enormous amount of data is
valueless, far from it. Big data holds incredible potential across a
range of industries, not just mobile.
In this report we'll define big data, outline mobile's connection to
it, and point to some of its practical applications and pitfalls. Along
the way, we'll answer some of the most frequently asked questions
about big data and mobile.
Click here for the charts and data associated with this report in
Excel

Copyright 2013, Business Insider, Inc. All rights reserved.

What is Big Data?

Internet-connected mobile devices underpin the growth

of data generation.
The term big data is clouded by the imprecise way it is often used
in business conversations.
Duncan McCall, CEO of PlaceIQ, draws a comparison with cloud
computing. Cloud computing is basically just a network, and it
simply means stuff stored on a computer server somewhere, when
it comes down to it. Yet the term "cloud computing" is frequently
invoked in soaring tones, as if the very technology possessed
magical properties.
Likewise, big data is just that: enormous and exponentially
growing amounts of data.

Copyright 2013, Business Insider, Inc. All rights reserved.

Big Data is not defined by any particular volume or size threshold.

As we've noted, it is defined by the three Vs: velocity, volume, and
variety large amounts of disparate data coming in very, very fast.
But there is something more to it.

Facebook's data center in Prineville, Oregon

"I like to say there's a fourth V: value," says Kipp Jones, vice
president of product at Skyhook.
In order for data to be meaningful at all, it needs to be captured
and stored efficiently. Then someone has to manage the data,
analyze it, and extract value from it. Data, big or not, doesn't add
up to anything worthwhile if it doesn't have value to someone.
Unstructured Data
Big data is also sometimes used to refer specifically to the deluge of
unstructured data that has been created in the past decade or so.

Copyright 2013, Business Insider, Inc. All rights reserved.

Unstructured data is best understood by contrasting it with the

structured kind, which refers to data that is captured and
processed according to a pre-specified scheme.

Structured data: In a recent report, JP Morgan identifies stock

trading as an easy-to-understand example of structured data: each
trade has a stock ticker, price, time, and volume associated with it.
Structured data is most commonly associated with business
applications, and thus is frequently used for the optimization of
business processes.

Unstructured data: This refers to data that is not easily

captured through traditional schema. According to CIBC, 80% of
the world's data is unstructured. Text messages, video files,
machine data (such as web server logs), and social media feeds are
examples of unstructured data.

Unstructured data can be further split up between

human-generated data and machine-generated data.
o

Human-generated data is spurred by a

user's action, such as sending out a tweet.

Machine-generated data is automatically

generated from devices, perhaps without
humans even being aware.

Machine-generated data is growing faster than human-generated

data.

Big Data and Mobile

Mary Ellen Gordon, director of industry insights and analysis at
Flurry, offers a nice framework to help differentiate big data from
what came before: "I think about it as the difference between a
sample and a census."

Copyright 2013, Business Insider, Inc. All rights reserved.

In other words, big data doesn't just offer a snapshot of what some
representative cohort of people are doing, it offers the chance to
mine millions or even billions of direct observations on
what everyone is doing.
Flurry, a mobile app analytics company, previously would have
surveyed users about what apps they are using, but now they are
collecting this information in real-time. "You actually have data on
all those people," adds Gordon.
Flurry currently collects data from 1 billion devices and 300,000
apps. Each time an app is used it generates an app session and
each session produces further data.

Big data is not exclusive to mobile, or even the Internet.

Companies across many industries collect large amounts of data.
Car companies, for example, now track their vehicles' performance
to improve engineering. Big data is also being used in genetics to

Copyright 2013, Business Insider, Inc. All rights reserved.

mine enormous DNA databases with the potential for

groundbreaking medical breakthroughs.
While big data spans industries, mobile is particularly well-suited
to a big data lens.
"Mobile lends itself really well to big data," says Gordon, of Flurry.
"Mobile devices at their core are sending data back and forth. And
now theres more than a billion of us walking around with those
devices.
Mobile big data isn't only a function of smartphone penetration
and consumer usage patterns. The data is also created by apps or
other services working in the background.
"It's being created by those devices," says McCall, of PlaceIQ, "but
more data is being created to serve them."
Even when we are ostensibly not using our phones, we are still
creating reams of data.
Technically speaking, its not that different from data created using
the traditional Web.
Factual founder and CEO Gil Elbaz told us, "The infrastructure
that you need to handle [mobile data] might not be that different
from any other kind of application."
Although it is getting a lot more media attention these days, those
we talked to say that big data is not necessarily new per se
companies have always dealt with enormous data sets which they
mine for insights.
The difference is that consumers are just producing more of it.

Copyright 2013, Business Insider, Inc. All rights reserved.

"It's just this continuum of more and more data, as society shifts
from the analog to the digital world," says McCall.
In other words, as we shift our behavior to digital channels, we
leave a trail of data documenting our movements and actions.
Not everyone is a fan of the big data nomenclature. "I think it
masks the purpose," said Doshi of Mixpanel. "Just because you
have a ton of data doesn't mean you're doing anything with it."
In other words, the adjective big in big data might be more a sign
of its weakness than its strength.
"There are diminishing returns on the amount of data you harvest
and what it can do for you," adds Doshi.

Just How Much Data Are We Talking About?

Copyright 2013, Business Insider, Inc. All rights reserved.

While consumers may have some sense of their computer's data

storage or output, we are collectively putting out so much data it's
difficult to wrap our minds around it.
IBM reports that 90% of the data that now exists in the world was
created in the last two years alone.
IDC estimates that data volume will grow from 2.7 zettabytes (that
would be trillions of gigabytes) in 2012 to 7.9 zettabytes in 2015 to
35 zettabytes in 2020.
What's driving that traffic?

Internet-connected mobile devices play a huge role. Not only from

the enormous growth in gadget shipments and device penetration,
but greater use among current owners, which is driven by stronger
processors, more storage, faster networks, and the general growth
of the app ecosystem. Per Cisco, mobile network connection speeds

10

Copyright 2013, Business Insider, Inc. All rights reserved.

more than doubled in 2012 and average smartphone usage grew

81%.
Cisco also found that global mobile data traffic grew 70% in 2012.
Just mobile data from last year was more than 12 times the size of
the entire Internet in 2000.
It's going to keep growing exponentially: 4G connections generate
19 times more traffic than non-4G connections. Although they
account for less than 1% of mobile connections globally, 4G already
generates 14% of mobile data traffic. These numbers will get
another boost when 4G is rolled out in China, the largest
smartphone market in the world, later this year.
It's not just smartphones, however. Cisco found that the average
tablet generates 2.4 times the traffic of an average smartphone.
With tablet sales in the midst of an enormous expansion, they will
generate an increasingly greater share of global data traffic.
Mobile video accounts for the bulk of mobile traffic. According to
Cisco, video accounted for 51% of mobile traffic last year, a true
torrent of unstructured data. Cisco predicts that video will grow to
represent 66% of mobile data by 2017, even as total mobile data
traffic grows twelve-fold.

So, How Are We Collecting All This Data?

There are several common sources.

Software development kits (SDKs): Essentially SDKs are

lines of code inserted into an app, which allow a company like
Flurry or Mixpanel or Skyhook to pull out the app's analytics.

Partnerships/Open Source: Many companies are willing to

open up, or sell, their enormous archives of data. Factual, for
example, gets a lot of its information from its own customers.

11

Copyright 2013, Business Insider, Inc. All rights reserved.

However, the data is often noisy and unclean it may be riddled

with inaccuracies and Factual uses statistical methods to clean it
up. Companies can also plug into the APIs of huge social feeds like
Facebook or Twitter, and even tap into government databases.

Proprietary: After Factual cleans up the masses of noisy data it

has, it becomes proprietary data. There are other forms of
proprietary data too: PlaceIQ creates "polygons" of important
places for data-mapping purposes. Of course, companies also use
their data to improve their own processes. Skyhook uses feedback
from location data to improve its own network. Google's primary
competitive advantage is its own unequaled cache of proprietary
data, which it uses to make its services more effective and
marketable

But How Will All This Data Be Used?

As much of our personal and business lives migrate into digital and
mobile, there's virtually no end to big data applications.

12

Copyright 2013, Business Insider, Inc. All rights reserved.

With the growth of fitness bands, for example, there is an

enormous amount of data that can be used to improve personal
fitness. Nike FuelBand, for instance, records all your workout
information, graphs it, and then tries to motivate you to compete
with your friends and improve your performance.
Other applications are less immediately apparent, but also useful:
Google can mine search queries and their location data to map flu
or other disease outbreaks, or the impact of disasters. A more runof-the-mill application is using data signals from cell phones to
accurately map vehicle traffic patterns.
In sum, mobile big data can be used for a dizzying variety of
purposes, but it is often used for the optimization and
personalization of mobile services. Another application with huge
promise is mobile advertising.
Optimization and personalization
App developers, for example, might use Flurry's analytics to
improve their apps. Retention is a key metric for developers.
Developers can compare their user retention numbers with all
other apps and apps within their own categories, to gain insight
into how they stack up, and what they might have to change to
improve their numbers.
Mixpanel wants to use mobile analytics to help clients deliver a
more personalized experience for their mobile customers. For
example, Mixpanel's clients might craft more personalized emails
or push notifications rather than blasting out generic mass
messages.

13

Copyright 2013, Business Insider, Inc. All rights reserved.

Factual operates a location database with 66 million businesses

and points of interest, which client apps can integrate into their
service in order to improve their own mapping and
recommendation capabilities.
Location
Location data is an essential component of mobile big data
perhaps the primary data type that differentiates mobile from
Web-based big data.
Location data is expected to help transform the mobile advertising
industry.
"Advertisers are looking to take advantage of interest based on
where you are, to bridge offline opportunities with mobile
experiences," says Factual CEO Elbaz.

14

Copyright 2013, Business Insider, Inc. All rights reserved.

Companies are leveraging the proliferation of location data by

mapping it, thus transforming it into a potent targeting tool.
PlaceIQ has created what it calls a tile system, geospatially
mapping mobile activity to create a comprehensive profile of a
neighborhood's mobile users. For example, the behavior of the
mobile population in Times Square is completely different
depending on the time of day, while other neighborhoods might
have more consistent mobile usage over the course of a day.
PlaceIQ's tile system can help facilitate the type of advertising they
should be shown.
Kipp Jones of Skyhook says that, using a similar method, they can
compare mobile behavior on city blocks in New York City and
Paris.
Even carriers are getting into the game and opening up their
massive troves of customer data to advertisers and marketers for
a price.
Advertising

Another view of Facebook's data center

Why are advertisers so eager to incorporate local data?

15

Copyright 2013, Business Insider, Inc. All rights reserved.

For one thing, studies have shown that mobile campaigns that use
local data outperform those that don't by a factor of two to three.
But, conceptually, high quality location data represents the holy
grail of mobile advertising. The promise of the mobile advertising
market, vis-a-vis the desktop, is that consumers carry around these
devices on them all day, so that it creates ideal occasions for ad
targeting.
The ability to deliver real-time hyper-local, targeted advertising
represents a potentially momentous evolution of the ad market.
Skeptics will argue that however effective the targeting, mobile
interactions are very brief, and the screens very small, hardly great
contexts for advertising.
Big data is expected to drive that change.
Big data also opens up opportunities to "close the loop," or
properly attribute consumer offline shopping behaviors to digital
advertising they have been exposed to.

For example, PlaceIQ ran a campaign for Starcom MediaVest

Group's retail clients that geo-fenced a certain area, showed a
sample group a mobile ad, and then tracked who entered the retail
store. Those who saw the ad were 50% more likely to visit the store
versus a control group who were not shown the ad. Among those
who saw the ad, those who clicked on it were 46% more likely to
visit the store.
Finally, the mass of location data can be used for that other
aspiration of mobile advertising: cross-device tracking.
Smartphones and tablets, which largely don't use cookies,
obliterated the advertisers' ability to create cohesive digital
profiles. As mobile ad company Drawbridge laid out in a recent

16

Copyright 2013, Business Insider, Inc. All rights reserved.

paper, algorithmic analysis of user data, including location and

temporal data, offers the key to tracking consumers from desktop
to mobile and vice-versa, unleashing the potential for even greater
targeting opportunities.
Long-term, uses of mobile big data are numerous and speculative.
For example, McCall, of PlaceIQ, posited that the enormous
amount of location data surrounding mobile data could improve
city planning. Factual's Elbaz cited anticipatory computing
computer-generated suggestions and recommendations as an
emerging use of mobile big data.

The Data Trap

Businesses need to be wary of the data trap: There is more data

than ever, yes, but mountains of data do not automatically
translate to better business decisions.

17

Copyright 2013, Business Insider, Inc. All rights reserved.

All that data can give you greater confidence, but Mixpanel's Doshi
says, "You don't need that much information to generate a
conclusion."
Data is important, undeniably, but picking the right data is even
more important.
"It's a constant battle," says Jones of Skyhook, "[In addition to] the
data we generate, we can start to combine all sorts of data sets
Twitter, Foursquare check-ins, Flickr image analysis ... The
amount of data feeds that you can plug into is somewhat
daunting."
Big data is sometimes about eliminating options, not plugging into
yet another torrent of data.
"You have to look ... and figure out the best value," Jones says. "If
you're a researcher and you just want to play with data, this is a
great time. But if you're a business, the tendency may be to grab
more data than you need, and that's not the best strategy."
Finding the right data
Picking the right data typically means defining a target or goal and
subsequently establishing disciplined parameters for the data you
want to collect or process.
"It doesn't necessarily have to be huge data to get value," adds
Jones, "Look at retailers, it may just be related to a store or a set of
stores."
Doshi pushes his clients to find a single key metric central to their
business and building out from there.

18

Copyright 2013, Business Insider, Inc. All rights reserved.

"For Amazon, they might want to understand how people are

buying and why they're buying everything is centered on
buying."
Once the central metric is established, companies can begin to ask
tangential questions to help sharpen their understanding of what is
driving the key metric. Amazon, for example, would want to know
what browsing habits lead to purchases, which traffic sources have
the highest conversions, etc.
Secondary sources of data should branch off from the central
metric, and be understood in relation to it.
Finally, data can sometimes be mirage-like, tempting businesses
into making decent piecemeal decisions (which might be positive
in the short-term), but ultimately miss the big picture and the
greater opportunity.

THE BOTTOM LINE

There is no exact definition of big data, but broadly it refers

to enormous amount of unstructured data that are being
produced in large volumes, at high velocity, displaying a
huge variety of forms and content.

Mobile is playing a big part in driving this data explosion.

Smartphones are personal sensors constantly transmitting
data, often when their users are not even actively using their
devices.

Broadly, mobile big data is often being used by companies

to optimize their products and create more personalized
experiences for their customers.

19

Copyright 2013, Business Insider, Inc. All rights reserved.

The biggest potential for big data may lie in mobile

advertising, however. Big data will help mobile ads deliver
on two promises: hyper-local targeted advertising and
cross-device tracking.

Big data is not an across-the-board cure-all. Data for the

sake of data is the wrong approach. Too much data can
obscure overall goals. Businesses should focus on one or two
key metrics at the heart of their strategy and branch out
from there.

20

Copyright 2013, Business Insider, Inc. All rights reserved.

About BI Intelligence
BI Intelligence is a new subscription research service from Business Insider that
provides in-depth insight, data, and analysis of the mobile industry. We publish notes
and in-depth reports about the business of the mobile industry as well as an
exhaustive library of charts and data that will help you stay on top of the key trends in
the mobile ecosystem. We help our subscribers make smarter strategy decisions.
To learn more or to sign up for a free trial of the service, please visit
intelligence.businessinsider.com.
Analysts
BI Intelligence has an experienced team of analysts led by Henry Blodget, CEO &
Editor-in-Chief of Business Insider. BI Intelligences team of dedicated analysts have
deep analytical and industry experience, and work with the Business Insiders 50+
journalists covering specific verticals, such as technology, advertising, and strategy, to
produce unique insight and analysis on the mobile ecosystem.
Copyright 2013 Business Insider, Inc. All Rights Reserved.
Proprietary and Confidential Property of Business Insider, Inc.
Licensed for Use By BI Intelligence Subscribers Only.
Access to and use of this proprietary and confidential information is limited by the
terms of conditions.

21

Copyright 2013, Business Insider, Inc. All rights reserved.