How data will transform business

I'm going to talk a little bit about

strategy and its relationship with technology.

We tend to think of business strategy

as being a rather abstract body

of essentially economic thought,

perhaps rather timeless.

I'm going to argue that in fact,

business strategy has always been premised

on assumptions about technology,

that those assumptions are changing,

and in fact changing quite dramatically,

and that therefore what that will drive us to

is a different concept of what we mean

by business strategy.

Let me start, if I may,

with a little bit of history.

The idea of strategy in business

owes its origins to two intellectual giants,

Bruce Henderson, the founder of BCG,

and Michael Porter, professor
at the Harvard Business School.

Henderson's central idea was what you might call

the Napoleonic idea of concentrating mass

against weakness, of overwhelming the enemy.

What Henderson recognized was that,

in the business world,

there are many phenomena which are characterized

by what economists would call increasing returns,

of scale, of experience.

The more you do of something,

disproportionately the better you get.

And therefore he found a logic for investing

in such kinds of overwhelming mass

in order to achieve competitive advantage.

And that was the first introduction

of essentially a military concept of strategy

into the business world.

Porter agreed with that premise,

but he qualified it.

He pointed out, correctly, that that's all very well,

but businesses actually have multiple steps to them.

They have different components,

and each of those components might be driven

by a different kind of strategy.

A company or a business
might actually be advantaged

in some activities but disadvantaged in others.

He formed the concept of the value chain,

essentially the sequence of steps with which

a, shall we say, raw material, becomes component,

becomes assembled into a finished product,

and then is distributed, for example,

and he argued that advantage accrued

to each of those components,

and that the advantage of the whole

was in some sense the sum or the average

of that of its parts.

And this idea of the value chain was predicated

on the recognition that

what holds a business together is transaction costs,

that in essence you need to coordinate,

organizations are more efficient at coordination

than markets, very often,

and therefore the nature and role and boundaries

of the cooperation are defined by transaction costs.

It was on those two ideas,

Henderson's idea of increasing returns

to scale and experience,

and Porter's idea of the value chain,

encompassing heterogenous elements,

that the whole edifice of business strategy

was subsequently erected.

Now what I'm going to argue is

that those premises are in fact being invalidated.

First of all, let's think about transaction costs.

There are really two components
to transaction costs.

One is about processing information,
and the other is about communication.

These are the economics of
processing and communicating

as they have evolved over a long period of time.

As we all know from so many contexts,

they have been radically transformed

since the days when Porter and Henderson

first formulated their theories.

In particular, since the mid-'90s,

communications costs have actually been falling

even faster than transaction costs,

which is why communication, the internet,

has exploded in such a dramatic fashion.

Now those falling transaction costs

have profound consequences,

because of transaction costs are the glue

that hold value chains together, and they are falling,

there is less to economize on.

There is less need for vertically
integrated organization,

and value chains at least can break up.

They needn't necessarily, but they can.

In particular, it then becomes possible for

a competitor in one business

to use their position in one step of the value chain

in order to penetrate or attack

or disintermediate the competitor in another.

That is not just an abstract proposition.

There are many very specific stories

of how that actually happened.

A poster child example was
the encyclopedia business.

The encyclopedia business

in the days of leatherbound books

was basically a distribution business.

Most of the cost was the
commissions of the salesmen.

The CD-ROM and then the internet came along,

new technologies made the distribution of knowledge

many orders of magnitude cheaper,

and the encyclopedia industry collapsed.

It's now, of course, a very familiar story.

This in fact more generally was the story

of the first generation of the internet economy.

It was about falling transaction costs

breaking up value chains

and therefore allowing disintermediation,

or what we call deconstruction.

One of the questions I was occasionally asked was,

well, what's going to replace the encyclopedia

when Britannica no longer has a business model?

And it was a while before
the answer became manifest.

Now, of course, we know
what it is: it's the Wikipedia.

Now what's special about the
Wikipedia is not its distribution.

What's special about the Wikipedia
is the way it's produced.

The Wikipedia, of course, is an encyclopedia

created by its users.

And this in fact defines what you might call

the second decade of the internet economy,

the decade in which the internet as a noun

became the internet as a verb.

It became a set of conversations,

the era in which user-generated
content and social networks

became the dominant phenomenon.

Now what that really meant

in terms of the Porter-Henderson framework

was the collapse of certain
kinds of economies of scale.

It turned out that the tens of thousands

of autonomous individuals writing an encyclopedia

could do just as good a job,

and certainly a much cheaper job,

than professionals in a hierarchical organization.

So basically what was happening was that one layer

of this value chain was becoming fragmented,

as individuals could take over

where organizations were no longer needed.

But there's another question
that obviously this graph poses,

which is, okay, well we've
gone through two decades:

does anything distinguish the third?

And what I'm going to argue is that indeed

something does distinguish the third,

and it maps exactly on to the kind of

Porter-Henderson logic that
we've been talking about.

And that is, about data.

If we go back to around 2000,

a lot of people were talking
about the information revolution,

and it was indeed true that the world's stock of data

was growing, indeed growing quite fast.

but it was still at that point overwhelmingly analogue.

We go forward to 2007,

not only had the world's stock of data exploded,

but there'd been this massive substitution

of digital for analogue.

And more important even than that,

if you look more carefully at this graph,

what you will observe is that about a half

of that digital data

is information that has an IP address.

It's on a server or it's on a PC.

But having an IP address means that it

can be connected to any other data

that has an IP address.

It means that it becomes possible

to put together half of the world's knowledge

in order to see patterns,

an entirely new thing.

If we run the numbers forward to today,

it probably looks something like this.

We're not really sure.

If we run the numbers forward to 2020,

we of course have an exact number, courtesy of IDC.

It's curious that the future is so much
more predictable than the present.

And what it implies is a hundredfold multiplication

in the stock of information that is connected

via an IP address.

Now if the number of connections that we can make

is proportional to the number of pairs of data points,

a hundredfold multiplication in the quantity of data

is a ten thousandfold multiplication

in the number of patterns

that we can see in that data,

this just in the last 10 or 11 years.

This I would submit is a sea change,

a profound change in the economics

of the world that we live in.

Now what does that imply in terms of business?

Well, I got a hint of this some years ago.

Back in around 2003 or so,

I was doing some consulting for the Pentagon,

for august institutions on the subject

of network-centric warfare,

and in that context I met a
gentleman called Jeff Jonas,

a brilliant engineer who had made his fortune

designing the security systems in Las Vegas.

Jeff said to me, "Next time
you're in Las Vegas, Philip,

why don't you stop by and I'll take you on the tour.

You can meet Nora. Nora will show you a good time."

N.O.R.A. was not his girlfriend.

N.O.R.A. is the Non-Obvious
Relational Awareness system,

a realtime fraud control system developed by Jeff

which supports all of the casinos in Las Vegas.

We were in the security room

of the Bellagio Hotel in Las Vegas,

and on the monitor I saw this happen.

A woman was playing Blackjack against the dealer.

There was nobody else at the table.

She was winning too much.

They know how likely that is, and this wasn't likely.

So the first thing they did was
they use facial recognition,

see if she's staying at the hotel. She wasn't.

Then they can kind of run the cameras backwards,

tracing her movements back through the hotel

to the parking garage, where they found her car.

They could then run N.O.R.A.
to find who owned the car.

The car was owned by Hertz Las Vegas.

Within a second or so, N.O.R.A. pulled down

the Hertz Las Vegas application.

Now they knew who the woman was.

Where was she staying?

Well, they pooled the data across the hotels.

It turned out she was staying
in a hotel across the street.

Had she gambled in that hotel? No.

Very strange behavior, staying in one hotel,

gambling in another.

Then came the really interesting thing.

N.O.R.A. looked for a connection

between the woman and the dealer,

because a very high fraction of fraud in Las Vegas

is committed when the staff are actually

in illicit collaboration with customers.

It turned out, what N.O.R.A. did was to look through

6,000 databases, public and private,

some owned by the Bellagio,

some by other hotels,

some police records, and so on.

It turned out that 10 years earlier,

this woman's brother had
been the dealer's roommate.

And it took N.O.R.A. six
seconds to work that fact out.

It cost the woman and the dealer six years.

This was N.O.R.A. in action.

It's what today of course we would call Big Data,

long before the term had been formulated.

Now notice some very interesting things about this,

most of all the fact that N.O.R.A.

runs as a cooperative across the entire of the strip.

These casinos, which are otherwise

competing aggressively with each other

actually collaborate when it comes to the management of their security systems.

They pool data into a common database

that is run essentially as a co-op

for this specific purpose.

Why? Because the scale of N.O.R.A.,

what N.O.R.A. is trying to do,

blows past the scale

that even a very large casino
can possibly do for itself.

The value chain is not big enough

to accommodate the economies of scale

that are inherent in this particular activity.

And that principle, I would suggest,

is actually a fundamental and pervasive one.

In essence, what happens is that because

of these colossal economies of scale in data,

what used to be value chains that ran separately

are compelled, in order to
achieve those economies of scale,

to create some kind of common utility,

some common resource, a co-op,

a pool, a vault of data within which

those insights can be gathered.

Now N.O.R.A. is a relatively trivial example

in the sense that if N.O.R.A. failed,

it wouldn't exactly be the end of civilization.

But consider something vastly more important,

where the logic in fact is exactly the same,

the logic of health care.

The first human genome,

that of James Watson,

was mapped as the culmination of the
Human Genome Project in the year 2000,

and it took about 200 million dollars

and about 10 years of work to map

just one person's genomic makeup.

Since then, the costs of mapping
the genome have come down.

In fact, they've come down in recent years

very dramatically indeed,

to the point where the cost is
now below a thousand dollars,

and it's confidently predicted that by the year 2015

it will be below a hundred dollars,

a five or six order of magnitude drop

in the cost of genomic mapping

in just a 15-year period,

an extraordinary phenomenon.

Now, in the days when mapping a genome

cost millions, or even tens of thousands,

it was basically a research enterprise.

Scientists would gather some representative people,

and they would see patterns, and they would try

and make generalizations about
human nature and disease

from the abstract patterns they find

from these particular selected individuals.

But when the genome can be
mapped for a hundred bucks,

99 dollars while you wait,

then what happens is, it becomes retail.

It becomes above all clinical.

You go the doctor with a cold,

and if he or she hasn't done it already,

the first thing they do is map your genome,

at which point what they're doing

is not starting from some abstract knowledge

of genomic medicine

and trying to work out how it applies to you,

but they're starting from your particular genome.

Now think of the power of that.

Think of where that takes us

when we can combine genomic data

with clinical data

with data about drug interactions

with the kind of ambient data that devices

like our phone and medical sensors

will increasingly be collecting.

Think what happens when we collect all of that data

and we can put it together

and use precisely the N.O.R.A.-type techniques

in order to find patterns we wouldn't see before.

This, I would suggest, perhaps it will take a while,

but this will drive a revolution in medicine.

Fabulous, lots of people talk about this.

But there's one thing that
doesn't get much attention.

How is that model of colossal sharing

across all of those kinds of databases

compatible with the business models

of institutions and organizations and corporations

that are involved in this business today?

If your business is based on proprietary data,

if your competitive advantage
is defined by your data,

how on earth is that company or is that society

in fact going to achieve the value

that's implicit in the technology? They can't.

So essentially what's happening here,

and genomics is merely one example of this,

is that technology is driving

the natural scaling of the activity

beyond the institutional boundaries within which

we have been used to thinking about it,

and in particular beyond the institutional boundaries

in terms of which business strategy

as a discipline is formulated.

The basic story here is that what used to be

vertically integrated, oligopolistic competition

among essentially similar kinds of competitors

is evolving, by one means or another,

from a vertical structure to a horizontal one.

Why is that happening?

It's happening because
transaction costs are plummeting

and because scale is polarizing.

The plummeting of transaction costs

weakens the glue that holds value chains together,

and allows them to separate.

The polarization of scale economies

towards the very small, small is beautiful,

allows for scalable communities

to substitute for conventional corporate production.

The scaling in the opposite direction,

towards things like Big Data,

drive the structure of business

towards the creation of new kinds of institutions

that can achieve that scale.

But either way, the typically vertical structure

gets driven to becoming more horizontal.

The logic isn't just about Big Data.

If we were to look, for example,

at the telecommunications industry,

you can tell the same story about fiber optics.

If we look at the pharmaceutical industry,

or, for that matter, university research,

you can say exactly the same story

about so-called Big Science.

And in the opposite direction,

if we look, say, at the energy sector,

where all the talk is about how households

will be efficient producers of green energy

and efficient conservers of energy,

that is in fact the reverse phenomenon.

That is the fragmentation of scale

because the very small can substitute

for the traditional corporate scale.

Either way, what we are driven to

is this horizontalization of the structure of industries,

and that implies fundamental changes

in how we think about strategy.

It means, for example, that we need to think

about strategy as the curation

of these kinds of horizontal structure,

where things like business definition

and even industry definition

are actually the outcomes of strategy,

not something that the strategy presupposes.

It means, for example, we need to work out

how to accommodate collaboration

and competition simultaneously.

Think about the genome. Think about N.O.R.A.

We need to accommodate the very large

and the very small simultaneously.

And we need industry structures

that will accommodate very,
very different motivations,

from the amateur motivations
of people in communities

to maybe the social motivations

of infrastructure built by governments,

or for that matter cooperative institutions

built by companies that are otherwise competing,

because that is the only way
that they can get to scale.

These kinds of transformations

render the traditional premises of business strategy

obsolete.

They drive us into a completely new world.

They require us, whether we are

in the public sector or the private sector,

to think very fundamentally differently

about the structure of business,

and, at last, it make strategy interesting again.

Thank you.

(Applause)