Big Ideas from Big (or Small) Data

Big Ideas from Big (or Small) Data Book Summit Canada Pete McCarthy The Logical Marketing Agency

Big Ideas from Big (or Small) Data | Book Summit Canada June 19, 2014 2

Who am I and why am I here?


What are we talking about and why are we talking about it (now)?

We are talking about big ideas.

June 19, 2014 Big Ideas from Big (or Small) Data | Book Summit Canada 4

Really, a process which may yield big ideas. Discussion of data is highly probable.

It is a capital mistake to theorize

before one has data. Insensibly one

begins to twist facts to suit theories,

instead of theories to suit facts.

– Sherlock Holmes, A Scandal in Bohemia

This is a big idea!


94% accuracy of opening weekend box office up to 4 weeks pre-‐release…

2013

So was this and seems to still be.


97% correlation between “Twitter chatter” and opening weekend box office.

2010

Especially when combined with this work.


Which adds (a little) more (seemingly correct) data to eliminate bias.

2012

This might be part of a big idea…


77% “predictive.” Backward-‐looking. Reliability of data?

2012

2013

1983

These were big ideas…and some still are…


Most big ideas build on prior big ideas – successful or not.

2010

2010

2002

2000

1994

Why we are here.


Because of what Google (and others) do. Because we can do similar things.

ü  What ü  When ü  Where ü  Which

ü  Who ü  How ü  Even a plausible

why!


What we talk about when we talk about consumer data

In essence, we are talking about useful research.


Some “types” of consumer research and the methods used.

Secondary

Industry-‐specific

Qualitative

Non-‐transactional

Snapshot in time

Bricks & Mortar

Unknown People

Unknown Person

Primary

“Whole World”

Quantitative

Transactional

Trended

“Digital/Online”

Known People

Known Person

|

|

|

|

|

|

|

|

Types of Research/Data

Methods of acquiring research data

1.  By surveying people

2.  By observing them

Research that yields data on audiences to solve below.


Big data, little data –generally pretty similar data. Just scale and use differ.

Aware & Will Buy.

Aware & Will Not.

Unaware & Just Might!

Unaware & Just Fine.

This is the gold mine of readers. It is the crossover hit. Especially true for niche and vertical publishers.

A must.

Content created/consumed by consumers.


Mary Meeker referred to the “data-‐creating consumer” as a top 2014 trend.

Major social platforms total registered users.


0

200

400

600

800

1,000

1,200

1,400

2004 2005 2006 2007 2008 2009 2010 2011 2012 2013

Millions

Facebook Twittter Google+ (Gmail) Pinterest Instagram

Registered users as of May 2013. Reported.

Several, culled by Search Engine Journal

US social network penetration by age + mobile.


As of May 2013. Via survey.

Pew Research: Social Media Update 2013 via Search Engine Journal


Canada-‐specific data.

Search Market Share June 2014 opt-‐in panel.

June 2014.

Top Social Media Sites Used in Last Month Canada “Digital” Snapshot Data

Source: Experian Hitwise Canada

§  86% internet penetration §  76% mobile internet penetration

§  56% smartphone penetration §  77% of owners research products on

phone, 27% buy on phone §  82% Social Media penetration

§  55% Facebook penetration §  <2 hours/day social media use

0% 10% 20% 30% 40% 50% 60%

Pinterest

LinkedIn

Google+

Twitter

Facebook

Canada and the U.S.


Sources: PWC Global Media Outlook, Census Data, Global Web Index Wave

60

7

0 20 40 60 80

U.S.

Canada

137

17

0 50 100 150

U.S.

Canada

254

30

0 100 200 300

U.S.

Canada

315

35

0 100 200 300 400

U.S.

Canada

Population (M) Ratio: 1:9 Internet Users (M) Ratio: 1:8.5

Facebook Users: Last Month (M) Ratio: 1:8

Twitter Users: Last Month (M) Ratio: 1:8.5

Trade Book Sale Ratios

Range from 1:15 to 1:10…

No “apples-‐to-‐apples”

data but directionally

these provide a sense.

A sense of proportion.

Canadian book consumers and retail.


2012−2013. Primarily via survey. (I’ve focused on the Business category.)

•  68% Business book buyers = male ! > 50% awareness = online ! Only 20% purchase impulsively.

BookNet Canada, “The Canadian Book Consumer 2013”


Some really useful places to gather consumer data.

§  Social Graph They know consumers. Online and offline. 360-‐degree view.

§  Ad Platform Open (APIs, Tools), app development, Oauth site sign on.

§  Constant A/B testing Fail fast, fix.

§  Result: Happy Users/Advertisers

Despite incredible concerns over privacy. Relevance trumps it.

§  Search (& lots else) Massive share. YouTube.

§  Ad Platform

Targeted inventory at an all time high.

§  Literally Building a Brain Yes. All products data-‐driven. Predictive.

. §  Open

APIs and tools. Oauth site sign on.

§  Massive growth Wild adoption and usage.

§  Ad Platform

Targeting.

§  Timely Almost “now.” Predictive.

§  Open (for now)

Can get at the data. Oauth site sign on.

A sampling of useful tools.


Social Analytics §  Simply Measured §  SproutSocial §  Social Bakers §  Followerwonk §  Commmun.it §  Bit.ly §  Topsy §  Social Mention

§  Facebook Ad Interface §  Facebook PowerEditor §  EdgeRank Checker §  SimplyMeasured §  Twitter Ad Interface §  Radian 6/Crimson

Hexagon §  HootSuite

§  Facebook Insights §  LinkedIn Analytics §  Instagram Analytics §  Etc.

Web/Email Analytics

Web/SEO §  Raven §  Compete §  Quantcast §  SEO Quake §  SEM Rush §  Google universal analytics

§  WordTracker §  WordStream §  Amazon comp authors §  Librarything tags/

comps §  Etc.

§  Google Analytics §  Omniture §  ExactTarget §  MailChimp

Mostly not huge, costly a la Adobe or Salesforce

§  Optimizely §  Etc.

And many, many more to fit nearly any use case

§  Google Trends §  Google AdWords §  Moz §  Soovle (autocompletes

in general) §  Seorch

I like how this guy talks about research and data.


Nate Silver. (I like others, also).

…if the quantity of information is increasing by 2.5 quintillion bytes per day, the amount of useful information almost certainly isn't. Most of it is just noise, and the noise is increasing faster than the signal. There are so many hypotheses to test, so many data sets to mine—but a relatively constant amount of objective truth.

Photo: Marius Bugge

Bayes’ Theorem

Foxes gather “big ideas”…quickly.


Photo: Marius Bugge

“The fox knows many little things, but the hedgehog knows one big thing.”

Hedgehogs are Type A personalities who believe in Big Ideas—in governing principles about the world that behave as though they were physical laws and undergird virtually every interaction in society.

Foxes, on the other hand, are scrappy creatures who believe in a plethora of little ideas and in taking a multitude of approaches toward a problem. They tend to be more tolerant of nuance, uncertainty , complexity, and dissenting opinion. If hedgehogs are hunters, always

looking out for the big kill, then foxes are gatherers.

One second on Bayesian statistics.


No test (I wouldn’t pass). The governing principle is the thing.

»  Bayesian statistics is a subset of the field of statistics in which the evidence about the true state of the world is expressed in terms of degrees of belief or, more specifically, Bayesian probabilities.

»  Bayesian statistics (if only practiced in spirit) sets one up to:

§  Statistical inferences

§  Statistical modeling

§  Design of experiments

§  Statistical graphics

§  Be human (encouraged)

§  Move quickly, get lots of data

§  Admit bias but try to verify

§  Change tack as indicated

§  Becoming “less wrong” (testing)

§  Becoming even less “less wrong,” over

time

§  Demonstrating/validating

We verify or discover the big ideas, as opposed to just having them.


Identifying and understanding audiences using data

I wonder how The Signal and the Noise is doing?


#1 Bestseller. In Statistics Textbooks….

#989 overall. Without being able to see POS, I don’t know if that signifies…

I might throw a “Business BISAC” at Amazon. It’s not a

textbook.

Nate Silver’s audience.


Wonder who they are. I have guesses but that’d be bias. Let’s look.

720k is a hefty Twitter following. He’s tweeted often and “on message.” Recency.

Where do they live?


Home locations of unnamed Silver Twitter followers based on a sample. Directional.

New York, LA, London. Is that Canada I see?

Canada?


It is indeed. But those followers are in Seattle. Drats!

Why no Canadian followers? Bug? Opportunity? (We know Canadians use Twitter.)

Google.ca auto-‐prompts me at “s.” That’s good.


1 1b

Book results are low and related.

Amazon is first book result. Way below the fold on any device.

How does the book look an Amazon.ca, Kobo, Indigo.


There a book audience but it feels small.

Two reviews feels low…

Good position. Seem like more consumer-‐

aligned categories

Would have expected him to be prompted above

Nate Southard…

What is the search interest like?


Canada – Spikes – Volume is on Him

Interest falls but stays. Book present. Google Trends Canada, US.

January 2007 – September 2012

September 2012 – May 2014

Interest falls fast. No book.

January 2007 – September 2012

US – Very Similar

Comparing raw search volume.


Canada Brand Search Volumes

US Brand Search Volume

1,400 reach in Facebook CA advertising vs. 62,000 in US

Ratios feel as if he is punching below weight.

More data on interest in Canada allows inference…


Silver does not enjoy the interest here that he does in the states.

3% is too small number, given expected ratios. Canada has about the population of California.

Hypothesis: he is under-‐indexing in CA.

Perhaps there is room for sales growth – in and

using social.


Efficiently growing audiences using data

Mine adjacencies.


Some potential adjacencies for Nate Silver.

One adjacent audience: Moneyball.


Google Adwords and Facebook confirm connection and show Canada reach.

=

=

50,000

196,000

Ride big waves.


Google Trends Canada.

“538” is Silver’s recently re-‐launched site, covering things from sports to politics.

There is Canadian search interest in 538.

He is predicting the World Cup winner in real time.

15M Tweets on World Cup in past month.

The World Cup is big in Canada (I did verify). Though it is an adjacency that is further away, Silver has tied himself to the World Cup explicitly. Hypothesis: It can likely be capitalized on to get people interested in him.

Reaching “look-‐alikes”


Some characteristics of his audience.

Regionality gleaned from search.


Are there attributes of the US locales that “match” Canadian locales? (DMAs)

Comp authors: adjacent fans and look-‐alikes.


Authors whom the consumer comps, as opposed to us. Preferably outside book spaces.

The intersecting folks are a great source of look-‐alike attributes.

Comp authors: adjacent fans and look-‐alikes.


We can use the Venn to find people to target who look exactly like the shared followers.

Thinking in terms of optimizing “funnels.”


Goal: sell The Signal and the Noise in Canada. One potential funnel (to test).

Segment

§  Male §  Like Moneyball §  And topics

related directly to Moneyball

Platform §  Facebook §  Mobile stream

Landing §  Kobo page

Creative §  A: Sports §  B: Business

This is funnel A. There should at minimum be a B, testing with at least one variable changed.

Measure costs to reach fans and conversion to sale (the goal here). See who is responding, adjust (more hypotheses) or “get out.”

This may not be a “big idea.”


But if it were to be successful it would be a nice one-‐off and could lead to learning how to develop a process of outsizing “American” authors in Canada.

»  One could systematically identify US authors with works on sale in CA §  Look for the delta in unit sales between US and CA. IF greater than norm, examine.

»  Do the same with authors with major digital presences in US without in CA. §  See what can be modeled in CA from the US presence

And so on…


Suggestions if you’d like them (along with 2 warnings)

Suggestions


»  Establish goals regarding audience identification.

§  What outcome would be ideal.

»  Involve organization around the approach.

§  Marketing, sales, publicity, IT need to align to gain maximum value.

§  Affects everything; physical distribution, ad creative, PR to metadata, etc.

»  Recognize that it is a process of testing and learning.

§  Failure (of a reasonable hypothesis) is not a bad thing.

»  Buy, build, find, learn the systems to support the work.

§  Capture learning at all times.

§  Scale when the value is there (eg. Big Ideas are coming and are repeatable).

May prove useful if data-‐driven, audience-‐centric marketing is of interest.

See warnings.

Two warnings

1.  This is relatively technical work but does not require one to be a “data scientist.” Just unafraid of technology, curious, and able to employ the logic.

2.  The more one does it, the faster it goes. It is not fast at first but is, in the end, likely more efficient and will yield big ideas.


—Nate Silver, The Signal and the Noise

Thank you


Big Ideas from Big (or Small) Data

Data & Analytics

oauth site sign

yield big ideas

world cup

aware amp

unaware amp

adjacent fans

bayesian statistics

comp authors