Eyes on Analytics

Apple, Animals, Analytics

2012-01-30T07:00:00.000-05:00

You may have read a lot about Foxconn last week.

Tl;DR summary:

Foxconn is the subcontractor that makes the iPhone and iPad.

Foxconn's CEO called his workers animals.

Foxconn, is probably racking up big time ILO violations.

Here's the key tl;dr quote, from a current Apple executive:

“We don’t have an obligation to solve America’s problems. Our only obligation is making the best product possible.”

That's a lot of focus. That's laser-like precision on a given mission. Because it follows that if Apple produces the best product possible, people won't care about anything else.

Indeed, isn't s/he right?

Doesn't free market and price competition makes hypocrites of us all? (Two-Buck-Chuck anybody?)

What if it didn't have to?

Implications for Analytics Practitioners:

NRCan runs the Energuide program which assigns a single, real number to most major appliances expressing how much electricity it uses over the course of a year, and in so doing, it makes the previously invisible, visible, to the consumer.

Is there some mechanism by which we could assign the unseen human cost of a given product and make it visible to the consumer?

How could the invisible be made visible, so that at the very least, our collective market can generate preferences that are not based on market price alone?

I make no normative judgements about industrialization in other countries, nor, a statement about the working conditions that people escape when they transition from subsistence agriculture into industrialization. I'm only asking if an analytical solution can drive an incremental disruptive dimension of consumer preference. ILO standards already exist, and have existed for decades. It's a problem-solution space that I know analytics practitioners have unique insight on.

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

Sent from my iPad

It's possible to dicern exactly what film someone is watching by analysing the power consumption of their TV

2012-01-27T07:00:00.000-05:00

You may be familiar with smart grids, open data, and Pachube.

I was reading a piece on smart data, when suddenly a wild quote appears:

“...a group of hackers who demonstrated in early 2012 that it is possible to discern exactly what film someone is watching by analysing the power consumption of their TV via their smart meter, as every film has a unique ‘fingerprint’ of electricity usage.”

Oh yes. Confirmed. It happened at a hacking for privacy event.

Reactions and Questions:

What an unintended consequence of the technology!

What other hidden signals might there be in other sources of data?

What good might come of re-purposing seemingly noisy/garbage data?

Just as William H. Perkin discovered purple dye in waste coal tar, who else might discover useful relationships in wasted data?

Another reason why it pays to be curious and creative.

(Your cable company knows more about what you watch - so the specific privacy risk - while interesting, is really quite theoretical in nature.)

**
I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

The Productivity Trilemma, Average is Over, and Analytics

2012-01-26T06:30:00.000-05:00

Thomas L. Friedman wrote a fairly good piece for the New York Times. The theme is linked to something that has kept public policy makers awake for a very long time - the Productivity Trilemma. These two themes explain part of the reason for the rise of Data Science and how Web Analytics must evolve.

To summarize Friedman:

The era of average people relying on doing an average job for average pay is over.

Technology is more efficient than ever at destroying average jobs.

Everybody has to get smarter.

To summarize the Productivity Trilemma:

Productivity growth causes growth in GDP, producing negative employment effects.

Real interest rates outpace real growth rate of GDP, causing regressive redistribution effects, leading to the impoverishment of debtors and the enrichment of creditors.

Governments attempt to keep employment high through deficit spending to compensate for employment effects, enriching creditors and ultimately impoverishing all debtors.

Policy theorists have known of this problem since the late 1990's (cited). I recall a paper from Europe dating to the 1970's though. We know this problem exists.

The implication for web analytics and data science:

Automation technology is destroying manual data entry and dashboarding positions.

Get smarter, get creative, and get into experimentation-as-a-value-add. Do it now.

Data Science is a creative outgrowth of BI and Web Analytics, maneuvering directly to be the destroyer of not just manual entry, but any thinking at all.

I can't solve the Productivity Trilemma. That's something for all of society to decide. In the meantime, the best defense against these forces is a very aggressive offense.

SOPA protest was a 'watershed event'

2012-01-25T07:30:00.000-05:00

According to Chris Dodd, the response against SOPA was unlike anything he's seen in his thirty years in politics. He called it a 'watershed event'.

Possibly.

Proponents of SOPA argue:

Americans are losing their jobs to foreign pirates.

National security!!!

It's Caucasian looting - all you kids just want free crap.

Opponents of SOPA argue:

Externalities.

Proponents want to believe that somehow Google made me oppose SOPA. What should be of even more concern to Chris Dodd was that Google had very little incremental effect. Their contribution to the movement was weak compared to what the real grassroots did.

There was no astroturf:

I learned of SOPA from one of the image boards.

It led to a slow moving reddit bloom a few days later.

We all gathered our pitch forks and went after go daddy.

Another reddit bloom encouraging blackouts ensued.

The cost of incremental votes lost outweighed the benefits of incremental lobby dollars.

The pattern was obvious to anybody watching the situation unfold.

The real test of whether this is a watershed event is vigilance. Chris Dodd will try again. Hopefully they'll have a plan for absorbing their own costs for their own enforcement, but with fewer externalities and 99% less rent seeking. They probably won't.

Piracy has to be addressed. The entire business model needs revamping. There's no doubt. It's just how it gets addressed, with as little damage to other sectors, is the real question.

I hope that this awoken GenY. The pattern of ignorance, I hope, has been broken now that something real and concrete in their lives has been attacked.

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

Changes to Google's Privacy Policy

2012-01-24T18:29:00.002-05:00

Google has announced changes to its privacy policy.

TL;DR:

They're rewriting it into human language.

Nothing about your Google Analytics account data changes.

"The only change for Google Analytics users under the new privacy policy is that now, information about how you interact with the Google Analytics interface may be shared with our other products."

Implications:

If the product is free, you are the product.

Software-As-A-Service (SaaS) analytics on SaaS users (meta-meta) is a major input in the product development lifecycle, so you can expect Google products to get better.

This paves the way towards a single Google Center of Excellence for internal SaaS Analytics.

Predictions:

Initial uproar.

Diminishing interest.

Business-as-usual.

Carry on.

Differentiation through negativity

2012-01-24T07:30:00.000-05:00

There's a big difference between skepticism and blind negativity. It's through negativity that many experts attempt to differentiate themselves from a herd. Expertise is often some sort of competition - a game by which some people are more expert than others. Over time, that negativity can accumulate in a community, causing stasis and then retreat.

Skepticism:

The sample size involved seems awfully low. We need more evidence that this relationship holds up before declaring that this is a natural law of marketing.

The author didn't consider a few factors from prior work in this field - probably a genuine oversight on their part - so I'd like to see this report replicated with those factors to see the link.

If I accept the authors assumptions, then yes, the conclusions are logically sound. However, I have a problem with the realism of one particular assumption. As a result, the model might not be predictive of events in the following sets of circumstances.

Negativity:

The study is stupid because correlation isn't causality - you can't ever say that these factors cause this to happen. It's impossible to prove, so we shouldn't try. And you're an idiot for trying.

I disagree with one point, so the entire thing just falls apart.

I disagree, therefore, the entire study is invalid.

It's good to be skeptical.

I've watched tremendous harm done because of negativity. I've watched it over and over again for the past fifteen years. It's always the same pattern.

People frequently choose to disassociate themselves from a spinning black hole of negative energy. What happens when experts don't have an audience to shout at?

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

Effective analytics is disruptive

2012-01-23T10:00:00.000-05:00

Effective analytics is disruptive because being smarter causes smarter actions.

Organizations do not, and probably can not, change as rapidly as the intelligence suggests. This alone can be a massive source of frustration, both for the analytics professionals, and for other areas of management within the organization.

Three key questions to consider:

Small failure is likely and common within most organizations - are you comfortable with those getting surfaced?

Small success is likely and common within most organizations - are you more concerned with sharing the resulting insights instead of investing in assigning credit?

Do you have a system for change management and updating strategy?

Analytics shines a harsh light on previously dark corners. And yet, knowing what you don't know, and what you would do if did know, is a healthy attitude towards the benefits of analytics.

Building an analytics team or a center of excellence

2012-01-20T07:30:00.000-05:00

No fewer than four companies in Toronto looking to build analytics departments. I'm excited for them.

A few points of advice as they move forward:

If you don't like the truth, you're not going to like analytics.

Effective analytics is disruptive and prompts change.

If you're not open to changing, then there's no point in being smarter. You're better off being dumb.

You'll hit a trough of disillusionment, usually because too many of the wrong people in the organization are looking for too many of the wrong numbers, getting too frustrated that nothing is telling them anything (that they want to hear or see), and that there isn't a transformation.

Some organizations never get out of that trough and give up. Plan for that and don't give up.

I wish them all the best and the luck as they try.

Why Marketing Science Isn't Physics

2012-01-19T07:00:00.000-05:00

Marketing Science isn't Physics.

One of the great things about the Marketing Science community is the sane approach to assumptions. Unlike economics, marketing science aims to make reliable predictions about the world, just like the other grown up sciences.

Consider:

Physics is just called physics.

Chemistry is just called chemistry.

Biology is just called biology.

None of these three fields use the description 'science' to affirm that they're science. They just are 'science'.

Next, consider:

Marketing.

Politics.

These are two sciences which are very young - especially as compared to physics. Marketing Science is really only 50 years old.

Marketing Science has special problems that are created by the subject matter itself.

Consider:

The same laws of motion that put a rocket into space in 1962 applied in 2012.

The same commercial that caused mass awareness in 1962 would not be nearly as effective in 2012.

The same commercial that caused mass awareness about Taco Bell, in America, in 1962, would not be effective in introducing Taco Bell to ~~Ecuador~~ Bolivia, in 2012.

Marketing science is particularly difficult because the phenomenon it studies doesn't stay put.

It fidgets. It evolves. It's highly susceptible to technological change.

We'd be a lot further along if we had a stable phenomenon and centuries to study it.

It just means that the types of laws and understandings we're all working to derive are that much more fluid and more dynamic. This feature of the object of study is the defining difference between physics and marketing science.

It's not any differences in the type of math or methods employed. (Marketing science is experimental and just as observational as physics.)

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

A Classification for Web Analytics Metrics

2012-01-18T07:30:00.001-05:00

I believe this classification was first enunciated by Alex Langhshur at eMetrics Toronto (2008). It's worth expanding upon.

Consider the following classification for web analytics metrics:

Pre-Click Metrics

On-Site Metrics

Post-Click Metrics

To unpack that:

Pre-Click Metrics refer to all activities that led to a visit to a digital owned property. (E.G. paid search keywords, referring domain, any traditional spend)

On-Site Metrics refer to all the activities that can be observed on the site. (E.G. Visits to specific pages, graph analysis / path analysis, time spent on site, and a host of very specific things like the nebulous world of engagement.)

Post-Click Metrics refer to all the activities that occur after the visit. (E.G. Money getting transferred to your bank account, an email address getting opted-in by the user, and even the nebulous world of recommending the service to a friend.)

The advantages of this structure include:

A clear, simple, linear chain of causality from beginning to end.

Enables the separation of the On-Site metrics as its own intermediate variable.

Enables clean segmentation among visitor attributes: returning visitor, returning customer, geography, and so on.

The disadvantages of this structure include:

Is too linear, since it does not explicitly call out the impact of repeat visitation.

Is too linear, as it does not highlight the importance of loyalty and retention.

Is too linear, as it does not highlight the interplay between bricks and clicks.

If you draw an arrow from Post-Click back to Pre-Click, you can get a nice cycle going. There's no reason not to. It just complicates the classification.

Useful?

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

How are we behaving in a multi-medium media world?

2012-01-17T07:30:00.000-05:00

I can tell you right now how I'm behaving in a multi-medium media world.

I have several tabs open as I write this:

Facebook - which contains a newsticker stream of activity.

RDIO - which is playing music, pipped right into my ear.

Blogger - where I'm typing this.

A half dozen tabs I haven't visited for a few hours, or, I have yet to actively click upon.

I also have an Adobe Air App loaded:

Tweetdeck.

That's just one device - my MacBook.

I attended an INFORMS conference in 2008 where a particularly bright professor presented findings from a new type of diary. His findings suggested that multi-medium media consumption was the norm. Ie. Many people reported listening to the radio while having the TV on while reading a magazine.

He argued for medium planning. I thought he was right and well ahead of the curve.

And this was pre-tablet.

My behavior can't be unique, but it certainly can't be common - yet. Or is it?

The challenge to this generation of marketing scientists and marketers is obvious.

Why seeing the distribution of data is important

2012-01-16T07:30:00.000-05:00

SPSS, R, and Python (matplotlib) have very functional visualization libraries because seeing the data is vital, even when armed with statistical methods.

The chart below, called Anscombe's Quartet, illustrates why:

All four data sets return the same summary statistics:

Their averages are all 9.

The correlation between x and y are all 0.816.

They can be described by the best fit linear regression equation y = 3 + 0.5x.

It's important to visualize the data, even when relatively powerful summary statistics are available, because:

Outliers are common in most data, deserve special attention, and can cause very large skews.

You may need something a bit heavier than linear regression to predict the relationship between x and y.

Summary statistics sacrifice specificity for simplicity, and as such, are not substitutes for understanding.

Conversion is an anomaly

2012-01-13T08:00:00.000-05:00

Depending on who you believe and the context, average site eCommerce conversion rates vary between 0% and 12%. That's not very helpful. In my own experience, defining conversion as number of completed checkouts divided by total number of site visitors, that rate varies between 0.20% to 2.00%.

That fact has important implications for analysis, bias, and making causal statements about what causes conversion.

Specifically:

When doing an experiment, the lower the conversion rate, the greater the number of visitors that are required to make a truthful causal statement that something causes conversion.

As a consequence, poorly converting sites that could benefit from experimentation the most are the most disadvantaged.

Methods that are more common in the machine learning community may actually be more appropriate than what we'd call 'traditional statistical analysis'.

As A Result:

If the traffic to a given site is low, it is even more important to test big things that matter, than it is to fiddle with something likely to be trivial. Take big risks.

It is preferable to increase the efficiency of the site by converting visitors into customers than it is to incur high incremental costs from driving more unqualified traffic.

We may have more success if we treat conversion as an anomaly detection problem as opposed to a regression problem.

Data Science is not defined by the tools it uses

2012-01-12T07:15:00.000-05:00

Yesterday, I wrote:

"Many [Data Scientists] will find some of their peers co-opted by tools, as it's far easier to be religious about the merits of a tool over another one than it is to exert any sort of real leadership or independent thought."

To expand on that point:

Data science is results oriented - the tool is the means to the end - it isn't the end itself.

Arguing the merits of Cognos against SAS is akin to the chefs spending an entire episode of Bravo's Top Chef arguing whether a boning knife or a birds beak knife should be used to cut a duck. (It doesn't make for good TV and it doesn't matter.)

The central tenet of good web design, and by extension product development, is 'don't make me think'. Great tools don't make you think. As a result, it's far easier to be a mindless proponent of something and to look no further.

I don't view the dumbing of Data Science as inevitable or even a good thing.

Do you?

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

Baby Faced Stratans and Graybeard BI's

2012-01-11T07:45:00.000-05:00

Steve Miller wrote of Data Science Maturity yesterday. It's a very good post.

To summarize:

He attended both Strata, a Data Science (DS) event, and Enzee, a Business Intelligence (BI) event, and noted just how young all the DS kids are, and how old all the BI adults are.

The DS kids come out of university armed with open source tools, the BI graybeards are all settled on enterprise tools.

He predicts that DS will merge with BI, largely as BI analytic data structures are unified under the BI banner and come to dominate organizations.

Editorial:

BI defines itself by tools whereas DS defines itself by methods and ends.

Many DS'ers will find some of their peers co-opted by tools, as it's far easier to be religious about the merits of a tool over another one than it is to exert any sort of real leadership or independent thought.

DS stands a chance of not becoming co-opted by tools because they're aware of what happened to BI and Web Analytics in previous generations - the real test will come.

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

The Three Reasons People Ask For Data, and The Three Questions Analysts Should Ask

2012-01-10T07:30:00.000-05:00

There are three broad categories of reasons why people ask for figures.

They know what they know and need evidence to support what they know. (Convenient Reasoning)

They know what they don't know, and genuinely need objective evidence in support or against somebody or something. (Decision Support)

They don't know what they don't know, and are looking for somebody to tell them what they should know, or what they should do. (Exploration)

There are three questions an analyst should ask whenever they get an incomplete request for data:

What problem are you trying to solve?

Who are you trying to convince?

What are you going to do differently if you had the evidence?

Editorial:

Not everybody who engages in convenient reasoning is evil - they're just building a business case - and conflicting evidence frequently isn't welcomed or viewed as helpful.

Not everybody who engages in decision support is indecisive or engaging in analysis paralysis - they have a hypothesis about the world, they have an inkling that something is likely to be true, but they're keeping an open mind.

Not everybody who engages in exploration is a pain in the ass. They're trying to understand the world better and might not even have a firm idea of what the core problem they're trying to solve yet.

I've done my best here to be mutually exclusive and comprehensively exhaustive.

What do you think?

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

Plants That Tweet When They're Thirsty

2012-01-09T15:52:00.000-05:00

A post made the rounds on HN over the weekend about a DIY kit that enables plants to tweet when they're thirsty. You can read the post for yourself here.

Summary:

Order a kit for $99.95, do some sodering, and you'll get a sensor in the shape of a leaf.

The sensor detects the relative dryness of the soil.

The sensor will send a signal that ultimately causes the plant to tweet you when it's thirsty.

Editorial:

It's a novel application of sensor technology + internet of things (IoT) + a social broadcasting utility.

It solves a practical problem that typically is solved by outloook reminders or having a Twilio App text you every two weeks.

It's another case study on a very long path to mass consumer relevance, so, while the snickers are certainly appreciated and eyes are certainly rolling - at least these guys are doing something pretty fun...plant analytics?

Can you think of other uses for a sensor in your home?

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

Social Media Management Systems and Strategy

2012-01-06T09:11:00.000-05:00

Yesterday, Jeremiah Owyang, with Andrew Jones and Christine Tran - from the Altimeter Group, published "A Strategy for Managing Social Media Proliferation”.

It's so huge and so packed with goodness about Social Media Management Systems (SMMS) that I couldn't do it justice in three bullet points. I extracted three choice quotes below.

Choice quotes:

"Corporations should not jump into social media without a clear goal...Instead align goals towards business objectives." (p. 23)

"Unlike one-way marketing of yester-year, company must be ready to engage with negative conversations, often taking them head-on." (p. 23)

"New media efforts will be scrutinized by management as budgets shift. Be prepared to measure." (p. 23)

Editorial:

They're correct in emphasizing the importance of strategy and process beyond the tools.

Twenty-seven SMMS providers were evaluated (!!!). Why so many? Like they say, low barriers to entry - after all, API's are so easy, right? (They're not. Easy to get in. Hard to do well.)

Social is a very different beast. The TL;DR is page 23. Then flip to page 14 to see an excellent radar of the space.

Blatant Call Out (I head up research at Syncapse):

In that ranking of 27 SMMS, Syncapse scored 'strong capabilities' in 4 out of the 5 use cases, and 'above average' in the other one. Syncapse was the only partner to score so well.

The report is a must read if you have any interest in this space, starting a strategy, or preparing a RFP.

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

Strategic Analysts, Technical Analysts, Analytical Analysts - A response to Stephane Hamel

2012-01-05T19:00:00.000-05:00

Stephane Hamel posted an excellent article today on "The Three Heads of Online Analytics".

To summarize:

To succeed in online analytics, we need an analytics competency center.

Start with the business in mind, get the technology in place, and then analyze to generate insights.

Analyze your weaknesses and zap them, try adapting 'paired programming' that's common in agile methods.

I commented:

There are T's, S's, and A's. T's are technical analysts. S's are strategic analysts. A's are Analytics (actual analytics, not fake BI reporting) analytics.

Hiring a competent T-A is exceptional. Finding a S-A is getting easier. Finding a good T-S-A is damn near impossible. Most of them are consultants.

Web analysts need to identify their weaknesses, be it T, S, or A - and move to vigorously eliminate them.

I thank Stephane for bringing it up and contributing.

Audrey Watters on Data Science in 2011

2012-01-04T19:00:00.000-05:00

Audrey Watters wrote a good article about Data Science in 2011 for O'Reilly Radar.

Audrey cites three big events / trends:

Hadoop, an open source distributed computing framework, became ubiquitous.

More Data, More Privacy Problems - citing the Apple scandal as an example.

Open Data at an Inflection Point. (With a great shoutout to my friends at BuzzData!)

Editorial:

Even Hadoop has warts (gasp!?!111shiftoneone), but so far there are good anecdotes about it working out well for many companies, and, you can expect a few horror stories in 2012.

The devices we're carrying and our own behaviors are generating more data than ever - and people - that means you - need to be aware of when they're consenting to sharing that data.

Open Data represents an awesome opportunity that I'm very optimistic about, and so much work to make data effectively accessible needs to be undertaken.

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

Steve Miller: Data Science Skepticism

2012-01-03T19:00:00.001-05:00

Steve Miller authored a very good article about Data Science Skepticism over at Information Management.

I've previously written about Data Science and shared an excellent video about what makes a great data scientist. Both posts are expanded primers on the emerging field.

The TL;DR version is:

A Data Scientist (DS) sits at the intersection of computer science, statistical methods, and business.

I won't define what Business Intelligence (BI) is.

There's an EMC study making the rounds. Steve Miller takes exception to some portions of that study.

To summarize Steve Miller:

Findings from the EMC survey made certain statements about BI's that are unnecessarily polarizing, and should be viewed with suspicion by data scientists (which should be their natural inclination anyway).

DS distinguishes itself by applying the scientific method to data, discovering insight and productizing that output for both customers and businesses.

BI also uses the scientific method, is focused on internal operations, and there's nothing wrong that at all - just recognize the continuum.

One of the most useful quotes in the article is:

"Others contend that data science distinguishes from traditional BI in its strict adherence to the scientific method that:

Observes a phenomenon or group of phenomena;
Formulates a hypothesis to explain the phenomena. The hypothesis takes the form of a causal relationship or a mathematical function (the more of X, the less of Y);
Uses the hypothesis to predict the results of new observations; and
Performs tests (experiments) with the predictions."

Editorial:

Marketing science and data science are not divisions of business intelligence, although, it is easier to sell that way. (And whatever - sellers are gonna sell.)

Business Intelligence is a division of Operations Research.

As a data scientist and a marketing scientist, I turn data into products, actionable insights and stories, and others are entitled to define themselves any way they want to.

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

Tyler Nichols: "I am done with the freemium model"

2012-01-02T11:13:00.002-05:00

Tyler Nichols writes: "I am done with the freemium model".

Tyler divided all the users of his service into two groups: free and paid.

He measured the behaviors of each group.

He found that the free group was detrimental to his business because:

They emailed more questions on average than paid people.

They hit the spam button when he emailed them with a follow-up, paid people didn't.

Free customers were not worth the maintenance costs they caused.

Hacker News and other communities replied (paraphrased):

Free people were not as engaged, and therefore more wreckless.

It was a santa letter generator, which has low repeat value after the season.

The plural of anecdote isn't evidence, you've added little value to freemium business model discussion.

Editorial:

Freemium apps can be monetized by way of ads.

The conversion rate from 'free' to 'paid' is generally very low, so the decision to go free must be placed into a broader strategic-competitor context.

Maintenance on free customers can be managed by way of automated autorespond emails pointing people to the FAQ, at the opportunity cost of conversion to a paid fee.

A shift in voice

2011-12-31T18:00:00.000-05:00

I used this blog to talk to very specific groups. Sometimes it's marketers. Sometimes web analysts. Sometimes it was candidates applying for a position. Sometimes it's data scientists, brandsters, and social analysts.

Sometimes this worked.

Sometimes I confused the hell out of different audiences at different times.

I'll continue to speak to web analysts through the research committee of the Web Analytics Association, in particular, through a new experiment we're launching and ongoing Peer Review Journals.

I'll continue to speak and collaborate with ultra niche communities - data scientists, marketing scientists, and open data professionals through christopherberry.ca.

Eyes on Analytics is shifting.

I'll be curating content from not just from marketing analytics, but also from further afield. My goal is post more interesting content, more often, with less reading. Finally, I'll be borrowing more heavily from the Economist Style guide, with less jargon.

Thanks for reading.

Hype Cycle Cleverness

2011-12-26T13:06:00.000-05:00

It's worth explaining The Gartner Hype Cycle. It's topical for 2012.

It works as follows:

Usually many people invent a technology during the same envelope of time. Somebody really gets hooked on the idea. That somebody executes the technology sufficiently well that it produces a technological trigger. And that gets the ball rolling. Awareness spreads through a single market, and then transmits into adjacent markets. Excitement spreads like fire.

People are quick to see potential. Enthusiasm is contagious, and opposing views are downvoted into gray obscurity. Innovators are visionaries. After all, I'm winking, pointing a finger at you, and making a 'click click' sound my voice. 'Hay, click click'. This is an impolite way of saying that 'ignorance increases'.

Hype builds to a saturation point. We're at the peak of inflated expectations. Calls that 'this is a bubble' aren't quite greeted with the same amount of derision. Pesky questions about monetization aren't met with statements of 'this time it's different' remarks. The enthusiastic roarers aren't as loud. In fact, they're becoming quieter. They're not popping by as much. They're not delivering sermons at conferences as often.

The dreamers and the hollow suits evaporate rapidly. They're off to the next big thing or something.

We ultimately end up in the trough of disillusionment. And it sucks. Everything is negative. And there's a very real danger that an entire technology may not make it out alive. Impatient firms abandon their efforts way too soon. The firms that stick by their initial strategic logic learn a lot more. Most technologies climb up through a slope of enlightenment.

Ignorance receeds.

People who were generally not associated with the initial hype find new uses for it. And, gradually, the technology enters into a plateau of productivity. People are generally aware of it, but the sizzle is all gone.

The Gartner Hype Cycle has its critics. It's not a cycle. It's not prescriptive. It's not really objectively measurable.

Yet, just because something isn't perfect doesn't mean that it's useless. I find it pretty useful. It's a model I use to understand a complicated part of the world.

I like the because I can mentally mash it up with Moore's 'Crossing The Chasm', which is a seminal and well studied concept. The Early-Majority is skeptical.

I really enjoy skeptics who get way ahead of the peak. It's as though they're racing ahead to the trough of disillusionment just so they can say 'ha, I told you so'. Is there a market for that? Are there people who really want to buy a piece of consulting that way? Or, more specifically, do such people reinforce the bias of the Early and Late Majority?

So yes, hype is hype. And if you're annoyed by hype, technology probably isn't the right sector for you. Maybe try steel. The pace of innovation in the steel industry is constrained by depreciation. It takes a generation for something to come in and go out.

In technology, it can take as little as 2 years for something to go from dark to hot and through to dark again.

I try to look past the peak and look at the plateau. Monetization doesn't have to be obvious from day zero, but there should at least be at least two business model canvasses that appear viable. The trough always comes. How should that be managed? And how far away is it?

That, in sum, is the Hype Cycle. What do you think of it? Is it accurate? Is it not?

Why Things Don't Mean What People Assume They Mean in Analytics

2011-12-08T14:46:00.001-05:00

I live and work at one of the most amazing intersections. It's also the cause of why things don't mean what people assume that they mean.

There are technologists - developers and computer scientists - who grapple with the limitations imposed by API's and big data.

There are marketing scientists - analysts and statisticians - who grapple with the limitations imposed by computability and understandability.

There are marketers - brand and channel - who grapple with the limitations imposed by budget and cognitive surplus.

It's pretty amazing how a technologist, a marketing scientist, and a marketer can all be right within their own silo, their own way of thinking, but collectively be misunderstood and wrong. The confluence of all three types of people generates continuous tensions, innovations, stresses and misconceptions.

Just to share one such example, take the concept of a response rate, a conversion rate, and an engagement rate. We've had peace, among web analysts, since 2009 on that topic.

And yet, technologists face the limitation that we cannot access certain measures at a unique individual level. Marketing scientists face that limitation, and have it compounded by natural activity inflation within the numerator. That is to say, new types of engagement are invented every quarter. Engagement rates drag higher not necessarily because the experience is better or the marketing more effective, but because more opportunities exist. Marketers face the limitation of having to understand, optimize, and ultimately prove that they matter. Which is also a potential cause of numerator inflation.

Systems don't exist because measurement exists. Rather, systems evolve to maximize some other result. It will always be that way. As a result, the limitations enumerated above will always exist. It's best to understand it and try to make use of it. This causes choices to be made. And like it or not, choices are made.

For instance, the marketing scientist will attempt something new, like some variation on the new LIV algorithm, to control for numerator inflation. In so doing, they'll make the calculation impossible for a marketer to understand conceptually, or to reproduce at all, and even harder for the technologist to implement. They might try for something far simpler, but in so doing, introduce even greater problems into the equation.

This happens again and again and again. It happened in 1993.

The use of the word 'people' in web analytics caused one such problem. Technologists at Netscape invented the HTTP cookie, within the limitations imposed at the time, to understand revisitation rates to the Netscape.com website. Web analysts always meant 'unique browser cookies' when grappling with that innovation 2 years later in 1995. It was explained to marketers (and maybe marketers explained it to themselves) as 'people'. The 'visit' and 'unique visitor' were massive advances over the term 'hit' (still in use by some very, very old people even to this day!), but didn't get down to that unique 1 cookie : 1 human being relationship.

The superiority of the cookie, and the linkage to direct transactional and research behavior, is a huge step forward for marketers. It's still very cute when a TV executive asks "why can't digital people get their measurement together?". Indeed, the TV executive is fully aware of the problems with ratings. They just gave up in 1954, because, well, TV doesn't exist solely to measure it.

Things don't mean what people assume they mean, at least in analytics, is because of the limitations imposed by reality are passed onto people who try to interpret them, and, the lack of time and brain memory space to remember that.

It's only a problem if somebody makes a bad decision based on an incorrect interpretation of what something means.

Hope and Action for 2012

2011-12-01T09:21:00.000-05:00

It's an annual tradition. We navel gaze. Every December.

It's as predictable as the tides.

So, let's talk 2012.

A Forrester author, Joe Stanhope, asked us what we wanted to be when we grew up. I replied, 'doing something meaningful', or something to that effect. I meant it.

Let's consider something really meaningful that's happening right now.

Joe painted a picture of accelerating medium fragmentation and bloatware trying to keep up. Indeed, I'm encountering more analysts gathering the pitchforks against the new-new media.

After all, if we can't even do x right, what business do they have to even attempt y?

Because it's there.

It's never been a better time to be a marketing scientist or an analyst. It's never been a better time to experiment. To run tests. To run them lean. And to peer into some of the most violent datasets and questions. It ought to be refreshing.

Convenient reasoning, focused against building business cases for the new-new media, will get analysts nowhere. If the old reporting has to be cut back so the new-new media can be explored, then so be it. Apply scientific principles to your own work and optimize. Razzle dazzle'em. *click* *click*.

Adapt, in 2012, or be consigned to the dustbin. Lack of a business case hasn't prevented any marketer from ever experimenting. And, with Europe on the brink and strife at home and abroad, never before has bold ideas and bold experiments been so necessary. Shrinking marketing budgets mandate broader experimentation.

How are we going to get out of this mess? By spending more on radio? Really? Is radio really the answer? (I'm poking fun at the one radio analyst I know. He agrees with me.)

So, don't sound like the radio analyst in 1992 with respect to the Internet. The economy sucked then too. And we came out booming. Sound progressive in 2012 with respect to whatever it is that's emerging, and win the decade.

Resist and lose the decade.

That's pretty much the choice.

Look at that fragmentation, would you? We got social. We got mobile. We got social apps. We got angry birds and in-game commerce. We're growing enough crops on Farmville to feed the whole of North Korea, nine times over. We got virtual reality and virtual payments. We got apps on DVR's. We got portable tablets with angry birds. We got netflix on your XBOX. Did I mention XBOX? We got XBOX! We got social apps on mobile. We got timeline coming up (you better get on that). We have an emerging Internet of Things (IoT) that promises to illuminate us all. The trend is more data in more places.

And consider the changes in usability. We're inventing new concepts quickly. The pinch. The three point touch. The swipe. Even the notion of the Newsfeed is relatively recent. What happens when we can engage with objects virtually? What happens when things become usable in a digital sense?

Who's going to grapple with it?

The BI guys? Really? They're still working on integrating data from my Intellivision into their datamart. And they're trying. They're really, really trying. But no. You can't rely on them for help.

New analysts? Maybe. The great fifth wave of analytical talent - those that got their start in 2008 - are so rare. The great recession has done more to wipe out the present crop of analysts who should be hitting their 4th year stride than anything else could. Directors and VP's of analytics will pay dearly in 2012 for the devastation that 2008 caused. And let's face it, there's a solid 9 month ramp, even under the best of circumstances, to get a new analyst walking on both feet. On the positive side, if you instill a positive attitude about medium fragmentation and experimentation, you have the chance to set them up for success in a big way. But no. You can't rely on them to save you.

It's you. You're going to have to grapple with it. You're the one who knows what a cookie used to be. And what a gesture is. And you're probably one of the few people that understands how all these other channels fit together - like the email program with the SEM program with the website program with the old mobile program. And each channel has a strategist. And they're all fighting for fewer dollars and less attention.

But not you. Who are you fighting with? With the marketer that wants to try a social app? Why? Is that really worth your energy?

My hope for 2012 is for Europe to pull through it. I'd like to avoid a second major recession and the nervous stuff that goes with it.

My hope is also for action.

My hope is that we'll go from viewing data as a tool for explaining the past and destroying business cases to using data to actively make change better, and make better changes.

We can start really mattering.

I'm going to stand off my soap box now, and ask you - what do you think? Am I pretty preachy about fragmentation? Am I right about the dire consequences of getting in the way? LMK what you think.

.

Thoughts on "The Closed, Unfriendly World of Wikipedia"

2011-11-23T21:36:00.001-05:00

Danny Sullivan wrote a pretty good blog post about an article getting deleted. You can read it here. I'm not so interested or outraged about it.

This spawned a Hacker News thread. You can read the whole thing here.

The comment I want to draw attention to comes to us from Phil Welch. It's so good that I'm quoting it below.

"Turns out if you throw together a few thousand neckbeards and convince them to play status games around building an encyclopedia, you get an encyclopedia.

You also get a whole lot of stupid politics, wasted energy, process wanking, flamewars, and acronym-laden cryptic discourses where words like "arbitration" have strange, Orwellian connotations. ("Arbitration" is Wikipedia's name for the process governing, among other things, removing administrator privileges and banning contributors for long periods of time.)

Contributing to Wikipedia is a usability disaster because of all the red tape, process, policy, and other crap the core group of contributors has constructed.

But the project is big and popular enough, and the work is easy enough, that people still throw themselves into it. You start by just making a few small edits to something, or adding some sources and information to an article. Then it goes on a bit, and you learn some of the terminology and process, and you start feeling personally invested in it.

I mean, it's Wikipedia!

It has lots of information about everything!

It's free!

The human race needs something like this!

Before you know it, you're in a "community" of core contributors. There are people with impressive titles like "arbitrator", "administrator", "bureaucrat". You start playing status games. You worry about your edit count. You try to get an article up to "featured article" status. You network. You try not to make enemies. You try to sound wise on "Articles for Deletion", or the "Administrator's Noticeboard", or a million different talk pages.

Eventually someone nominates you for administrator. You answer a few questions and other contributors vote for you.

Let's assume you played politics well enough leading up to this point. You win! Now in addition to all the crap you were doing before, you get to play with your administrator tools as well.

Fun, eh? Well, for some people it is. But at that level, Wikipedia is 90% politics.

There's an awesome encyclopedia there for sure, but man, you don't want to see how the sausage is made."

-------------------

Phil Welch's narrative is enlightening and refreshing.

What's really amazing is how generalizable Phil's narrative is. The same can go for a number of Toronto's community of communities.

Wikipedia has power because it's so heavily trafficked. Quantcast lists it as the seventh most visited website in the United States. The edit wars on some pages is incredible. Wikipedia is widely thought to be 'collective truth'. It isn't really. It's just a good starting point.

Many communities organize themselves around systems. The power of communities is that they can modify those systems. Political scientists recognize this as path dependence. It's real - just ask Thelen.

Path dependence is measurable and predictable.

What would you use it for? Well, that's up to you.

DJ Patil on the traits of a good data scientist

2011-11-07T16:55:00.000-05:00

How to think about the attribution problem

2011-11-04T06:30:00.000-04:00

Joe Stanhope wrote a good piece for Forrester. If you have a subscription to Forrester, read it. It summarizes the state we're in, and has a few very good points on the last page.

In that piece, web analysts themselves list 'attribution' as a major challenge.

This is a wicked problem. All the energy you put into untying that knot only causes it to become tighter. But let's try this again, together.

If you haven't seen this previous post, it's new to you. I drew out a conceptual model report, in part to demonstrate how cause-effect can be embedded into a report.

Alright - so that's a conceptual model. I believe that paid spend causes paid visits. I believe that affinity score is a predictor of returning and lost customers. And I believe that non-working dollars should be part of your profit figure. I also don't believe in returns.

A lot of the math here is pretty tame. It should be pretty obvious. And it served its purpose.

So, if you're staring at that diagram and you want to think about which levers you could pull to make that profit number increase, what would you do?

What's the relative strength of each lever?

I don't think of this problem as one of 'giving credit'. I have no dog in this fight - with the paid media person fighting for budget against an insurgent customer affinity group. I'm as objective and pragmatic as possible in setting up the model.
There are methods to execute attribution modeling - all technical. But crafting the solution means having a clear model that links cause and effect. I'd argue that we should be thinking in terms of a system - and that it's not couched in language around 'credit', but rather, how to make the entire system more effective.

Systems Thinking and Marketing

2011-11-01T06:28:00.000-04:00

The complexity in measurement ramps with the complexity of the channel. In this post, I'll write a bit about an interpretation of systems thinking, and how I apply it to marketing and marketing modeling.

We all seek to minimize complexity and maximize predictability. We want to minimize risk and maximize empowerment. We want to synthesize a huge amount of information and boil it down to a handful of levers. Levers cause empowerment and they enable people to make really good decisions.

Systems thinking is an actual thing now.

Some organizations already have models in place, and are all fairly standardized. Not every organization has them. Understanding them is pretty important.

This is my approach:

I write a load of variables out onto cards. I talk to a lot of people to get all the variables onto the table.

And then you're confronted with complexity.

I arrange all the variables in a way that cause and effect makes sense. You want to put the outcome you really want to achieve on your far right, and you want to put the variables you have the most control over on the far left. And then you start piecing things together

It's very natural to believe you have far more levers at your disposal than you really do. Areas on the left will start to migrate towards the center.

It's not entirely about your gut. Certain relationships, like the one between age and income, are very well known. Spend and reach is usually well known. Other relationships, like between recall and creativity, are known directionally but not certain.

You've now created a model.

Classically trained statisticians are taught to avoid reinforcing variables in their models, as it leads to a very particular error in regressions. I gravitate towards reinforcing variables. I can deal with the regression later on - but it's truly an artificial barrier. When you're laying things out, you want to pay special attention to such variables and dynamics. This is where you can get some of the best efficiency and/or effectiveness. A little difference there can go a long way elsewhere.

It's also where a lot of the most optimistic thinking comes in. It might be attractive to think all reinforcing variables go on reinforcing indefinitely. But in this house, we obey the laws of thermodynamics.

(Obligatory Simpsons Reference)

This is also the root of the 'virality' argument. That somehow 2 people will each tell 2 that will each tell 2, and before you know it, the entire planet knows of something. If virality really worked that way, we'd all be aware of everything, ever. But we're not.

Reinforcing effects typically have a time limit. Even halos dissipate. Even virality dissipates.

Once it's laid out, you can run a few simulations on it to get a sense of the range and impact of the variables. Then, for the purpose of communicating a model with the fewest number of words, delete the least predictive variables and levers, and communicate a simplified version of the model.

You do the usual 'what-if' scenarios and encounter assumptions about the world that don't quite make sense. You have to aggressively inquire about why people think the way they do. Assume the best in people. Be aware that sometimes people forget that targets are a logical consequence of building a business case. Not everybody thinks ahead, and, in some organizations, there is truly no linkage between targets and business cases.

Remember that the model you generate is one in trillions. Your goal is to generate the most predictively accurate model you can based on what you know. It is all but certain that more accurate models exist, we just literally don't know what we don't know at this point. It's not a question of 'wrong', but a question of 'better'.

Good enough is defined as good enough to make a confident assertion about a set of causes that have likely effects. You're not aiming for perfect because perfect isn't possible for another few hundred years (or certainly some other time scale that exceeds your life).

Once you have a model, you have a system. You can use that system to write powerful recommendations that link actions the firm can take with outcomes that are likely. A system ultimately leads to a general understanding, and we move into a process of normal science.

Evidence accumulates that certain relationships are growing weaker, for instance, between TV Spend/GRP and Sales. New channels emerge and fragment customer attention and behavior. In short, systems evolve and much of what was true in 1990 isn't true in 2011. Much of what is true in 2011 won't be true in 2020. It's not that anybody was wrong. It's only bad if we fail to update our present view of a system with new knowledge.

That's my take on systems thinking and marketing. Useful?

eMetrics NYC 2011 Recap - three major takeaways

2011-10-22T14:47:00.001-04:00

The eMetrics NYC conference / data driven business week, is said to have attracted some 1600 people. The WAA Industry meeting coincided with it. It was a success.

Three major takeaways that stick out in my mind.

Janus Faces for Janus Audiences

I've now seen two versions of Peter Fader. The academic Fader and the industry Fader. The one you see at an INFORMS Marketing Science conference is the academic Fader. He's the kind of guy that'll smile as he tells his skeptical detractors to take their perspectives and reconsider them. His insights into models and the way theories have been constructed are original. He doesn't mess around with the bits at the edges. He's good at the comparative method among models, even if that's not quite how he'd describe it.

Then there's the industry Fader. He's accessible and engaging, even for a Friday afternoon. He presented a pretty good case of the way that marketing scientists in academia work with industry practitioners, and went on to highlight a variant of the shop-till-you-die model. It was interactive and participatory - the very opposite of INFORMS MS - and it was excellent.

This is probably the most radical example of the 'know your audience and adapt' that I've ever seen. It's inspiring - just how far a model can be contextualized and explained to a very different audience - to that extent. Incredible.

(Finally, for me, it's a key datapoint for the evolving partnership between academia and industry, for mutual benefit.)

Mobile has arrived

It's notable just how little hype there was for mobile this conference. Every year is supposed to be the year that mobile breaks out. That it's going to be the year of the hockey stick. Discussions around app and tablet usage were muted, and rare.

I saw loads of tablets. Loads of smartphones in use. There was tremendous engagement on Twitter and Beluga throughout the conference.

The reason for the lack discussion, possibly, was that mobile has already broken though.

Have we started asking interesting questions yet?

We're well positioned to take advantage of the next 5 years

For most of us in the East, the October eMetrics is our major collaboration. Not everybody made it. Those that did included long time collaborators and conspirators. And it was intense.

I was very happy to share a stage with the only other data scientist there - Michael Healy - and that panel led to a very spirited discussion with Jim Novo and one of SAS's leading text miners (IE: Richard Foley). The debate changed an important perspective I've long held on data quality. That discussion couldn't have happened anywhere else.

The sophistication of discussion was high. I spoke to a number of directors about their challenges with social. They were having the same problems I was having, and I was happy to share how I worked with developers to solve them. Many of the solutions reside within Syncapse Platform, and are getting better every day. I found that others were also really forthcoming with the problems, solutions, and mistakes they had made, in a range of fields. Again, those discussions couldn't have happened anywhere else.

Based on the quality, caliber, and energy I felt at eMetrics, I'm certain that we're well positioned to take advantage of the next 5 years. We have incredible challenges, opportunities, and technologies to take advantage of.

What's next

Patrick and I thank, again, everybody who came out to the 'Communicating to designers' preso. June Li remarked that the style is completely different from talking to executives. And I'm inclined to agree. It was very unusual to have a design discussion at eMetrics, and yet, I wonder - why should that be the case?

I'm up for another six months of contributing to eMetrics Toronto. I'm particularly interested in stories from the client side that follow the dramatic structure. No, seriously. It's great storytelling.

The next WAWTO is October 26 at the Wellington. I'm looking forward to seeing many designers, web analysts, developers, hackers, IA's, data scientists and marketing scientists out for the event.

Insight Revisited

2011-10-18T18:48:00.000-04:00

I wrote a definition of 'Insight' on December 29, 2010 that read:

"An insight is:

New information
Executable
Causes action
Profitable

Or, more detailed, an insight is:

A piece of information that you didn’t know before, which -
Can feasibly executed, culturally acceptable and of a scale relevant to the firm, and -
Causes a decision to be made that wouldn’t have been made otherwise, and -
Results in profit or a sustainable competitive advantage"

If held to that standard, insights are incredibly rare.

Depending on who you talk too, an insight may mean:

A bullet point factoid

An element that combines all that is known about a segment, compressed into the foundation what motivates and inspires them

A simple statement of cause and effect

So, any of the following have been called 'insights' in the past:

Sales increased by 4%
Young men 18-25 compete to get laid and look for edges
Scoring better Google PageRanks increased organic traffic by 9%

It's all possibly all 'new to you'. The rest is all context. Bullet one isn't actionable. The second bullet point is actionable if a company can take action by solving that particular segment problem or aspiration. The third is getting there - what's the unit of measure for 'Google PageRanks' - and is it a lever I can push?

There's a lot of confusion out there in the market about just what an insight is. It's a problem.

Whenever you're asked for more insights, please ask, 'what do you mean by more insights'. It's the only way to get into their head. Yes, it's aggressive inquiry, but you won't be chasing your tail.

The Most Measured Era in History

2011-10-12T17:56:00.003-04:00

We're living in the most measured era in history.

Are you the beneficiary of any of the data you're generating?

You optimize what you measure.

One of the most data intensive self-improvement projects I undertook was in 2005. I recorded everything I ate and every exercise I did. And did I ever optimize - to the point where my joints couldn't keep up with the muscle and bone growth. It was a massive amount of work to record all that detail, the weights of various things and then to cross reference with the USDA database. Then it all had to get loaded into SPSS for analysis. It was brutally time intensive. But it did generate incredible evidence-based insights about the way my body worked.

Those insights formed heuristics. The no-fry principle. The 100g daily protein target. The outer aisles principle for super market shopping.

The friction and time intensity ultimately meant that measurement receded.

It's 2011. The recording and the cross-referencing should be made better why way of mobile apps, so I'm looking forward to less friction on that front.

The transferability of data from that device into a format that can be read by SPSS or Python is a lingering problem.

Ideally, a machine would be able to make some sense of the data on my behalf. That is another, wonderful, opportunity.

In many ways, I should be the beneficiary of the data I generate. If I want to benefit other groups, like researchers, then so much the better.

It's my hope that more developers will partner up with statisticians to produce incredible data driven systems that provide real utility to people. That people can use to make themselves better and save a lot of time. There will always be other beneficiaries of that data - after all - if the product is free, you are the product. That's not necessarily a bad thing, especially with consent. It's a good thing.

We should all benefit from the data we're generating ourselves, for ourselves.

The fight for the Data Science Soul Begins

2011-10-06T19:25:00.000-04:00

This is a pretty good summary of the definition of data science. Some statisticians seem to be incensed. Some people say that this whole thing is invented as an O'Reilly buzzword. And there's consternation, fear probably, over the devaluation of actual craft.

Sound familiar? Ah, the great Web Analytics debate of 2007. Yes. We've seen this.

Nothing like a fresh Gartner Hype Cycle in the morning, is there?

But lets consider what technology is causing, and the role that data scientist will play, in driving that cause.

Accessibility to data is expanding. What used to be the jealously guarded by people who didn't want to be educators, is now liberally spread. It doesn't really matter that most people don't know what the figures means, does it? At best, it's making big parts of the world smarter. At worst, it's merely reinforcing pre-existing ignorance about what people conveniently want to believe.

Pop-business literature is good. More people are aware of the potential. It's awesome.

And there's no resisting this market. Everybody wants data because they believe it will make them better. That it will make the smarter. That it will result in sustainable competitive advantage.

And it can. Data is preresquisite. Understanding, well, that's something else.

Data scientists will use the very best of computer science (computability, algorithms, scalability), the very best of usability (IA, UX, Infometrics) and the very best of statistical analysis (models, probability, learning). How they do so, and what are the best patterns for success, will fuel an entire generation of operations research. A sort of meta-meta study of the meta-meta of competing on analytics.

Too much meta?

Not enough. Not nearly enough.

While the fight for labels will begin, and then persist, for the better part of this decade - the proof will be in the experiences, and, one degree of causality out, the results.

***

I'm Christopher Berry.
I tweet about analytics @cjpberry
I write at christopherberry.ca

Google Analytics Premium, Enterprise, and Disruptive Innovation

2011-09-29T16:08:00.000-04:00

Google Analytics Premium was announced today. Finally. It wasn't really a secret.

What do you get for an enterprise fee?

Dedicated support, a number to call, no data caps, some attribution modeling (nice), and now, 50 custom variables.

There's good literature around disruptive innovation in web analytics, with a very specific vocabulary and model build around.

Is this disruptive or incremental?

The increase to 50 custom variables is purely an increment on an established dimension.

Enterprise support is incremental from the previous version of support. They did have a type of support. It was vague. But it was there. So that's an increment.

Increasing data caps is incremental.

The guarantees are incremental. There was protection in the past. There's more protection now.

The attribution modelling is a new. It could be disruptive.

Is attribution modelling disruptive?

It's certainly a new competitive dimension. It's likely to be far more transparent than most MMM firms. And it's a natural extension of Google's core competency. It helps people understand why things are happening. Original features that assists people in making better decisions down the line could, potentially, probably, be considered disruptive.

Watch this feature.

It's squarely in their sweet-spot.

Industry

Fan boys abound. Omniture Fan Boys will point out a superior report builder and data slicing capabilities. Coremetrics Fan Boys will point out usability differences and a superior eCommerce support. Webtrend Fan Boys (you tireless people) will argue it's the best from a compliance standpoint for government. Google Analytics fan boys will scream 'game changer!'.

Google Analytics was always a reasonable alternative to Omniture and Coremetrics. (I did my first enterprise install of GA in 2008.) There was just a lot of discomfort within enterprise on support. It persisted. It drove preference. The last of those specific enterprise concerns are now addressed.

In effect, the incremental improvements are along known primary dimensions, and allows them to compete better.

You, The Customer

There's value if you're an enterprise manager.

For 95% of the user base, the release isn't aimed at you.

When firms compete, you win.

Towards Disruption

The Adobe Vision was that designers would value analytics, and that synergies would be realized through integration.

The Google Vision is to assist people in making better websites, better experiences, and better results through data. They believe in democratic data access.

The IBM Vision is sense making.

Each vision contains a bias about which dimensions are most important, and most likely to generate a sustainable competitive advantage. The vision is a grand hypothesis.

Which do you buy into?

Credit where due

Phil Mui, WAA Research Committee member and group product manager over at Google, has an excellent track record of both incremental feature improvement and some disruptive innovation. I congratulate him and his team for getting this out and launched.

Facebook GraphRank

2011-09-25T13:10:00.000-04:00

A lot happened at the F8 developers conference last week, the most significant was changes to the Facebook GraphRank and the Social-Product Graph.

Instead of offering a single degree of freedom, to 'like' anything or remain silent, it will be possible for people to state (verb) + (noun) something. I have 'read' + 'this book'. I have 'watched' + 'dexter'. I have 'eaten' + 'breakfast'. And, I hope, I have 'bought' + 'this phone'. This goes to the notion of 'friction'.

Frictionless

The term 'frictionless' was used a lot at the conference and this has significance.

Friction is resistance to sharing information. It's caused by technology, experience interruptions, and by, yourself.

Let's start with you.

There's a pretty good model, put forward by Wendy Moe at the last INFORMS Marketing Science Conference, that people undergo two stages when interacting with social media. The first stage is considering if they're going to say anything at all. The second stage is how they'll moderate their opinion for the audience and their peers. A lot of evidence was put forward at the 2010 INFORMS MS Conference about the 'need for self-expressiveness' amongst individuals, and youth in particular. Your subjective opinion of 'over-sharing' depends a lot on your attitude towards self-expressiveness. Another way of saying that is sensitivity to privacy, or need for privacy.

Your perception of Mark Zuckerberg's language around 'expressing yourself' is likely heavily moderated by that very same attitude. Do you feel the need to post everything you do, all the time? Do you feel the need to update anything at all?

You cause friction. Technology causes it too.

The like label on the like button is such a technology. The verb 'like' forces an editorial or an endorsement, and, applying that Wendy Moe model, if somebody were to share an opinion, it's going to be moderated. You have only two options. Click like, and generate a social endorsement, or don't click like - which generates no social signal whatsoever.

Reporting facts, without the editorial, ought to increase the volume of sharing. They can certainly increase the volume of data volunteered by reducing the self-censoring, self-moderation step.

The addition of objective 'reporting the facts' verbs will contribute a large amount of data to the social graph. And that's not trivial. The volume of information about what somebody experiences will increase, and it will be organized through the graph.

Changes to the allow technology are another factor. The photo below caused Mark Zuckerberg's heart to sink. It's easy to see why. If Super Mario Bros was released during the era of Facebook, this is what you'd probably see. What a terrible experience.

The publish allow is common. It causes friction, and, it ruins many user experiences. It has its origins in privacy concerns. It's at this point that a 'utility' becomes an 'unwanted technical solution'.

Arguably, if you have an app from Netflix, or from RDIO, on Facebook, what you listen to will be added to your social-product graph seamlessly.

The Benefits To You, The User: Relevance

This information is arranged on a product-social graph. A graph is a mathematical representation of vertices and edges. You're a vertex, and you're connected to your friends and the things you like. You will also become related to the things you've read, games you've played, songs you listened to, and so on.

All of that information can be used to filter the newsfeed to surface relevant information. It should also enable data scientists to help you discover new music, better TV shows, and just in general have a better time with the long tail of content. The objective is better experiences.

You've been the beneficiary of such technology long before Facebook was ever created.

Consider how Google uses the web graph to generate relevant experiences, specifically, how to find content. Google's key insight was to add information to webpages based on how they are connected to one another. They use this information, among many other pieces of information, to make recommendations to you on search terms. Google returns hundreds of pages of results for some words. Most people click on one of the top 3 returns. People are generally satisfied with the results returned by Google. Google's algorithm is called PageRank, and it operates on Graph Data.

Facebook, too, uses an algorithm to optimize your newsfeed. It's called GraphRank, and it will operate on more complex information. In theory, the more you share through Facebook, the better GraphRank will become, personalized to you.

The aim is to compete on relevance.

Competing on Relevance

GraphRank will be taught through user control. Users will decide who their close friends are and who their acquaintances are, and list them as such. Even though most people 'don't make lists', many people will make lists if they figure they'll get a return on that effort. Users can easily decide to censor those that share too hard or share too much. It may very well be the case that there's no such thing as sharing too much, it may be the case of sharing not enough relevant information.

Brands, too, will have to compete more than ever on relevance. The degree of user control means they can slap a brand with censorship without 'unliking' them. The flip side is that never before has there been so much intelligence to understand an audience (non-PII). This linkage between analytical intelligence and communication strategy is better than ever.

It used to be that being boring was a good way to reduce risk. Being boring may actually be more risky. Nobody is going to be rewarded with endorsement and engagement by being boring. But what constitutes 'interesting', what constitutes relevance, varies by audience.

What constitutes 'interesting' can be observed directly from the product graph, especially with expanded linkages and data.

Privacy and Utility

When the product is free, you are the product.

Whenever I publish a web page, a blog posting, or put up a site - it is scanned and indexed by Google. Google helps people find my site. I am the product. But I get utility. I have the ability to suggest to Google not to scan my website (blogger excluded), however, I have to do something to opt-out, instead of doing something to opt-in. Of course, I run the other way. I use Google Analytics because it's worth the value exchange.

Whenever I publish an update to Facebook or exhibit a valuable behavior, it is scanned and indexed by Facebook. Facebook helps people find things I like, and developers who create new apps will help me discover new content. I am the product. But I get utility.

So long as Facebook educates people, and people educate themselves, about the privacy they are exchanging to Facebook in exchange for utility, we have a value exchange. There is consent from the user. There is consent from Facebook.

Users ought to have control over what they share, and what they don't. Utility will adjust as a result. Tradeoffs.

Conclusion

A lot happened at F8 to GraphRank.

There will certainly be a reduction in friction. There will most certainly be an uproar around privacy and informed consent. There will most certainly be an explosion of really awesome, useful, functionality forthcoming to Facebook.

Random Finds You

2011-09-21T15:39:00.000-04:00

The longer you look for something, the more chances random has to find you.

To demonstrate:

I drew the blue lines in a pseudo-random way using Google Correlates awesome Search By Drawing function.

That is to say, it's impossible for anything that I draw to be truly random. I'm biological and my brain-hand coordination is nowhere near the same degree of randomness as background radiation.

However, I drew the lines without a conscious design. It wasn't governed by a model or a theory, like the one that powers the S-Curve.

I made 5 runs. Google returned 3 correlations, pictured below.

There are a few pretty good R's here - all exceeding 0.66.

Moderate correlations can be erroneous in very specific ways - missing small peaks or having slight phase variances. And it's possible that Google Correlate won't return anything with a correlation 0.6000 (the lowest I managed was 0.605.)

Google didn't return correlations for two attempts.

I drew those lines without a specific theory in mind. Yet, correlations emerged. I suppose, after the fact, you could construct some sort of theory to explain why a random doodle works. But it's not informed by a theory.

A lot of pure data discovery generates seemingly random findings, and there are methods to assess variance. It involves cutting your data into two, exploring your theory on the training set, and validating any emerging theory on your testing data set. That reduces the risk. It doesn't eliminate it altogether.

Such studies have been executed on massive health files to turn up correlations between astrological sign and breaking bones. In fact, in much the same way that buying thousands of lottery tickets increases your chances of winning, attacking a dataset millions of times with varying functions can turn up significant correlations that survive the testing data set! Over-training ahoy!

This doesn't invalidate exploration.

I'd argue that it's far less risky to go into a dataset with a theory / model in mind. The likelihood of getting caught in a random trap of your own imagination is less likely.

The future belongs to the companies and people that turn data into products

2011-09-15T15:25:00.000-04:00

"The future belongs to the companies and people that turn data into products."
- Mike Loukides, O'Reilly Radar, June 2010.

One of my favourite thinkers, Mike Loukides, repeated and expanded on that today. And a good thing too.

It's not as though product development alone is easy. Even when armed with the tears of hundreds of thousands of developers as a reference guide, you're bound to contribute several of your own to the corpus.

Google Search is the most familiar example of turning data into product. And it has a few effects you know about.

We all know several dozens of people who can explain SEO in high detail. I know only four people who can describe the mechanic behind PageRank in high fidelity - and they no longer practice SEO at all (or exclusively) anymore. Yet, users of Google don't care how relevant results are delivered. They just care that relevant results are delivered. Marketers don't care how the algorithm works. They just want to be ranked first.

How do we make the data dissolve into the experience?

That's the question. There's no single answer.

Just lots of social and physical technologies and very open surface of possibilities.

There's an intersection at aesthetic, execution, and market. (I'm not the first to note this.)

It's is a lot like solving a linear system of equations while painting the Mona Lisa while shouting 'just set it and forget it'. It's relatively difficult for an organization to perform all three functions, and have them balanced both in absolute terms and temporally.

To expand on that point - everybody is subject to the tyranny of the Production Possibility Frontier. Resources are limited. Not everybody values aesthetic. Others do. Choices have to be made. And hidden in the nooks of those degrees of freedom is error.

Moreover, firms always confront the need for short-run profitability and long-run sustainability. It's very easy to incur a massive amount of technical debt when you're trying to keep the lights on and meet payroll. It's very hard to say no in the short run for long-run sustainability. Such tradeoffs are accentuated by the stock market in particular, and within startups in general.

It's because it's risky that the rewards are huge

Turning data into product is hugely rewarding. And can be hazardous.

Ultimately, hopefully, it's through far better analytics design and data science that we can make things better and solve major problems.

A brief commentary on tagging issues

2011-09-13T13:58:00.000-04:00

Four Daily Criterion

2011-09-12T17:18:00.000-04:00

Darryl Ohrt consistently produces very divergent and relevant content. I've been visiting his blog regularly for about a year, and it's pretty awesome.

He founded an agency, Humongo, some 15 years ago.

He recently left.

He wrote about it in AdAge.

What caught my eye were four questions he asks himself:

"For me, the ultimate decision came down to a few factors. I found that I couldn't answer "yes" to all of these questions:

Are you doing something that scares you nearly every day?
Are you having fun every day?
Are you proud of what you've accomplished at the end of the day?
Are you consistently learning something new?"

These are pretty good criterion.

The first criterion is about comfort zone. If you're not leaving the comfort zone consistently, you're not growing, and chances are, you're risking very little.

The second criterion is about sustainability. If you're not having fun, you're going to burn out.

The third criterion is about progress. If you can't point to any, and your no closer to your mission, then you're not progressing.

The fourth criterion is about growth. If you're not adding to yourself, you're not growing.

The unit of 'day' is an interesting one.

Clearly, for Darryl, he had gotten into mud for quite some time. It had reached a threshold.

Since we're all subject to be anchor-and-adjust phenomenon, it might be good to record how you feel at the end of each day. Answer each of the four criterion, and if you quantitatively see a negative trend, take action.

We spend a lot of time measuring the satisfaction of everyone else. Perhaps it's time to start measuring your own.