eMetrics is coming Toronto next week. There's still time to register, and I have a discount code if you want to attend. Tweet me at @cjpberry and I'll shoot it on over. This will be my third eMetrics in three years, and as such, I'll offer a few predictions.
The panel I'm moderating on Wednesday will go off swimmingly. There will be some controversy as the panelists tussle over what's really important in the qual/quant mix. There will be enough sparks to ignite some lively debate that evening.
The whole Syncapse Measurement Science team will be there in force that night and on Thursday. They're going to see just how other people present their material and they'll have quite a few takeaways.
Web Analytics Wednesday on a Thursday night will be messy. When isn't?
On Friday I'm presenting on Word of Mouth social analytics and talking publicly about new idea. I will be presenting real data and with a real idea. You have to come to the session to find out.
Then there's a dynamite panel right after featuring Jim Novo, Stephane Hamel, Jacques Warren, John Lovett and myself. It's the first major panel in Canada following the Carrabis posts on the matter and should be really relevant.
Major stories emerge in the hallways during the conference. Several major projects get started, deals are done, products are sourced, and thoughtcrime is exchanged. At previous conferences disagreements bloom into much of the blog discussions (like this one).
I think the most major discussion will be the sentiment problem and some of the broader reconciliations between brand and action. Certainly, Sterne has put forth his point of view. Avinash has offered his. My team and I have offered up ours. There will most certainly be an effort to reconcile the views into a coherent model.
eMetrics continues to be an important gathering of people, product and ideas. It's an interesting gathering of different types of people.
Agendas will be set. They always are at these events. Like it or not. Products and features will be announced. They always are. And knowledge will be spread. It always is.
So I'll repeat the action I'd like you to take. If you want to go, tweet me at @cjpberry and I'll shoot you a promo code.
Wednesday, March 31, 2010
Wednesday, March 17, 2010
300 Years of Science and Word Of Mouth
Pat LaPointe wrote a pretty interesting article for MediaPost Publications. You can check it out here.
My response is pretty much 'Yes, And...'
I don't understand why some people are making inductive inferences that online word of mouth is somehow reflective of offline word of mouth. (As a certain company appears to be making). I share his concern and skepticism.
Let me unpack that.
A whole generation of quantitative market researchers are supposed to understand that if you take a small, random sample of a population and expose them to a treatment, then you can make an inductive inference on how the entire population will react to that same treatment. The probability that the inductive inference is accurate is a function of the sample size and basic probability.
This is why when we randomly call 100,000 people, and get 1000 responses, we make this inductive leap that those 1000 people are reflective of 300,000,000 million people.
This is the field called sample statistics. It's an inductive science. And it has it's problems. And it has it's benefits.
I'm stunned that any firm would take a word of mouth dataset from the Internet and infer that it is reflective of what goes on offline. I don't believe that there an inductive leap can be made that way. (Moreover, there's a deeper problem with the self-selection bias that happens in the survey methodology, but that's another rathole for another time).
However, I will say that online word of mouth analytics has much more in common with data mining than sample statistics. Remember the reason why sample statistics was invented in the first place? It was because we couldn't possibly collect, store and run algorithms on such massive datasets.
Now we can. Giddyup.
Now, the next part.
I disagree that Word of Mouth (WOM) research is still in its infancy. We've inherited a very rich base of literature and understanding about how WOM really works, especially as applied to marketing and commerce. Being ignorant of that literature doesn't mean that we're all in our infancy. It just means that most people are in their infancy in terms of understanding.
Contemporary WOM research is now some 60 years old - and has been intensely studied from the 1960's on. The number of databases we have to go off of is very deep, and debates within the Marketing Science community are nuanced and relevant. I'll be touching on just two of those debates at eMetrics Toronto on April 9th.
There is much triangulation in methodologies in grappling with social media measurement, and indeed, brand measurement. We have approaches that span computer science, brand measurement, direct measurement, database marketing measurement, linguistics, decision neuroscience, psychology, marketing science, web analytics and data mining - just to name a few. We're there. We've been there for a long time now.
I don't believe that anybody really wants to hear how complex it is. These are problems that leaders will solve. The market will always reward people who will make the complex - simple.
In sum, prepare to be even more annoyed, Pat. Breathe deeply. We're in for a very exciting decade.
.
My response is pretty much 'Yes, And...'
I don't understand why some people are making inductive inferences that online word of mouth is somehow reflective of offline word of mouth. (As a certain company appears to be making). I share his concern and skepticism.
Let me unpack that.
A whole generation of quantitative market researchers are supposed to understand that if you take a small, random sample of a population and expose them to a treatment, then you can make an inductive inference on how the entire population will react to that same treatment. The probability that the inductive inference is accurate is a function of the sample size and basic probability.
This is why when we randomly call 100,000 people, and get 1000 responses, we make this inductive leap that those 1000 people are reflective of 300,000,000 million people.
This is the field called sample statistics. It's an inductive science. And it has it's problems. And it has it's benefits.
I'm stunned that any firm would take a word of mouth dataset from the Internet and infer that it is reflective of what goes on offline. I don't believe that there an inductive leap can be made that way. (Moreover, there's a deeper problem with the self-selection bias that happens in the survey methodology, but that's another rathole for another time).
However, I will say that online word of mouth analytics has much more in common with data mining than sample statistics. Remember the reason why sample statistics was invented in the first place? It was because we couldn't possibly collect, store and run algorithms on such massive datasets.
Now we can. Giddyup.
Now, the next part.
I disagree that Word of Mouth (WOM) research is still in its infancy. We've inherited a very rich base of literature and understanding about how WOM really works, especially as applied to marketing and commerce. Being ignorant of that literature doesn't mean that we're all in our infancy. It just means that most people are in their infancy in terms of understanding.
Contemporary WOM research is now some 60 years old - and has been intensely studied from the 1960's on. The number of databases we have to go off of is very deep, and debates within the Marketing Science community are nuanced and relevant. I'll be touching on just two of those debates at eMetrics Toronto on April 9th.
There is much triangulation in methodologies in grappling with social media measurement, and indeed, brand measurement. We have approaches that span computer science, brand measurement, direct measurement, database marketing measurement, linguistics, decision neuroscience, psychology, marketing science, web analytics and data mining - just to name a few. We're there. We've been there for a long time now.
I don't believe that anybody really wants to hear how complex it is. These are problems that leaders will solve. The market will always reward people who will make the complex - simple.
In sum, prepare to be even more annoyed, Pat. Breathe deeply. We're in for a very exciting decade.
.
Tuesday, March 9, 2010
Pixels and Fail
Consider the impact of the mechanical clock and the curved lens on early analytics.
The mechanical clock enabled consistent, scale, time.
You won't optimize what you won't measure. And Europeans most certainly started optimizing time.
They've been optimizing work per time unit, productivity, since the renaissance.
Countries that didn't have a method of measuring productivity simply didn't optimize it. Worse, cultures that didn't value the standardization of time simply didn't value productivity. Why care about productivity when you have loads of population to toss at a project? It put whole swaths of the globe at a competitive disadvantage.
The curved lens, aside from giving us astronomy and microbiology, enabled great strides in miniaturization and productivity. A skilled worker could work through to middle age thanks in no small part to glasses. The pocket watch was to then as the cell phone is to us today. Ever smaller. Ever more accurate. Ever more personal.
Therein lies a judgement about culture, time, incentives, and productivity.
These are great inventions that had a huge impact on competitive advantage.
Is creative destruction within an economy so violent because cultures simply won't change their value systems in response to productivity shocks? In that sense, are they really made of pixels and fail?
.
The mechanical clock enabled consistent, scale, time.
You won't optimize what you won't measure. And Europeans most certainly started optimizing time.
They've been optimizing work per time unit, productivity, since the renaissance.
Countries that didn't have a method of measuring productivity simply didn't optimize it. Worse, cultures that didn't value the standardization of time simply didn't value productivity. Why care about productivity when you have loads of population to toss at a project? It put whole swaths of the globe at a competitive disadvantage.
The curved lens, aside from giving us astronomy and microbiology, enabled great strides in miniaturization and productivity. A skilled worker could work through to middle age thanks in no small part to glasses. The pocket watch was to then as the cell phone is to us today. Ever smaller. Ever more accurate. Ever more personal.
Therein lies a judgement about culture, time, incentives, and productivity.
These are great inventions that had a huge impact on competitive advantage.
Is creative destruction within an economy so violent because cultures simply won't change their value systems in response to productivity shocks? In that sense, are they really made of pixels and fail?
.
Monday, March 1, 2010
Hal Varian is right.
"Chief information officers (CIOs) have become somewhat more prominent in the executive suite, and a new kind of professional has emerged, the data scientist, who combines the skills of software programmer, statistician and storyteller/artist to extract the nuggets of gold hidden under mountains of data. Hal Varian, Google’s chief economist, predicts that the job of statistician will become the “sexiest” around. Data, he explains, are widely available; what is scarce is the ability to extract wisdom from them."
Source: The Economist, Feb 25, 2010
Yes, Hal. Yes.
Statisticians are certainly a sexy lot.
I coined the rather curious term stratistician - a cross between a strategist and a statistician, over a lunch with Mark Dykeman on Friday. A lol ensued. Then silence. I think too many of us would confuse the term with former WWE superstar Trish Stratus though.
Not a bad impression to press into that brain. (And yes, that was the tamest picture I could find of her. Go do a Google Search.)
Personally, this notion of the scientist-practitioner, the data-scientist, the stratistician - the notion of an analytics beyond reporting - has been a major source of inspiration. I feel like there's a path out for so many of us who can't seem to find....wait for it....stratisfaction... in our daily lives. Last pun I promise.
(I went there)
Data is, indeed, widely available. The incidence of people who actively practice statistics while actively practicing programming is very thin. Thankfully, the languages of statistics and programming are starting to come together, and this is very good news. It's going to become increasingly easier to maintain both skillsets at a competent level.
The storytelling is another story. As is business strategy. I don't think you can ever have both of those skills completely mastered all the time. People always want to be told the same thing five times and in a different way that keeps them entertained and engaged. Business strategy has fundamental principles - and the particulars tend to drift.
Being that quadruple threat: knowing business strategy, knowing storytelling, knowing statistics, and knowing programming - is possible. It's illusive. And it's possible.
We've got a big challenge.
But at least it's going to be sizzling.
I have resisted the urge to type out one last pun.
Your welcome. ;)
Source: The Economist, Feb 25, 2010
Yes, Hal. Yes.
Statisticians are certainly a sexy lot.
I coined the rather curious term stratistician - a cross between a strategist and a statistician, over a lunch with Mark Dykeman on Friday. A lol ensued. Then silence. I think too many of us would confuse the term with former WWE superstar Trish Stratus though.
Not a bad impression to press into that brain. (And yes, that was the tamest picture I could find of her. Go do a Google Search.)
Personally, this notion of the scientist-practitioner, the data-scientist, the stratistician - the notion of an analytics beyond reporting - has been a major source of inspiration. I feel like there's a path out for so many of us who can't seem to find....wait for it....stratisfaction... in our daily lives. Last pun I promise.
(I went there)
Data is, indeed, widely available. The incidence of people who actively practice statistics while actively practicing programming is very thin. Thankfully, the languages of statistics and programming are starting to come together, and this is very good news. It's going to become increasingly easier to maintain both skillsets at a competent level.
The storytelling is another story. As is business strategy. I don't think you can ever have both of those skills completely mastered all the time. People always want to be told the same thing five times and in a different way that keeps them entertained and engaged. Business strategy has fundamental principles - and the particulars tend to drift.
Being that quadruple threat: knowing business strategy, knowing storytelling, knowing statistics, and knowing programming - is possible. It's illusive. And it's possible.
We've got a big challenge.
But at least it's going to be sizzling.
I have resisted the urge to type out one last pun.
Your welcome. ;)
Subscribe to:
Posts (Atom)