Marketing Analytics

Are You an Average Data Analyst?

2008-02-27T05:31:00.000-08:00

A Quote from the KDnuggets Newsletter:

Einstein was an "average" data analyst, with tendency to make trivial arithmetic mistakes and some clumsiness in data recording.

From Iglewicz, B., (2007) "Einstein’s First Published Paper," The American Statistician, Vol. 61, No. 4, 339-342. - thanks to Bruce Ratner

Something is wrong here.....

2008-02-18T08:51:00.000-08:00

So I've read on Techcrunch that Yonja raised $12.5 M in funding. I used to be a member to this "Turkish" social networking site - haven't used it over a year. Judging from alexa; I wasn't the only one.

Based on this alexa chart Yonja's traffic reached it's peak in the first half of 2006. The decline accelerated in the second half of 2007 - and I bet that's when Turkish surfers realized facebook is accepting registrations from everyone.

I never understood what Yonja's competitive advantage was. It was mainly a social networking site for Turkish people, albeit designed in English - a Turkish version was created later. They had a paid subscription model where people could have their profiles posted on the homepage therefore could get more traffic - which is usually popular for matchmaking sites but not social networking sites.

It doesn't take a lot of sophisticated time series modeling to predict what's going to happen to Yonja. It brings memories from my old Lycos days - which ended in 2003... My prediction is Yonja will disappear from the social networking scene by the end of 2008. Then again, my prediction may be totally off... I will follow up on this post.

CRM for BlackBerry... Now We're Talking....

2008-01-21T12:09:00.000-08:00

Recent articles published on IDM and WaMN there's a new CRM software available with a twist. Alltel has a new CRM application targeting Small and Medium Businesses with BlackBerry devices in need of contact, activity and deal management.

The new software is called SalesNOW, and it helps Blackberry users to manage their sales data, contacts, deals, activities and emails on the move. Users can also access and share information with their teams.

SalesNOW has three different versions:

There's a 30 day Free Trial - and I will definetly try this one out - however plans seem a bit pricy...

Netflix Marketing Metrics

2008-01-18T12:22:00.000-08:00

In his data mining blog Kevin Hillstrom provides projections for Netflix up to year 2012. This is a hot topic for a lot of people since we all know Netflix has a reputation for being a very analytically driven company. (I have been ignoring the Netflix Data Mining Challenge for too long - I shall post about it in the near future - the leader board is still showing 8.x % improvement; here's Steve Krause's blog post about the challenge).

I charted Acquisition Cost per Subscriber over Annual Pre-Tax Income... Is this a good picture? You decide...

Another Comment on "Advanced Marketing Analytics"

2008-01-18T07:05:00.000-08:00

I came across this article on crm2day.com, written by the CEO of Pluris -I have never heard of them before, although I lived in MA (where they have their headquarters) for 7 years and worked in Marketing Analytics for 5- a CRM agency / consultancy.

Probably the boldest statement in the article is:

Analytics is no longer a “nice to have” tool but a “need to have,” integrated component of a marketer’s overall execution platform.

However the article only lists the traditional direct marketing models that have been deployed since the 1960s

Knowledge about a customer’s likeliness to purchase each and every product.
Knowledge about a customer’s likeliness to upgrade from a current product/service to a new package/service.
Knowledge about a customer’s expected churn rate and timeframe.
Knowledge about a customer’s expected lifetime value.

Another point that's being made -again not a surprising point; actually a pretty well known issue- is:
...the analytical capability and integration must be truly automated, saving the modeler over 80% of the his/her time dealing with data integration and scoring issues and letting the advanced analytic resource focus on providing value.

Rest of the article is about optimization of marketing activities as well as inbound channels. It's worth a read...

The Missing Link: Marketing Analytics

2008-01-09T04:19:00.000-08:00

D. Santaro's article on DM Review is a bit dated but has a lot of important information in it. His main observation (which I strongly) agree is "most CRM spending focused on information-centric technology components such as data warehousing, data aggregation, content management and reporting; or operational customer touchpoint elements such as contact center management, sales force automation, personalization or marketing automation".

Nowadays; Analytics, which has been the "secondary subject" up to now is moving fast ahead to become the #1 priority for many organizations. The article contains a neat chart for different vendors:

The article also has a very straight forward definition for predictive analytics: "organizations reach these goals by applying basic math and understanding that any consumer behavior (Stats 101: Dependent Variable) can be predicted with statistical precision using historical observations of actual behaviors and interactions (Stats 101: Independent Variables)". Here's the chart that goes with it:

IBM alphaWorks Services

2007-04-05T18:21:00.000-07:00

Try IBM alphaWorks - you will like it...

Web Analytics: Webtrends Leads

2006-08-28T10:51:00.000-07:00

According to the Forrester Wave of Web Analytics -requires registration for download- Webtrends is leading the web analytics category. (Please keep in ming that this report is available from Webtrends and always check Wikipedia for a list of cognitive biases)

Other Web Analytics software players evaluated include:
Visual Science
Omniture
Core Metrics
Web Side Story
Sane Solutions
IBM Surf Aid
SPSS NetGenesis
NetRatings

According to Forrester Webtrens lead the industry in Strategy (product direction, commitment, reference client strength, pricing and licencing) and Market Presence (company financials, installed base, international presence). However Omniture has a slightly favorable score in terms of current offerings (capabilities, service and support).

Visualisation: Edward Tufte

2006-08-25T14:08:00.000-07:00

Edwatd Tufte is one of the leaders in information visualisation. Wikipedia defines him as "an expert in the presentation of informational graphics, such as infographics, charts and graphs." When presenting information (yes, that includes marketing analytics) one should always keep Tufte's motto in mind: "ink should only be used to convey significant data and aid its interpretation".

Tufte authored three books on the subject: "The Visual Display of Quantitative Information" (1983), the follow-up "Envisioning Information" (1990) and "Visual Explanations" (1997). According to an article on Salon he has described the trio as being dedicated in turn to "pictures of numbers," "pictures of nouns" and "pictures of verbs". Needless to say I highly recommend his work.

Edward Tufte is also an outspoken critic of Microsoft PowerPoint. In his article (titled PowerPoint is Evil) he argues: "PowerPoint's pushy style seeks to set up a speaker's dominance over the audience. The speaker, after all, is making power points with bullets to followers. Could any metaphor be worse?"

Analytics Related E-mail Groups

2006-08-25T14:05:00.000-07:00

List of analytics focused e-mail groups on the world wide web.

Google Groups:

Yahoo Groups:

E-mail me from analyst@nadirler.com if you would like to make suggestions.

Davenport vs. Raden

2006-08-24T19:58:00.000-07:00

A while back Tom Davenport (author, professor and director of research for the School of Executive Education at Babson College in Boston) published a white paper titled "Competing on Analytics". He basically argued that " increasingly, analytics will have a primary rather than supporting role in competitive strategies".

(You may view the full paper from the following places:Optimize Magazine ; Babson Knowledge Blog -may require registration- ; Spotfire)

According to Davenport "you know you compete on analytics when..."

1. You apply sophisticated information systems and rigorous analysis not only to your core capability but also to a range of functions as varied as marketing and human resources.
2. You not only employ analytics in almost every function and department but also consider it so strategically important that you manage it at the enterprise level.
3. You not only have committed to competing on analytics but also have been building your capabilities for several years.

among 7 other conditions (the full list can be viewed on Babson Knowledge).

However his approach drew critism from the BI circles. Neil Raden (founder and president of Hired Brains) posted his thoughts in an article titled "Power to the People: Analytics for the Masses" (Raden's article can be accessed from his company web site).

Raden argues Davenport's approach focuses on creating an analytical elite (PhD level statisticians) delivering predictive analysis with buy-in from Sr. Management. Instead, Raden says, analytics can be utilized by all sorts of people in organization generating incremental value all over the place (which would eventually lead to sizeable success for the organization). For more information on Raden's philosphy you may also want to review his Analytics Manifesto.

Juice Analytics is another blog (run by Chris and Zach) that disagrees with Davenport's article.
They suggest:

Democratizing data in your organization, making it easier to put more eyes, more experience, more brains against your data is the challenge of the next ten years in analytics.

which conflicts with Davenport's view of centralizing data and analytics staff.

Here are my two cents:
1- This is a good discussion to have. Everyone seems to agree on value of analytics :) The focus is on the approach.
2- Depending on the organization there are upsides and downsides to both approaches. My final answer is "It Depends!".

Statz Rappers

2006-08-19T18:12:00.000-07:00

Open Source Data Mining Software: Proximity

2006-08-16T08:45:00.000-07:00

I had downloaded PROXIMITY ~ 3 years ago.
It's offered by Umass Amherst.

Official Page

PROXIMITY offers a more flexible data representation than many other approaches to knowledge discovery and machine learning. Other techniques assume their input data instances are structurally homogeneous, identically distributed, and statistically independent. These assumptions lead to a “flat-file” data format in which each instance is represented by a vector of features, and all the vectors have the same length. Relational knowledge discovery is designed for the many real-life domains that do not meet these criteria. Objects in a PROXIMITY database can be structurally heterogeneous. They can have a variable number of attributes, and a variable number of values for each attribute. The relationships encoded by links violate the assumption that each object is statistically independent of the others.

Online Courses: Data Mining

2006-08-03T08:24:00.000-07:00

Courses are listed in order of importance.

1) Shapiro's Course on KDNuggets: This is a very comprehensive course on Data Mining by Gregory Piatetsky-Shapiro. It starts with an introduction to Machine Learning and Data Mining and then proceeds to teaching classification (decision trees, etc). Other subjects covered are prediction problems, clustering and visualization.

2) Madigan's Course at Rutgers: This is a "Data Mining Special Topics" class and has information on advanced topics like Bayesian Network, Neural Networks, Bagging, Boosting, etc.
The class page also has links to popular sample data sources on the internet.

3) Patel's Course at MIT: MIT Open Course Ware is a great source of information even if you are not a MIT student. This is an introductry level data mining course including lecture notes, assignments, etc.

4) Introduction to Data Mining and Knowledge Discovery: This is an introductry field guide to Data Mining - feels like the very first chapter of a Data Mining book.

SAS vs. R

2006-07-27T19:48:00.002-07:00

SAS and R are both statistical programming packages. SAS also has additional modules for data management and connectivity.

Many of my friends tell me there's been an uptick lately in new books that focus primarily on R. It also sounds like non-academic statisticians are accelerating their switch to R (mostly from SAS).

It feels like SAS has become really focused on the enterprise-side and not doing the cutting-edge statistical computing work anymore. In contrast R package has some interesting statistical work.
Statisticians (especially in academics) are becoming more and more savvy programmers with every generation. Statisticians who have also done work in C, Java, or scripting languages can appreciate the flexibility in data structures that R offers.

SAS has really become an enterprise database with advanced mathematical/statistical functionality. Most people simply don't need all of the enterprise functionality (or the pricing) it offers and are better off with the combination of R and flat files or an open source DB like MySQL.

However as long as there are enterprises doing statistics, SAS will exist. Many of the employers I know of run SAS shops. However, R is a favorite among biostatisticians, it's becoming better documented, and it's FREE. All good reasons for academic use.

Part of the reason that enterprises prefer SAS is the support- the really fabulous documentation and tech support. With R, you have the R mailing list. You also have to study the R FAQ really well. It was also pointed out to me recently that a company called Revolution Computing supplies a Red-Hat version of R called RPro.

Also, in certain industries, SAS seems to have reached some sort of "trusted" status.

I think that if R truly wanted to compete in that space it would need to improve the database support and some of the memory requirements (which have been improving or at least changing recently) along with maybe some sort of Red Hat-like company to provide the documentation and support (and, if the licensing could be worked out, an "Enterprise-level" user interface, that likely means point-n-click access to certain modules). For addressing SAS's "trusted" status you could go with something like the OpenBSD auditing projects---since the code is totally open you can actually do things like peer review audits. SAS also has pretty extensive reporting facilities, but I think that R's Sweave like tools will be better in the long run (for those who don't know about it, very handy. It lets you do things like include the commands for generating figures in the LaTeX document itself s.t. processing the LaTeX document automagically generates the figures, tables, etc and inserts the appropriate LaTeX code that gets passed on to the LaTeX toolchain to generate the DVI, PDF, etc).

R encourages the good software development practices, especially
the use of test cases and documentation. There is no SAS equivalent to
R CMD check. This is extremely important. You are ensuring that your code
does what you think it does, both today and in the future..

R's graphics are superior to SAS.

Via Sweave, xtable and friends, R makes it very easy to practice
literate programming.

Analytic Marketing

2006-07-18T22:08:00.000-07:00

CRM
Data Mining
Media Mix Modeling