***** Well-written, technically expert and great business sense. Recommended.
Avinash Kaushik’s Web Analytics 2.0 deserves its five stars. It is packed full of the wisdom of his expertise and experience. Many field experts could write a technically competent ‘How To’ book, however, what really makes Web Analytics 2.0 stand out is that it maintains great awareness of the needs of the business. Again and again, Kaushik reminds the practitioner to focus on identifying actionable outcomes and to make clear what the value of these outcomes are. I would recommend this book not just to web analysts but to any modern marketing professional.
In the spirit of the book, lest it self-combusts, here are my Critical Few takeaways:
The 10:90 rule: Spend just 10% of your budget on analytics tools, the other 90% on the smart people who can help you mine information from the data. Hire people with good critical thinking rather than for expertise in a tool. A capable person can learn a new tool quickly.
Components of Web Analytics 2.0: Clickstream Analysis (Web 1.0) can tell you What is going on online, Outcome Analysis puts a Value on what is happening, Multivariate Testing explains Why customers behave a particular way, Qualitative Analysis (e.g surveys) allows you to tap the Voice of the Customer, Competitive Analysis tells you how you are doing compared to your competitors.
‘Hits mean nothing’: Today’s important metrics are connected to business outcomes, for example, conversion rate, depth of visit, visitor loyalty. Learn which actions can be taken to change customer behavior.
‘Segment or go home’: Segmentation and testing are the key to understanding customer behavior. Experiment and learn. Waiting for complete or perfect data leads to paralysis by analysis. Don’t get hung up on this. ‘Move fast, think smart’ – an educated mistake is better than no action at all.
‘Don’t puke data’: Provide information, not data. Focus on the Critical Few metrics (3 or 4), not the insignificant many. Connect the outcomes to business performance. Recommend actions. Recall the Web Metrics Lifecycle Process: Define, Measure, Analyse, Take Action, Improve/Eliminate.
*** Pleasantly written, concise, but lacking substance
Seth Godin has a gift for branding the ideas he wants to convey in pithy, memorable terms, the phrase ‘Purple Cow’ being a prime example.
A Purple Cow would be a remarkable thing. Hence Seth’s invocation to marketers and innovators: build the ‘remarkable’ into your products and services and do so from product inception and design. Do not wait until the product is made to involve yourself in the marketing of the product. The reason the remarkable is necessary is that consumers are saturated with information and products. It is hard to attract their attention or interest. You should aim to win over a niche market with your remarkable offering first and to use the buzz generated to attract the attention of the larger market.
Seth asserts that there is not a lack of remarkable ideas, rather a lack of the will to execute them. He states that practitioners have to be brave to stick to the Purple path because conventional business practice will steer them to develop products which are inoffensive but boring.
This is the third Seth Godin book I have read, after Lynchpin (four stars) and Poke The Box (four stars), yet, I want to give this book only three stars. Seth writes as stirringly as ever and the physical book is as remarkably well-produced as his other books. While I agree with the main theme, there is little tactical guidance to help the innovator build and manage a Purple Cow process. The edition I read had a ‘bonus’ chapter of post hoc examples of businesses and organizations offering remarkable products and services. I felt this was just padding.
If you are looking for a more comprehensive knowledge resource for this subject area, one which is full of tactical advice, I recommend Geoffrey Moore’s masterful, ‘Crossing The Chasm’, which is itself referenced multiple times by Seth in Purple Cow.
**** An enjoyable read and good introduction to the Predictive Analytics industry for those with a non-technical background
This book would suit those who want to understand the context of Predictive Analytics and its applications. Readers are presented with many examples of PA in action and high-level explanations of different analytics approaches. The author, a successful PA practitioner, writes in an engaging style and illuminates the topics like a journalist who understands his subject matter well.
This book is not for those aspiring to acquire technical skills in Data Science. There are other resources which are better suited to that – try Coursera’s online courses: ‘Machine Learning’ from Stanford or ‘Introduction to Data Science’ from the University of Washington (both free) instead. Data Smart by John Foreman, although it eschews formal theory, is an excellent introductory text for gaining understanding of Data Science through hands-on practice with data problems.
My takeaways from Predictive Analytics:
“‘Knowing is not enough. We must act’ – Goethe”
‘An estimated 40% of London Stock Exchange trading is driven by algorithmic trading’
‘35% of Amazon’s sales come from recommendations.
‘The objective of Machine Learning is induction’, i.e. reasoning from detailed facts to general principles. Induction is much harder than deduction (‘reasoning from a general rule to a particular instance’) because assumptions are involved. If the assumptions are over-simplistic, we have inductive bias.
Decision Trees and Ensemble Models are explained at a high level, liberally illustrated with examples, in Chapters 4 and 5 respectively. The story of how IBM’s Watson computer won TV’s Jeopardy is well researched and engagingly told in Chapter 6.
Uplift Modeling, the science of segmenting clients into those who will respond positively or negatively to marketing communication campaigns, is well described in Chapter 7. I appreciated the accompanying 2×2 matrix illustration.
In conclusion, Predictive Analytics will provide the reader with a wide-ranging description of the technology and its applications, several examples of analytics in practice and some high-level familiarity with the algorithms and terminology employed.