Web Analytics 2.0: The Art of Online Accountability and Science of Customer Centricity 1st Edition

Web Analytics 2.0 is not a sequel to Kaushik's first book Web Analytics: An Hour a Day.The latter was a hard core offering that covered all aspects of the subject.2.0 is a more general book that covers a wide range of topics related to and around Web Analytics.The coverage of Social Media and Mobile analytics is sparse and that's my only gripe. Considering that both topics are quite hot and that Social Media has gained maturity it would have been helpful to have both these covered in depth. That said the book is pretty robust in its coverage of a wide mix of topics. The list of tools mentioned is also quite exhaustive.Key Takeaways· Paid Web Analytics providers are better than the free ones if you need advanced reporting. The other reason is that the paid tools integrate well with other allied offerings/tools. (A project that I'm working on validates both these points)· Data needs to be actionable. No point collecting old data if the business cannot use it· Keep an eye on the competition using Google Insights For Search(contains search keyword data on[...] only), Google Trends (contains broad web usage data), Compete, Hitwise. Also check Google Ad Planner and Quantacast since both use self reported data. Most analytics tools now allow you to benchmark against specific verticals.· Use tools like page level/site level surveys to gather user feedback(kampyle, uservoice, opinionlab). The Voice of the Consumer is necessary to fill in the gapsNow if only we could get key sales and marketing folks to read this book and understand how much data is there for them to useToolsWeb Analytics: Omniture, WebTrends, CoreMetrics, Google AnalyticsMobile Analytics: Bango Analytics, [...], [...]Experimentation and Testing: Google Optmizer, Omniture Test and Target, Optimost, SitespectVoice of the Customer: 4Q, iPerceptions, ForeseeResults, EthnioCompetitive Intelligence: Google Insights For Search, Google Trends, Compete, Hitwise, Technorati, Google Ad Planner, QuantacastAnalytics Tags Audit: SiteAudit(ObservePoint)SEO gaps, Web Application Performance Management, more : Maxamine, CoradiantPage level/site level surveys to gather user feedback: Kampyle, Uservoice, OpinionlabUsability: Ethnio, UsertestingAnalyze Actual Online Experiences: Tealeaf, ClicktaleInformation Architecture: OptimalSort, [...]Visual heat maps: Feng-gui.com, CrazyeggKeyword Analysis: Google Adwords Tool, Wordtracker, KeywordSpyOnsite Behavior Targeting Platforms: Audience Science, kefta, Netmining, BTBuckets(free)Paid Search Tools: Marinsoftware, Kenshoo, ClickEquationsFor this and other Web Marketing articles, my blog: [...]

Although it is only January, I can say with a pretty high degree of confidence, that the analytics book of the year is Avinash Kaushik's new book Web Analytics 2.0. As a follow up to his first book, Web Analytics: An Hour A Day, I think Avinash out did himself with this book.Web Analytics 2.0 is a nice balance of theory and practical tips, but more importantly, provides guidance for readers with a wide range of skill and experience. Over the 14 chapters of the book, Kaushik covers almost every aspect of web analytics, from competitive analytics to optimization, even guides for picking a solution vendor and starting a career in the industry.Overall the book is an easy read for anyone interested. Avinash's causal writing style and frequent examples makes the text engaging and entertaining. For the most part, you can dive into any area of the book without missing too much context from the rest of the book. Of course I recommend reading the whole thing, but even if you just read one chapter, say on social measurement, you'll still get a lot out of it.On top of being a well-written resource, all of the profits from this book (and his previous one) go to two charities: The Smile Train and The Ekal Vidyalaya Foundation. Get the book and start the new year off right with a quick education on the in's and out's of web analytics.

Lots of good information, but there are no descriptions for any software or how to get the reports seen in the book. I am trying to recreate these reports using Google Analytics, Coremetrics and Omniture. It seems that most of the reports are the standard reports out of Google Analytics, but I am having a difficult time recreating some of these with other software.I think this was a great book, but I have a few things I disagree with:Page 85, he says if he could only have one report, it would be Outcomes by All Traffic Sources. This report shows Goal Conversion Rates, but he does not describe what these are. In Google Analytics, these are custom, so this could be anything.I am disappointed, he does say it is important to measure ROI, but does not talk about how to do this. The author says that you can do this by comparing the data from Google to your campaign data. It is not that easy. You have to know how much was spent, and you have to know how much incremental revenue came in from SEO/PPC efforts. It is not an easy task. Test and control or some other method should have been addressed. In calculating ROI for PPC in chapter 11, he assumes that all visits from PPC are ones you would not have without the ad. Not necessarily true.In Chapter 7, testing is finally addressed. I disagree with his method of testing the impact of PPC by turning it off and on completely; this does not take into account any seasonality that may occur naturally in web traffic. This is also a problem if there is a lot of variation in web visits and sales over time. Why not try test and control markets: turning it off in some regions and have it on in others? This method would allow you to compare the on and off markets and find incremental sales.In the marginal attribution model from page 368, you change the spending for one type of online marketing, then attribute any sales higher than last month sales to the additional marketing. In my experience, web sales tend to have a large variation in sales from month to month making it difficult to say what the cause of any increase is without any kind of confidence bounds.The "controlled experiment" on page 375 is a really bad example. The ad is run at the same time in all markets and then compared to pre and post ad time periods. What if at the same time as the ad, some celebrity tweeted that they loved your product or some news program aired a warning about your product. There are too many uncontrollable situations to compare pre and post ad sales. You should have test and control markets to compare sales in the same time period.On page 377, the Author says: "The analyst at Walmart.com can use the previous URL to track how many people use the website and then visit the store." A view the store locator on the web does NOT equal a visit to your store. In his example, a user on walmart.com views a camera and then the store locator. It is very possible that the customer viewing the camera at walmart.com may also go to target.com and find the same camera at a similar price and find that the target store was much more convenient to visit. There is no way in this case to tie a store locator and product page view to an offline purchase. Using a discount code or unique offer would provide a better method of tracking online to offline behavior.In Chapter 14, the BMI is introduced. But on page 419, the author says this method is preferred because it has a scale of 0 to 100. It actually has a scale of -100 to 100.If 5 responders all gave a Not Satisfied or a Not At All Satisfied, the score would be [(0+))-(5):]/5*100=-100. The other method, weighted means can also give a scale of -100 to 100 if the right weights are used.Not Satisfied At all:=-1Not Satisfied =-.5Satisfied=0Very Satisfied= .5Extremely Satisfied= 1With these weights the scale is also -100 to 100.