Review – Web Analytics An Hour a Day

Avinash Kaushik’s new book Web Analytics An Hour a Day has been on my bookshelf for a couple months (not for lack of interest, but rather because I was too busy with the launch of our new site, my web analytics course, holiday and family visits). Finally I got a chance to sit down with it over August and give it a proper read. A few observations:

What I loved:

1. The structure of the book works for both a beginner or more advanced analytics reader. There’s a place for everyone to get started and a lot of great actionable ideas you can use today.
2. Avinash is truely a customer  focussed guy. I loved how throughout the book he puts the visitor/customer first and really wants you to understand that in order to get the most from your data, you need to be constantly trying to understand your visitor’s intent.
3. Throughout the book he tells you “how” to do stuff. Take for example p. 147 where he suggests a deep dive into search engine related traffic, then proceeds to show you a model you can use. While I would have liked more of this (and at times more detail or additional examples), the “how” is rarely seen in many analytics books.
4. “Month Two” of the book – the whole chapter is great. Trust me, read it.
5. “Month Three” on Search (internal and external). Lots of suggestions here worth applying to my day-to-day work (look out Taylor).
6. His reminders to “never present metrics without context”. Something I first heard from Eric Peterson.
7. “Month Eight”, in particular the section on path analysis.

What needed more explanation or could have been better:

1. Bounce rate (Percent of traffic that stayed on your site for fewer than X seconds). I’m a big fan of bounce rate (I measure as Single Access Visits to a page / Entry Page Visits to the same page). Avinash recommends the use of time, not number of pages, to compute bounce rate. The challenge I often have with this is that I don’t have a site with visits less than 10 seconds/30 seconds/1 minute and the determination of the time increment to use can be a challenge itself and for the new analyst. Looking at clickstream and outcome data to determine the threshold (as suggested on p.144) is unclear. Would have liked some more insight into making this selection.
2. The six-sigma section was a bit dry. Robbin Steif had a similar comment on her LunaMetrics blog the other day.

What I learned

1. While we’ve been tracking content effectiveness and monitoring changes, I do need to more attention to my top 10 entry pages and leverage them to highlight promotions/campaigns.
2. I need to do a better job incorporating other data sources into my analysis (e.g. call centre data, CRM, etc.)
3. SEO – leverage your internal search key phrases
4. BONUS: While reading a section on conversion rate and viewing one of my reports, I discovered was calculating conversion rate for our site incorrectly in the report.
5. I have a lot still to learn as an analyst, but that’s exciting.

In closing

• Kudos to Avinash donating his royalties on the book. As someone who formerly worked in publishing I still think he should have self-published it in a PDF format like Eric’s Big Book of KPIs or 37Signals Getting Real — he would have likely raised more money.
• I really liked the book. I kept wishing Avinash was sitting next to me available for questions, ready to take over my keyboard and mouse and show me “how” he does his work. Perhaps one day I’ll get that chance.
• BTW Avinash: is that me on p.348?