I deliberately haven't referred to the book title yet since readers might exclude it because it seems either too geeky or specialised from the web analytics in the title or because it appears to be a simplistic, a ‘dummies guide’.

Avinash Kaushik’s Web analytics An Hour a Day (See 5* reviews on Amazon US), is is quite simply an excellent way to assess and improve your organisation's or client's performance and capabilities across all aspects of digital marketing. 

If you're not sure about the book, check out Avinashes blog 'Occam's Razor' for sound advice.

It is very much a book to help anyone involved in Internet marketing, either as a marketing manager with digital specialists in their team, or campaign manager get better results from their online campaigns. It will help small business decide on the most cost-effective tools and approaches, but will also help larger organisations select the best systems, resources and responsibilities. Consultants will probably get the most from it, to check their approach and pick up new tips.

Unless you’re a devotee of web analytics blogs or the Emetrics summits, you probably won’t recognise the name, but Avinash is well worth listening to since:
•    Worked for several years with major software vendor Intuit to drive online sales and implement web analytics within a large organisation
•    Recently appointed by Google as its evangelist for its Google Analytics web service.
•    Well connected with all the vendors
•    Most important, he’s a very practical guy and the book is packed with very specific guidelines and tips and he demythologises the many impenetrable metrics by presenting 'what is it?; 'why should you care' and 'what should you care about'.

Here are some of my favourite parts as a reminder of what I learned and felt I would like to pass on to others of my Internet marketing training course.

1. Types of marketing site

• E-commerce sites
• Support sites
• Blogs (this is the best treatment i have seen of measuring blog effectiveness)

A criticism is that it doesn't explicitly cover B2B relationship-building sites or brand-building sites in this section, although they are referred to elsewhere.

2. In-depth coverage of specialist topics

Important topics that are skimmed over in many books are covered in more depth here including:

• Internal / on-site search (p194/202)
• Multi-channel marketing (p225-236)
• Testing (a whole chapter including AB and multivariate, p237 to 262)
• Building management dashboards (p275 - 289)
• Web 2.0 analytics (p314 to 325)
  

3. Getting rid of the garbage

Avinash helps you cut through the clutter by sections on Identifying true KPIs and rejecting measures that initially seem useful as you can tell through section titles such as:

• Path Analysis: What is it good for? Absolutely nothing
• Conversion rate: an unworthy obsession

Kaushik (2007) suggests you apply the ‘So What Test’, p122 to all so called ‘key performance indicators’. Meaningless non-actionable KPIs that go in managers dashboards are a personal bugbear of mine, so it its good to see him show how you can potentially reducing reporting on potentially non actionable KPIs.

Examples given include % of returning visitors or top exit pages, but instead will focus on segmented KPIs such as conversion rate for top search keywords (which will identify phrases which are driving a lot of traffic, but are not converting).

He's also a big fan of bounce rates which I tend to rave about, so it's good to see their value reinforced.

4. Managing web analytics inside an organisation

- The 90:10 rule – Avinash says on p81 that 90% of your budget should go on resources to encourage the adoption of web analytics, to work with marketers and technicians to find the best reports for them.
– Avinash offers job descriptions for different web analytics role on p89
- Top web analytic blogs to stay up-to-date on p86

5. Tagging pages for browser-based analytics

An example of the very direct best practices, e.g. on p113 he summarises these aspects and then goes into more detail, clearly explaining each:
•    Tag all your pages
•    Place tags last [to avoid disrupting the users’ experience if there is a slowdown on the data collection server] and to mean that page views are only recorded when the whole page has been downloaded
•    Place tags inline [obviously not in frames, or surprisingly tables]
•    Identify each page with a unique definition [important for dynamic pages which may have different parameters]
•    Use cookies intelligently [i.e. first party, with relevant categorisation of page groupings or value event pages]
•    Consider link-coding issues [i.e. make sure you are counting links in Javascript wrappers, links with different query strings for tracking which go to the same page]
•    Be aware of redirects
•    Validate data capture accurate
•    Correctly encode rich Internet applications,  [for Flash or Ajax implementation additional coding is necessary to assess Javascript effectiveness

Avinash Kaushik’s Web analytics An Hour a Day (See 5* reviews on Amazon US).

Dave Chaffey