December 2006


29 Dec 2006 02:12 pm

DSCF3968 smallAndy Beal of the Marketing Pilgrim has tagged me for a new blog tagging meme that has just started. Andy was sweet enough to play when I tagged him for the personally tag game so it is only fair that we play along for his game!

The 2007 Predictions tag game was started by Mashable. I think it is more dangerous than the personal blog tag game because in this one you actually have to come up with something intelligent (and one's “performance” can be measured at the end of time!).

It took me six minutes to write this and I am sure it will have turned out to be a career limiting move! My 2007 predictions for our beloved Web Analytics universe:

  1. 2007 will be a banner year for Web Analytics education (books, blogs, seminars, conferences) and Web Analyst salary improvements (of course I have to say that!).

  2. We will see increasing sophistication of and availability of “free” web analytics tools in the market (welcome Microsoft Gatineau). This will be good for current vendors: more people to sell to and move up-market. This will be bad for current vendors: will have to create more obvious differentiators, than just number of reports.

  3. Pay Per Click (sometimes called SEM) analytics will continue to be important but cede mind-share and control to Search Engine Optimization analytics (which is in a rather sorry state in all current web analytics tools).

  4. WebTrends MarketingLab and Omniture’s Discover will release updated versions of their “data warehouses” that will be closer to (and hence more integrateable) standard business intelligence data warehouses (allowing us to do deeper and richer end-to-end analytics with other data in our companies).
    [Important: I have no knowledge of either company’s product road-map.]

  5. More Analysts and Web Decision Makers will realize the futility of torturing clickstream data to get into the heads of their visitors and embrace qualitative measurement options (surveys, testing, usability and so on). No totally radical changes but steady progress towards the Trinity mindset.


    One Bonus:

  6. More website business owners will realize there is this thing called Social Networks and panic when they can’t be measured. Rest assured, targeted solutions will follow in 2007.

In turn I’ll tag our esteemed world leader Mr. Peterson to offer up his top secret list of 2007 web analytics predictions (I am positive it will be super!).

Do you agree with the list above? Is it achievable? Is it too conservative? What trend does it ignore? Please share your own predictions via the comments field.

[Like this post? For more posts like this please click here.]

27 Dec 2006 12:32 am

DSC02241 small1Occasionally we revisit a well established / standard metrics or report and take a fresh look at it. The first metric we look at was Visitors (Visits, Unique Visitors). In the second one of the series we will look at a very standard metric Exit Rate and one specific report, Top Exit Pages.

All web analytics tools contain the Top Exit Pages report. In terms of mechanics the web analytics tool take the last page viewed in each visitor session and the report simply shows that pages that appear most frequently (by a raw count) as the last one in visitor sessions.

On paper this metric (remember it is different from bounce rate though some practitioners are confused about bounce rate for a page) is supposed to show “leakage” of your website: where do people exit from once they start their session. It should illustrate pages that you should “fix” to prevent “leakage” and get customers to buy more or sign up more or get deeper into the site or generally do what a website owner would have customers do.

3

Is it worth it?

    For the most part you should not care about this metric, for most websites it tends to be a hyped up metric that tells you little while, on paper, claiming to tell you a lot. This is especially true if the report is at an aggregated level (as in the above screen-shot which shows top exit pages for All Visitors of the website). If you pause for a moment and imagine a typical customer experience on a website you’ll see that this is not telling you much, if anything at all.

    Yet it does seems obvious that if you knew where people were exiting from on your website that you could simply fix that “page” and all would be kosher. In reality visitors come to your website for a whole bunch of purposes and it is often ok that your top exit page on the website is the page that shows your best selling product (it will be that page) because a big chunk of visitors want to read about the product and buy it in retail. Just as an example.

    Another factor going against making this a valuable report is that the conversion rate for most website (e-commerce or otherwise) is around 2%. That means approximately 98% of the traffic will exit at places you don’t want them to exit (examples of good places to exit: checkout page or lead submission page or a support faq page). When such a huge amount of traffic is exiting (leaking) from your website, and most likely from your most viewed pages, it is extremely difficult from the raw exit rates on those pages to parse out success or failure.

    So if 50% of people who see your product_details page exit, what percent of that is “good”, those that read reviews and will buy some place else like a retail store, and what percent is “bad”, those who came to buy and you upset them with tiny six size font on that page? How do you know from simply the exit rate number?

    It is often, not always (see exceptions below), a futile exercise and you are better off at using other mechanisms (for example surveys or Usability) to figure out why people exit your site at different locations.

Is it ever worth it? 

    As always there are exceptions to the rule, atleast two in this case…..

    1) Segmenting this report for various traffic streams or campaigns or customer types can redeem it a little bit and potentially highlight trends that might shed some insights. If you get micro nuanced enough in your clickstream to focus on a small group of visitors whose Primary Purpose on the website would be obvious through clickstream data then this report can be of some use. (And only because in such micro nuanced detail it is knowing intent that makes the difference.)

    2) The only solid exception to the rule are structured experiences that are of a “closed nature”. For example the Cart and Checkout experience. You add to cart, you click start checkout, you fill out your address and credit cart, you hit submit and see the thank you page. In this structured experience it can be insightful to measure which page is the top exit page and why might that page causing “leakage” and how to fix it (multivariate testing to the rescue!).

Conclusion:

    As a general matter of habit we should consider avoiding the aggregated top exit pages reports so easily available in our standard web analytics tools. More time to spend on other stuff that might be more valuable.

    If you really do need to understand why customers exit from your websites then consider deploying other listening mechanisms that are available such as Surveys or other Qualitative Analysis options.

Your turn.

    What do you think? Do you agree? Disagree? What’s missing from the above analysis? Have you found the top exit pages report to be valuable? Please share your critique and feedback via comments.

PS: If you have suggestions for other standard metrics or reports you would like addressed then please email me (blog at kaushik dot net) or leave a comment.

[Like this post? For more posts like this please click here.]

Next Page »