The Pre-Commerce Blog

Web Analytics Confidence Game

Why would a major web analytics vendor want to undermine the confidence of the web site analytics market? Perhaps because they see it as a way to differentiate themselves in an increasingly crowded and competitive market. In a marketing effort that would make Frank Luntz proud, Webtrends has been orchestrating a PR campaign suggesting that cookies aren't reliable enough to use for high-end analytics. Or more directly, as summed up by a line in their own press release:

Report distortion from cookie rejection is much greater if the web analytics solution heavily relies on cookies for purchase histories or campaign responses, or, as the solution's only method to sessionize visits.

They may be right. I certainly don't know and it's very clear that all web analytics vendors have some work to do, because of the technology of the internet, to make their reports more accurate and stable. It's just the methodology by which Webtrends is propogating this 'don't trust cookies' campaign that I find interesting. A few hours ago I watched a White House Spokesman on Chris Mathews incessantly repeat the phrases 'up or down vote' and 'delay tactics' incessantly in a clear effort to shape the debate without addressing the actual issues. I'm a little surprised to see these advanced (and effective) yet somehow discomforting tactics being used to position analytics software.

Jupiter made the first splash about cookie deletion with their report a few months ago. Then there was some claims and counter-claims about the accuracy of the study, which seamed to end with support for the initial report.

Around that time, Webtrends was on a national roadshow - I attended a New York date - where they were only too happy to talk about the study and casually but in some detail go through all the issues and risks with cookies. Since then I've noticed several articles linking them to bad mouthing the cookie, until today an eMarketer article came out citing them as the source in this study of 'marketer confidence'.

I'm not sure why it matters if marketers are confident or not. It only matters if reports are accurate or not. Todays article reminds me of the MSNBC daytime polls - 'Did Micheal Jackson Take Advantage of Young Boys? - email us your thoughts' - does it matter what anyone thinks, especially while there is a jury of 12 people listening to the actual facts and arguments and making that very decision?

Seeing the report, and the blogosphere echo of it, made me realize that Webtrends had set their sites on a long term campaign to discredit cookie-based tracking. As I said, perhaps they're right and doing a service for all of us. I can say that they’ve inspired me to send an email query to our primary analytics vendor (Omniture) to get their take on the issue (I'll report here when I get an answer back).

If Webtrends is either a) right or b) able to convince everyone that they’re right, I assume that all the top tier firms would quickly support new tracking methods that don’t rely on 3rd party cookies so heavily. It’ll be interesting to watch, and to see if the ‘don’t trust cookies’ story gives Webtrends an advantage in the analytics wars.

Search Results Go To The Dogs

An article on Metasearch has some interesting stats on the differences between page one results on the top search engines.

...a study conducted by Dogpile in collaboration with researchers from the University of Pittsburgh and Pennsylvania State University, "Missing Pieces: A Study of First Page Web Search Engine Results Overlap." compiled a random sampling of 10,316 keyword searches across Google, Yahoo!, and Ask Jeeves, returning a total of 336,232 unique search results on page one.

Only 3% of returned results (totaling 10,712 links) achieved first-page rankings across all three search engines, 12% of first-page results (39,959 links) were shared by two of the three engines, and 85% of results (285,561 links) appeared on just one of the engines' first pages.

(you can) try this yourself. Go to missingpieces.dogpile.com to test the Missing Pieces tool, and see how results overlap.

This is interesting both because it shows how hard it is to rank well across the top search engines at the same time, and it highlights the fact that none of these search engines are really very good at finding 'the best' results. They just each have their own algorithmic biases and we're all too lazy to turn to page two or (usually) to go try another search engine. I think 24 months from now the major search engines won't look anything like they do today (in terms of results or user-interfaces) and we'll look back at this first 'heyday of search' and wonder how we all got along with such terrible results.

Paid Search In The News

Two good articles caught my eye today. The first is about the folks who work for Google writing Adwords ads (from the LA Times):

She crafts text ads to intrigue Web surfers because advertisers don't pay Google unless the ads are clicked on. She has only three short lines — of 25, 35 and 35 characters each — and a link to make her pitch.

The other is about a topic near and dear to my heart - why paid search gets 95% of the budget when it only delivers 15-20% of the traffic. In this case, wise man Gord Hotchkiss calls it the 70/30 rule:

I asked the audience which section of the page they normally look at first. Almost every hand in the audience went up when I got to the top organic results. This was no great surprise. From our research into search user behavior, I was pretty sure this would be the case. Then I asked who in the audience dedicated at least 30% of their search marketing budget to organic optimization. A very few hands went up, probably less than 3% of the audience.

Posted by Craig Danuloff at 7:02 PM