Long live cookies!

Second quarter circulation figures released last week by the electronic arm of the Audit Bureau of Circulation (ABC-e) were more controversial but no more flawed than usual - on that front it hit rock bottom a long time ago.

ABC-e faces a familiar problem: the chasm between what is technically feasible and politically required.

It is technically feasible to measure page impressions, or the number of pages requested by readers and delivered to their desktop. The means of doing so approach being foolproof, with established methods of measurement and a full audit trail to back it up.

In isolation, page impressions tell you nothing about readership.

Phillip de Wet, Journalist, ITWeb

Page impressions, averaged over a three-month period, stands alone as a means to determine how popular any Web site or group of sites is.

But the political requirement is to count unique users, the number of actual readers.

In isolation, page impressions tell you nothing about readership. One reader who requests 50 000 pages every month and 50 000 readers each requesting a single page every month give exactly the same result.

This makes it fairly meaningless to advertising agencies that require a certain reach or want to hit the same audience repeatedly, as is their wont. Their calculations require a readership figure and it is understandable that its absence would be a disincentive to spend.

While ABC-e, or rather its predecessor Abis, reported only page impressions, it saw membership stagnate and then decline in tandem with the death of the Internet bubble. Some surviving members did not see a return on their audit fees while the figures they got were meaningless.

Of course, ABC-e bowed to the pressure and implemented "best effort" readership counts, which opened an entirely different can of worms.

Despite what anyone may tell you, there is no way of accurately measuring a Web site`s readership. Even a password-protected site, such as an online banking facility, cannot be sure that a user has not shared his password with his wife.

When it comes to a freely accessible site, the nightmare becomes exponentially worse. When a reader requests a page, the only identifier his computer supplies is an IP or Internet protocol address. But a thousand employees using the same proxy server could be represented by a single IP address, while a dial-up user is assigned a new IP address every time he connects and could look like 10 (or 30, or 100) users.

To get around this, a site could try to drop a cookie on the reader`s machine; a second identifier that is associated with a specific PC. Beside the minor problem of multiple users on a single PC, you are also faced with users who refuse cookies to protect their anonymity, proxy servers that accept only a single cookie, firewalls that destroy cookies on sight, or any combination of these that thousands of systems administrators can come up with.

There are other technical and audit problems, and no solutions to any of them.

The result is exactly the dodgy circulation figures the ABC-e is meant to prevent. Different publishers use different metrics, all of the metrics produce what is essentially educated guesses and wild variations, and obvious discrepancies do nothing to reassure advertisers.

There is nothing anyone can do about the fundamental technical problems with online audience measurement, although the effects could be mitigated. For one, the ABC-e could enforce a single standard among its members - if it had not been de-fanged by fear of losing more members and becoming entirely irrelevant.

ABC-e could also educate the advertising community about the problems inherent in unique user measurement and how it should be interpreted - if it was not strapped for cash because of the precarious financial position of many Internet companies and a lack of members.

It is a chicken-and-egg situation in more than one respect. Without advertisers demanding membership, the ABC-e will remain a fairly exclusive club. But advertisers won`t require membership when it does not result in usable figures. Without increased advertising revenues, online publishers and the ABC-e will be short of disposable income. Without a fair amount of money being spent, advertisers will not be reassured about their investment in online advertising.

There are no easy answers, as similar bodies in Europe and America have found. But the next time you see ABC-e figures, at least spare a thought for the poor publishers, advertisers and auditors behind it and don`t turn off cookies on your computer.