pageviews

This page considers questions about a default metric for commercial sites.

It covers -

• introduction
• page views
• time at site
• ad views?
• activity

     introduction

As preceding pages have noted, there is disagreement about use of terminology and about measurement of what people are seeing (or perhaps not seeing, if for example they are using ad-washing software as part of their browser and email tools).

Uncertainty - sometimes disguised by loud protestations of expertise (usually on a 'black box' basis) - is a feature of much audience measurement, which as noted in a detailed profile elsewhere on this site is bedevilled by questions regarding -

• whether the right people or activities are being counted
• whether claims by the counter can be trusted, particularly in environments where it is difficult for observers to identify methdologies or gain independent benchmarks and where claimed "definitive" figures from one specialist may be half those of a competitor
• the interpretation of figures supplied by marketers or by third parties
• whether causation has been confused with correlation.

Businesses, advocacy bodies and government organisations that pay to appear on web sites are concerned with whether that investment is justified. Are they getting an appropriate return from the ad. Is anyone seeing the ad? Are the right people seeing the ad (illustrated by Betsy Bloomingdale's perhaps apocryphal quip to young Rupert Murdoch that "your readers are my shoplifters")? Are viewers being correctly influenced by the ad? Could the advertiser get better results by paying to appear on another site, one with for example greater numbers of visitors or more of the desired visitors?

A key feature of competition among metrics service providers is thus the search for authoritative online measurement mechanisms: authoritative because they have a clear empirical base that is not readily subverted or because they can be sold to ad buyers as accurate and cost effective.

     page views

As noted in the discussion of marketing, the simplest common metric l is returning unique visitors - essentially how many distinct users visit the site/service during a period of time. That time might be a day, a week or a month.

Uniqueness reflects attention to whether a visit is being undertaken by a human (ultimately the entities of interest to the advertiser, as humans make decisions and spend money) or by a search engine. It also reflects attention to whether there are recurrent visits, with 'return' traffic potentially indicating customer loyalty but potentially also indicating that one or more individuals (or their electronic surrogates) are boosting numbers by recurrently clinking on a link or doing the same search.

Unique visitation is difficult to meachure definitely because few people individually identify themselves every time they visit a page/site. Visitors may use different computers at one location or visit from several locations. Visitors may 'lend' logins or other access codes to friends and associates (eg provide a colleague or child with access to a 'dark web' subscription site); some codes may be appropriated from unsecured machines. Consumers who are privacy savvy - or merely fearful - may delete cookies from their machines.

     time at site

Some metrics specialists have responded by promoting measures based on 'time at site', a revamping of 1990s notions of web stickiness.

That stickiness reflected the earlier 'walled garden' approach promoted by private network operators such as AOL, in which a captive audience would never stray outside the particular network and thus would be more likely to encounter a message from the advertiser.

Nielsen for example announced that it would rank websites by time spend on the site, news greeted by enthusiasts as signalling "the death of the page view".

'Time spend' is however a problematic metric, one that may result in inappropriate comparisons. Critics have noted that simple time counts do not identify what is happening to the visitor and that a visitor may behave differently on various sites because those sites have different functions. Not all sites/pages are the same.

Moreover, the metric privileges sites that are slow (the visitor waits and waits) or have failed to heed design and accessibility principles and thus, for example, have confusing/inadequate navigation cues.

One critic implied, fairly or otherwise, that the metric is biased towards major corporate customers, some of whom are presumably delighted that their ranking has improved with a shift from page views to time spend. Google, for example, drops from third to fifth place in US rankings because although it is frequently visited its design is efficient and its function means that people look and move on. That is contrast to sites with an instant messaging (IM) function or a video-sharing function (eg YouTube), where users linger. One analyst thus comment that AOL would have the highest rank in the US as of mid 2007 because it scored 25 billion minutes, ahead of Yahoo!s 20 billion and Google, although in page numbers it would have ranked sixth.

     ad views?

comScore announced in 2007 that it was developing a new measure that would simply count how many ads are featured on each site.

The expectation was that such a measure would allow an existing or potential advertiser to infer the site's "scope and its potential to make money", with advertisers being able to readily compare sites. comScore would integrate the data with other measures such as ad impressions per page, ad-views-per-minute and indicators of where visitors go after encountering the page/ad.

     activity

Others have argued that the most meaningful metric is simply activity: what people do (assuming that the measurer can differentiate between actions by people and actions by databases).

A simple activity metric is to count by clicks, for example the number of people who click on a link to visit an online resource or to gain more information after encountering an ad. A more sophisticated metric is measurement of purchases that can be directly attributed to the advertisement (eg where a particular consumer clicks on an ad, enters a retail space and proceeds to buy a product or to request a report or a follow-up call by a customer representative).

Reliance on click-throughs is however contested. As early as 2001 financial services site MarketWatch.com grabbed 15 seconds of fame by abandoning the click-thru rate as a measure of effectiveness. It stopped providing click-thru data to advertisers and was reported as encouraging clients to look at other techniques for determining the effectiveness of their online promotions.

Retailers have used more specific measures, for example 'bounce rates' on visits to sites and 'abandonment rates' on facilities such as online shopping carts.

Although in the offline environment it is rare for a consumer to fill a shopping trolly but abandon it before going through the checkout that is common in visits to etail sites. Site operators have accordingly collected data on metrics such as the profile of abandoned versus purchased items, number of items per abandoned cart versus completed transactions and the ratio of abandoned carts to completed purchases. That data may allow inferences on matters such as design and accessibility. The information typically is not collected by third parties for provision to advertisers or other users: it is essentially a management tool for the site operator.



::