The Wall Street Journal has provided a good summary of how different web measurement companies produce conflicting figures on website usage and the difficulties this can give to both advertisers and publishers. As it says, the science of tracking Internet usage remains far from perfect due to the different tracking methods and measurement criteria being used.

Advertisers rely on data from companies like comScore and Nielsen Online to determine where to place their adverts, but it has long been known that comparing data between both companies is fraught with problems. In the same way that web analytics software will measure site usage data in different ways, it is better to look at comparative data within the same system, rather than to compare figures between two different methods of data collection.

Both of these leading web measurement companies use panels of online users to collect data and then extrapolate out to the likely universe - in much the same way that TV and radio audience figures are measured - so that the end results are statistical estimates. This means that there can be wide variations when the visitor numbers are in millions, plus both of these US companies lack the capacity to measure total international audiences.

Both comScore and Nielsen Online are attempting to address those shortcomings by looking for ways to increase the size and depth of their panels, investing in technology and expanding overseas. Nielsen is also trying to combine its web research with usage data from other media collected by its parent company, such as mobile-phone and television measures.

However, the current data remains the best there is for advertisers at this stage and therefore a good starting point for media planning. Ultimately their own response data on the ROI of any advertising will provide more meaningful data on how to develop an ongoing campaign.