Click-Through As The Wrong Metric

This unfolding year will be the first year of real web accountability. And for those of you still looking for sufficient justification to advertise on the web, 1999 will become the Year of Web Advertising ROI.

So, what is the right metric for ad results? It depends. Even in this most measurable of media, getting a fix on advertising results can be troublesome.

For one thing, a click-through rate is not necessarily a good indicator of advertising ROI, and certainly should not be used as a basis for pricing ad buys. Here’s the truth: Click-through is nothing more than a convenient, easy-to-get-a-handle-on measure of an ad’s ability to create an impulse response.

Click-through rates say nothing about results. It’s like counting the number of T-shirts given out on a street corner. You don’t even know if you have reached your intended audience.

Another clear reason that click-through is the wrong metric: The rate is dropping steadily. A couple of years ago, average click rates were in the 2 percent range. NetRatings measurement of ad click-throughs shows that from May to November 1998, click rates dropped from 1.3 percent to just above 0.5 percent.

Does that mean that web advertising isn’t working? Hardly. It just means that, on average, there is a lower impulse reaction to click on ad banners than there used to be.

A more meaningful measure to look at is advertising’s effectiveness in building site traffic. It could be that while people are not responding impulsively to banners, that the banners nonetheless have an impact on brand perception.

So let’s get specific.

To check the impact of sustained advertising on site traffic, NetRatings analyzed 20 advertising campaigns and corresponding sites for the months of June, July, August and September 1998 to see if there was a correlation between regular advertising and growth in unique audience size.

In 15 out of the 20 cases, we saw a consistent pattern of audience growth during months when advertising campaigns ran. The remaining five cases either showed an inconsistent pattern or a decline in audience.

It’s critical to note that while this pattern suggests a clear link between advertising and audience growth, direct causality is unclear. We didn’t perform a statistical analysis that controlled for other significant factors that could impact audience growth, like off-line advertising and the impact of search results and site links. Also, we examined sites and campaigns that reached at least 5 percent of the total web population. So smaller sites and campaigns were excluded.

Among these 20 cases, there were 19 advertisers. Eight campaigns ran for two months, five for three months, and seven for all four months analyzed.

Of those eight sites that ran campaigns for two months, seven showed an increase in audience size during the campaign months as well as a corresponding decline in audience during non-campaign months. The remaining case showed no clear pattern of correspondence between impressions and audience growth month to month.

Of the five sites that ran campaigns for three months, three showed a relatively consistent pattern of growth in audience over time ranging from 10.5 percent to 23.5 percent increases from June to September. One case showed changes in audience that were inconsistent with the presence of or changes in impressions; and another case showed a small decline in audience from June to September, but had a relatively small ad campaign.

Of seven sites that ran campaigns for all four months, five revealed a generally consistent upward trend in audience across the four campaign months ranging from 8.2 percent to a whopping 215.5 percent. One case showed both increases and declines in audience month to month, which were apparently unrelated to the continuous, although declining, advertising for that site. The remaining case, with an average monthly percent reach of 41 percent, showed a small decline in audience size during the campaign months.

So what’s the bottom line?

In our view, the study shows that there is strong evidence that web advertising influences site traffic, which — in the face of declining click rates — may seem counter-intuitive.

More importantly, it clearly sends a signal: If 1999 is the Year of Web Advertising ROI, care must be taken to choose the right metric.

Related reading