Time for a new hit

When banner ads started cropping up on the Internet in the mid-'90s, the term "hit" was all the rage. A "hit" (as in, a hit to a server) was the reigning measurement of a site's popularity. Hits attempted to show how many times a page was viewed by a user. As the online measurement industry evolved and site designs grew more complex, "hits" fell out of favor because they included all server calls, even those where users ditched a page before it fully loaded. Counts were therefore inflated.

It also became apparent that some independence was necessary to have believable performance numbers. Over time, engagement with a site evolved to measuring "page views," which were typically collected from independent, third-party sources. When combined with the "unique visitor" count, page views (or PVs) provided a metric of who saw the site and how much content or how many ads they consumed.

Enter the first measurement dilemma: There was no industry standard for how a page view was measured. Different firms used different methodologies, which differed still from how sites' internal logs were measuring traffic. No one was right or wrong, just different. Do you count PVs from outside the country? From spiders and bots? Do you count when the page is requested, or when the page is fully loaded? Although the industry has radically improved standardization in recent years, many questions remain about the efficacy of the page view as a method for tracking engagement or impressions.

Before I joined Yahoo!, I was President and CEO of comScore Media Metrix, where we spent years grappling with the challenges of measuring a dynamic web environment. It's 10,000 times more complicated than measuring television, and it's only gotten more complicated in the last year or so, with the adoption of cool technologies like Ajax and Flash and the growth in video. They've changed the way Yahoo! and other companies are designing products for consumers, but they will also translate into page view declines.

When you view our new Yahoo! Maps, for example, Ajax allows you to drag the maps around and zoom in and out without having to wait for the page to reload. But for all that convenience, the new experience translates into just one page view. Our users tell us they much prefer the new maps to those of old, so suddenly counting page views seems far less important.

Yahoo! has long reigned in page views. But we knew we'd eventually cede this honor as we incorporated more consumer-friendly technologies into our products. And that’s OK with us, if it means our services will be better built for how people want to use the Internet today vs. in 1998.

Page view counting has been a key measure for a decade but just because it was once the obvious solution, doesn’t mean it’s the best one now. A couple of reasons why:

• PVs aren’t a good reflection of web activity in 2006 and beyond. It’s a broadband world and page views are irrelevant to some of the most frequently used Internet services like instant messenger, VoIP, or video, in addition to technologies such as Flash and Ajax. More page views might actually reward sites for poor site design in light of these new technologies.
• PVs have never been consistently measured by third parties or by sites themselves. Everyone has a different definition of when and how a page is counted.
• PVs don’t represent ad inventory. In the early days of the Internet, page views were used to represent available ad impressions, but the reality is that page views and ad impressions are actually counted in different ways and don’t correlate. PVs also have little to do with available inventory with the different types of ad units available today using text, audio, video, etc.

The bottom line is that the page view has outgrown its usefulness. The industry needs to embrace change and develop new metrics that measure this new world more accurately. We all need to help to wean the industry off the crutch of familiar metrics in favor of more accurate and representative ones. We all need to be smart about these new metrics — the measurement companies, major publishers, and advertisers.

I remember a quote by Albert Einstein, who said something along the lines of: "Everything should be made as simple as possible, but no simpler." We can’t oversimplify complex web measurement. We need more innovative ideas and execution to measure new technologies and incorporate user behavior into measurement standards.

At Yahoo!, we're working closely with the measurement firms and our internal analytics teams to ensure we're creating the most accurate representation of user activity on our sites. Nobody has all the answers, so we need everyone pushing.

Peter Daboll
Chief of Insights