• by Jim Meskauskas , May 22, 2002

Often the forgotten metric, session lengths could become as important to media buyers as impressions when selecting websites.
Since the early days of monitoring traffic on the web, one of the metrics that metering services have provided for sites and agencies is session length.

Even the earliest PC Meter reports (the precursor to Media Metrix) had data points that indicated how much time a user spent, on average, with a particular site.

At first, no one was sure what to do with this information. To many, it was at best an interesting factoid. Then, for a brief moment, sites started including this tidbit of data in their sales pitches. Site after site would come in and present, almost always including mention of the amount of time a user spent with their site. This was done for a few reasons. Most notably it was to point out how loyal and dedicated an audience a particular website had. The more time the average user was on the site, the more committed a user they were. Some websites sales organizations might have thought that average-session-length data might suggest an audience’s attention level. The greater the average time spent on the site, the more attention a user was paying to the content and, thus, an advertiser’s message. And perhaps others thought it would persuade media buyers that exposing the site’s audience to their client’s message would be a good opportunity.

All of these have certain degrees of validity, to be sure. Thinking back on it, gaming and game content sites were most fond of pointing out how much time the average user spent with their content. TEN, Heat.net, HappyPuppy—they all wanted to tell us how much more time users spent with them than with other sites. This was initially done for lack of a better point of product differentiation. When you’ve got a thousand websites sites out there all fighting over the same nickel, you’ve got to have something unique.

But when all is said and done, no one has yet figured out a way to do anything meaningful with session-length data, so it has gone by the wayside. It has become the parsley served with your restaurant burger. And so the session-length data point has languished as a forgotten honorable mention.

The issue of session-length data continues to be raised, however, usually by clients new to the Internet. And from time to time, I come across a site that will still mention it.

So why does it keep coming up, and what might it mean? While some have suggested that session-length is a useless metric, I do not agree.What it means given the varieties and contexts of content is unclear, but to dismiss it out of hand because we are unsure of its implication would be unwise.

There is no doubt that the amount of time spent with a medium has been an important factor in media planning over the years. Those with experience in traditional media may remember the "Media Quintile." This is a way of looking at a particular target or demographic's media usage based on time spent. A quintile is an even segmentation of a universe into five parts, each representing a "bucket" of media usage: heavy, medium-heavy, medium, medium-light, and light. These designations are based on time spent with a medium. Now, a subset of how much time is spent with a particular medium is how much time is spent with a specific vehicle within that medium. This is not easy to obtain with most traditional media (for broadcast, we assume program length to be user's possible vehicle session length, and then Nielsen breaks that into five-minute parcels), but if planners could, they would take session length into account when planning media. With the Internet, we can break session length down not only by vehicle but also page. Whether or not this is meaningful to you depends on A) what the advertiser is trying to accomplish, B) what kind of media property your advertising runs on (dictated in no small part by the objective), and C) how the inventory is packaged and sold. Session length might mean nothing to a buy on Yahoo!, which sells massive amounts of impressions and has more of a "drive-by" audience. But this might be very important for a site selling fixed placements (sponsorships, Surround Sessions, audience-based inventory) like Salon.com, NYTimes.com, or some other vehicle used more for branding and awareness.

Though it isn't clear that session length should have an effect on the pricing of inventory, it shouldn't be assumed not to have an impact on what a client's advertising can accomplish, especially if that advertising can be shaped to take advantage of the time spent with the medium and the vehicle.

Opinions on the subject vary, but when pressed, everyone has a certain visceral belief that the amount of time one spends with a vehicle in a particular medium has an impact on the individual who is being engaged. Exposure equals influence—this is why parents don’t want their children hanging out with the "wrong crowd," walls in sanitariums are painted certain colors, and practice makes perfect. In traditional media we are forced to use certain surrogates to determine what and how our advertising might be working. We use demographics as surrogates for psychographics. We use lifts in awareness or purchase intent as surrogates for projected sales. We use vehicle exposure (to what extent a certain magazine or program might have a chance of being "seen" by an audience, for example) as a surrogate for advertising exposure.

The most complete picture of what our advertising might be doing for us could be provided by knowing the amount of time spent with our ads, but that’s something that is impossible to determine in a meaningful way with traditional media. However, session length provides online advertising with a metric that has no equivalent in traditional media, and thus it can, for the time being, serve as a surrogate for the time spent with an advetisement.

The difference between broadcast and online is that in broadcast, vehicle length doesn't have any correlate of message persistence. Basically, I can't have a "durable" message with the programming. The closest thing would be those ghostly network IDs and logos that float in the bottom right-hand corner of the TV screen during programming. Online, the advertising can potentially endure throughout the course of a visitor session.

This isn't to say that is how advertising works online, but for things like sponsorships, which provide the equivalent of signage at a sports stadium, the session length could prove a valuable metric for advertisers.

If an online publisher wants to experiment with selling inventory based on "circulation" or unique visitors (that is the audience-based media currency I've written about, on and off, for the past year and a half) rather than on raw impressions, the session-length metric might be usable as a way to differentiate the value of the impressions served rather than their frequency.

This may be a way of evaluating advertising on its own terms, but I'm not sure that advertising can be, in an evaluative sense, truly separate from the content within which it resides. I guess direct response is something that, to a certain degree, serves as its own medium, where advertising is evaluated solely on its merits. In light of the tenuous connections we are still willing to draw between the advertising event and the brand engagement, reliance on the correlative aspects of advertising, such as the content in which we encounter it, will continue to be important to how advertising gets evaluated.

I haven't worked everything out yet about what a session-length valuation might look like, but it's a problem I'm interested in solving. And when I do, y'all will be the first to know.