Time spent on newspaper Web sites

[Brian Cubbison]

I'm curious about how time spent on newspaper Web sites is measured. If the same kind of numbers are compared each time and they show a drop, that's a legitimate concern. But I wonder about how the total is measured.

You usually see it as a small-seeming total, such as 20 minutes a month, which would come out to less than a minute a day. But I wonder about two things:

1. It's unlikely that very many people will visit the site at 3 a.m., and there's really about a 16-hour window of prime Web site visiting: 8 a.m. to midnight, let's say. Perhaps the real prime time is from 9 a.m. to 5 p.m. weekdays when most people are at work. The monthly total might actually represent a larger amount on any given day than we think. It's also skewed by traffic that comes in through Google, finds what they were looking for, and leaves. More traffic = worse numbers seems wrong.

2. It seems like a very indirect way to measure effectiveness for advertisers. Time spent watching a video, for instance, is time not spent seeing an advertiser's ad, in most cases, or time not spent clicking through to an advertiser's site. Google measures time spent on its site in the fractions of seconds, yet it makes billions of dollars on its little text ads that don't blink, flash or even have images. Why those billions of dollars were lying around waiting for Google to scarf them up, or whether the floral shops and hardware stores are actually doing billions of dollars of business through those ads is for another time.

If only there was a magic way for advertisers to know: Buy an ad here, and here's how much your sales will increase.

[Brad King]

Brian:

The thing about advertising -- time on site matters very little to advertisers. What matters is audience (uniques) and lead generation (how many people are clicking). I've spent a fair amount of time meeting with ad groups in the last 5 years and they all ask the same questions -- the questions that site producers ask: how many people come, how often to the come, where are they going and what are they doing.

With syndication, readers, lifestreams and the like -- sticky is the wrong question to ask. You don't -- or shouldn't -- care how long people stay. You can if they are coming for the right reasons, to the right places and if your ads delivered are relevant.

You also don't want spikes -- hour to hour or day by day. It makes it very hard on your ad serving software.

What you do want -- particularly for local papers -- is the ability to create an AdSense type environment for your local market, allowing businesses to bid and buy keywords and other search terms.

[Zac Echola]

I've heard talk about selling ads based on "drive time" but I think Brad's right when he says advertisers don't care about that as much as conversions and impressions.

As a content producer, I'm concerned with time on site not as a stand-alone metric, but how it correlates to other metrics like page views, bounce rate, uniques and others.

If you have lots of uniques, but are short on time on site you might be well optimized for search engines but irrelevant to readers once they hit your site. A high bounce rate might occur. (Pay walls will do this)

If you see growing uniques but also have low page views and low time on site, you might want to start finding ways to feed in more relevant content to article pages.

If you have few uniques and lots of time on site people are reading for content, but your audience may be small. How can you grow that community while stemming an increase in bounce rate? Or maybe your site is just hard to navigate, but you have a product people want.

Lots to think about, but I use it to help get my bearings on other data sets.