By Dan Kennedy • The press, politics, technology, culture and other passions

Repairing the Web’s broken meter

There is no bigger issue facing the news business today than how to make the Web pay. And there is no bigger obstacle to solving that problem than figuring out how many people are visiting, how long they’re sticking around and the like.

As I found out last year when I was reporting a story on young news consumers for CommonWealth Magazine, the internal numbers compiled by Web sites like and can be as much as three times higher than the numbers reported by Nielsen/NetRatings, the source of the leading apples-to-apples statistics used by advertisers.

The dilemma: Nielsen says it’s more accurate to ask people which sites they’ve visited than to look at a given site’s statistics, because an enormous percentage of those statistics are based on automated hits from search engines. News-business folks respond that Nielsen greatly undercounts the number of people who log on from work and from overseas.

Now, according to this article by David Cohn in the Columbia Journalism Review, help may be on the way. The Media Rating Council, a nonprofit group that helped standardize television and radio ratings nearly 50 years ago, has turned its attention to the Internet in an attempt to figure out all the metrics that should be of value to advertisers: how many people, how many different people, how many pages they’re calling up and how much time they’re spending with a given site.

These problems are far more difficult to solve than you might imagine. As Cohn points out, increasing numbers of privacy-minded people are setting their browsers to eliminate cookies every time they quit. The result: they’ll be counted as “unique users,” rather than return users, whenever they visit a particular Web site.

And if someone leaves a Web page on her screen while she goes on a 10-mile bike ride, how is that supposed to be measured?

The news business thrived on the lack of knowledge over whether any given subscriber would pore through the paper that day or toss it in the recycling bin unopened. Online, advertisers demand to know a lot more than that. So far, it’s proved impossible to answer their questions. Maybe that will change soon.

Discover more from Media Nation

Subscribe to get the latest posts to your email.


One more on Bailey


Saturday morning roundup


  1. Anonymous

    If someone is on a page for an hour, it would seem simple to exclude. When you review the time and the page views, that seems to make more sense. No?

  2. Neil

    “Time spent” does seem like a useless metric. With tabbed browsers and multiple monitors getting more common (and with Linux, four desktops on each of those multiple monitors, to boot), it’s easy to leave lots of pages up in tabs, while never actually looking at them beyond the initial glance. Once open in a tab it’s as easy to leave them there as it is to go to the trouble of closing them. I usually open links in new tabs where they accumulate until once a day, or every couple of days, I might go through and delete them.If it’s not possible to distinguish bots from real human “side door” users, it is at least possible to accurately track hits whose source was elsewhere in your own site. Though that number will be decreasing over time due to increased use of the side door, it’s at least a consistent metric that seems to me would be of interest to advertisers, and would come closest to reflecting the equivalent of a print reader. (Ie, it’s a person who looked at more than one page on your site.)

  3. Irv Arons

    Dan,Why can’t websites use something like I use — Sitemeter. It counts how many hits, how many pages and how long people look at my web Journal.Irv Arons

  4. Dan Kennedy

    Irv: I use SiteMeter, too. Presumably a lot of those “people” are not people — they’re robotic search engines. We may all be overestimating our readership by a factor of two or three.

  5. Aaron Read

    Well, let’s make something of an analogy here to put things in perspective. Arbitron has relied on the near-useless diary method of reporting radio and TV viewership for over forty years, and only JUST NOW is finally rolling out the Portable People Meter; which promises (and so far in tests has delivered) a quantum leap forward in accuracy of listener/viewer measurement.The web has only been around in mass media form for, what? 15 years at most? So we’ve got another 25 years of guessing before we really come up with a realistic system of metrics for the web, eh? 🙂

  6. Anonymous

    All of the statistics you mention are already available in the website’s site logs. The problem is the site owners won’t turn them over to outsiders for analysis, and even if they did, there’s no way to tell if they’ve been doctored. All major search engines identify themselves as such and are easy to exclude. Minor ones are so minor as to not be worth counting anyway. We do this all the time for our internal metrics. We don’t give those metrics to outsiders, but it’s not rocket science.Part of the problem, too, is that the Herald (and maybe the Globe) both automatically refresh the page after a minute or so, so there’s an automatic increase in page views even if you’re not looking at it. Or even if you are, the page refreshes out from under you. I’m not saying that’s good or bad, but it’s dang annoying to be reading something when it refreshes.

  7. Wes

    Spyders are constantly crawling the web and it’s hard to weed them out.When the markets are open I often have Marketwatch on one screen for hours as it is constantly updating.Ads? I see a rare few as do almost everyone I know – they’re filtered.

  8. Anonymous

    For all the reasons stated, chasing the Holy Grail of “hits” is kinda like running around in a circle.A much smarter metric, and one that newspapers can never match, is tracking ad click-throughs, as Google does with its AdSense system. Advertisers don’t care whether I read the article; they want my eyeballs on their ad and my attention (via their own web site) once I read it. That’s easy to measure reliably.

  9. Dan Kennedy

    Anon 9:10: I think what you’re saying is that the advertiser doesn’t have to pay unless the reader clicks on the ad. I think that’s wrong, and it’s one of the things that have given way too much power to the advertiser.From time immemorial, advertisers have agreed with the proposition that putting their message in front of readers was worth spending money. Now, all of a sudden, they claim it’s worthless unless the reader actually takes action, right now.I disagree.

  10. Anonymous

    Sitemeter works for me — but then I can discard the viewer who spends an hour on 1 page — then I’m only HUMAN! I know there was an interruption, the dog for a walk, a phone call.Pretty funny that advertisers who spent mega bucks on birdcage liners never figured out not everyone was reading.

  11. Anonymous

    I can’t think how many times we got too busy with commitments to read newsprint daily and recycled those ads.Now that we’re talking internet ads, they’re finicky? More discussion would be informative.

  12. Anonymous

    Hi Dan:I’m “anon 9:10” – I don’t sign in with my blogger ID because it is a shared account I use for my upcoming book. My point was that click-throughs are a great MEASUREMENT tool, not a way to bill for ads. I fully agree that would give too much power to advertisers and I can think of several ways to set a base rate with “bonuses” for clicks that would help offset the problem.But in the meantime we would have a truly accurate measure of who was interested in the ad, something we are forced to tacitly assume and gloss over when selling most other forms of advertising.-dan hamilton

Powered by WordPress & Theme by Anders Norén