Newspapers Auto-Refresh Pages and Hurt Advertisers

Attention businesses who advertise in major online newspapers across the country. I have some good news and some bad news for you:

The good news: Online advertising is very measurable.

The bad news: By my measure, a huge portion of the ads you’re paying for are never seen.

Here is how I came to this conclusion:

This is a ranking of the top-30 online newspaper properties in the United States according to Nielsen based on unique visitors:

Top Newspaper Websites by Visitors (Dec 2008)

As you’d probably expect, the top-5 are:

1. The New York Times	18,187,000
2. USA Today	11,420,000
3. The Washington Post	9,470,000
4. The Los Angeles Times	7,963,000
5. The Wall Street Journal	7,235,000

And here is the same list sorted by a different metric, Time on Site, which is a measure of how many minutes the average visitor to the site spends when visiting:

Time on Site of Top-30 Newspaper Sites (Dec 2008)

The NY Times leads this list as well, but the other four sites manage to beat out some of the top visited sites when measured by time on site.

1. The New York Times	0:33:03
2. Star Tribune	0:32:20
3. The Houston Chronicle 	0:19:20
4. Detroit Free Press	0:17:33
5. The Boston Globe	0:16:55

Glancing through the data, here are a few things Neilsen is essentially claiming:

1. The average visitor to the Minneapolis StarTribune spends 22 minutes longer on site than the average visitor to the Wall Street Journal.

2. The average visitor to the Detroit Free Press spends 2.5 minutes longer on site than the average visitor to the Washington Post.

3. The average visitor to the LA Times spends only half as much time as the average visitor to the Washington Post.

Here’s the deal: I don’t believe any of those statements are actually true. I think they’re statistically accurate based on how Nielsen measures website traffic but are heavily gamed by America’s top newspapers.

Auto-Refresh Hell

Many (actually, most) major newspaper websites have adopted a practice called auto-refresh where they automatically refresh/reload their website’s pages on regular intervals without their visitor’s permission. Among the top-30 newspaper websites, forced auto-refresh frequencies vary from once every 30 minutes to as often as every five minutes. This provides no added value to visitors (in fact, it makes the sites less usable in many cases) but it achieves something important for publishers: inflation of ad impressions.

Here is a breakdown of forced auto-refresh frequency: The further to the left a site is, the more often they force unnecessary refreshes upon their audiences (as high as every 5 minutes). Sites to the far right do not participate in this despicable tactic:

Top-30 Newspapers Auto Refresh Time (Minutes) March 2009

Three sites, The Washington Post, Washington Times, and Dallas Morning News, are currently reloading their site’s pages every 5 minutes. Three more sites are force refreshing every 10 minutes: StarTribune.com, LA Times, and NY Daily News.

Destroying Online Advertising One Auto-Refresh at a Time

I often hear that online advertising doesn’t work from companies who’ve experimented with it through newspaper websites. It turns out that there is a very rational reason for why it doesn’t work: People don’t see a significant portion of the ad impressions.

For example, in the case of the Minneapolis StarTribune, it’s quite common for people to keep the StarTribune open in one of their browser’s tabs overnight so they can check the weather first thing in the morning. If they do that, the StarTribune will continue to refresh the user’s browser every 10 minutes all night long, generating 48 page views over 8 hours of “Time on Site” while the user caught some zzzzz’s. Advertisers end up paying for ads that no one viewed (48 x all of those ads on the page).

By charging for ads no one sees, major media sites are diluting the value of online ads and destroying the perceived viability of online advertising. Clearly, the value of online ads is considerably higher when they’re viewed by humans.

Breaking it Down

As I mentioned above, not all newspaper sites are serving their advertiser’s ads when no one is looking. Forcing refreshes all day and night appears to be a popular way to game the stats, but destroys advertiser’s returns. Below is a ranking of time on site with the forced refresh crowd on the left and the non-refresh (people click to refresh the page themselves if they’d like to) sites on the right:

Time on Site of Top-30 Newspaper Sites (Dec 2008)

What’s notable here is that no newspaper in the country is able to generate 30+, 25+, or even 20+ minute long time on site metrics without gaming the stats with auto-refresh scripts. The Houston Chronicle leads the pack among non-auto refreshed time on site stats by this measure. Outside of Chron.com, no website breaks the 10 minute time on site barrier without gaming their stats, which makes me wonder how Chron.com did it (their auto-refresh status was checked in March 2009 while the stats from Nielsen are from December 2008).

Frankly, this makes me wonder if half the ads the StarTribune.com serves are ever seen. Heck, it may be worse than that.

Combining the above stats generates this perspective:

Top-30 Newspapers by Time on Site (with Auto-Refresh Rates)

Which shows, again, that only one of the top-10 newspaper websites in the country appears to be doing so without forcing automatic 24×7 refreshes upon their visitors.

Who’s the Most Pathetic?

If I has to label one site the most pathetic among the top-30 most visited newspaper websites in America, it would be the Washington Times. It ranks 29th out of 30 for time on site, ahead of only the International Herald Tribune. But it beats the IHT on the pathetic scale due to it’s ridiculously high 5 minute refresh interval. At that rate, one would expect them to rate higher for time on site. They must churn through many visitors in under 5 minutes who never make it to the first forced refresh of their browsers. Admittedly, the Washington Times is not as pathetic as the Drudge Report’s pathetically high 3-minute refresh rate.

What Should Advertisers Do?

Websites of large newspapers have huge audiences so it’s hard to ignore them as a potential ad publishers. However, before trusting them with your money, ask a few rational questions:

1. Are you willing to pay for ad impressions automatically served while people are sleeping with a browser open?

2. Are you willing to pay for ad impressions automatically served while people are in meetings or at lunch during their work day?

3. Will you pay for ad impressions automatically served on a user’s home computer while they’re at work?

4. Will you pay for ad impressions served in open browser tabs other than the one the user is viewing?

Do your best to figure out what percentage of your ads are actually being seen by humans, and ask for fair price adjustments. There is definitely a fair price for the ads since humans do see some percentage of them. Then measure the performance of your ads against other online marketing opportunities and re-negotiate your ad buy as needed.

Or, newspaper publishers could start treating their advertisers with respect by only serving ads that will be seen. Sadly, I don’t think that’s a cultural fit for today’s newspaper industry who has a habit of stretching the truth about circulation.

26 thoughts on “Newspapers Auto-Refresh Pages and Hurt Advertisers”

  1. Reuben, in this case, I opened up all of the top-30 websites in separate tabs and turned off my WiFi before driving across town. The sites that refreshed were then displaying dead pages while the ones who weren’t were still displaying their content and advertiser’s ads. Then I took a quick peak at the JavaScript and/or meta refresh tags to figure out the increments. It wasn’t a big time commitment but a case of knowing what to look for and how to measure it.

  2. it would be nice if the page stop refreshing every 60 seconds so I could look at a graph, let alone write a comment.

  3. Would it not make more sense to just put I frequency cap on the impressions you run on sites that use the auto refresh more frequently than the other ones ? that way a user can only be exposed to only a max number of advertiser impressions per visit ?

  4. Tha !, that’s an interesting way to work around the problem. It seems like a lot more work than removing a single line of code from newspaper publisher’s websites.

  5. I suppose it would be a lot more work for these papers. However I’m just trying to help out the consumer base on this one. While these companies are jockeying for position with auto refresh. It would seem to me that they could do something as simple as putting another extra line of code on my ad campaign, so as to not waste more than “X” number of impressions on someone sleeping at there P.C.

  6. That can definitely be done. Most popular ad serving platforms do have frequency caps. I imagine what that would do is roll between all of a publisher’s advertisers who are eligible for impressions, thus creating an equitable dilution of advertising value.

  7. Ed you know this has bugged the crap out of me but I have never really dugg to deep into it. When they refresh do they serve new ads? If this is true it is a bit of a double edged sword…some ads may never be seen but chances are some advertisers might get one more shot at an eyeball than they would have without auto refresh.

    Id image there is some content that is more valuable to advertisers than others, in that case I would guess that they rotate more ads on more popular stories.

  8. ryanol, the refresh scripts are simple page reload scripts, so refreshing the page manually should generate the same results. I’m seeing new ads on refreshes. Not all new in all spots each time but they do rotate.

    I don’t see how this helps advertisers at all. To me, it just creates deflation in inventory prices based on the percentage of ads that go unseen due to page inflation.

  9. In a past life as a marketing director in Multi-housing I purchased a decent amount of (around 100K) display advertising. Many sites had housing broken up by different geographic regions.

    Each section has its own ROI…say for example I have a townhome community in Plymouth, MN.

    For example one site I advertised on had subpages for the western twin cities and a sub subpage for the city of plymouth. I would get way better performance for a rotating ad on the “western twin cities” that was only shown 1/6 of the time than static (full time) ad on the more targeted “plymouth” page.

    Display advertising is expensive and is a little trickier to track than CPA, CPC We used tracking phone numbers each assigned to specific campaigns for ROI tracking. Than tracked call to appt, and appt to rental conversion ratios.

    I am not saying that I like the practice just that it might not always be a net negative for advertisers.

  10. Ryanol, I’m not arguing that impression based advertising doesn’t work. Only that it doesn’t work when it’s not seen and people buying ads in that model should understand how they’re being treated by publishers. If you figure out that your cost per lead is X, and you’re comfortable with that number, then it doesn’t really matter what kind of shenanigans are happening, I suppose.

  11. I think that’s odd Mike, if I’m engaged in an article, and simply reading through it ,I don’t know if it would take me more than 5 minuets a story. I’m no expert and maybe Ed could enlighten me, but if you refresh it yourself or jump to a different article dose the timer refresh ? or does it continue ?

  12. I’d look at it like this–
    –if I was like ryanol (and I pretty much do view pages like this), I would click on the story and read the story and ignore the ads…because I am there to read the story, period, and have trained my eyes to find the content. So like ryanol mentioned above, I wouldn’t even notice the ads changed, all I know is the damn page reloaded annoyingly.
    –then if I was more open to ads, but I am still there for the story, so I will spot the ad but still be there to read the story and want to come back to see the ad, but guess what…it has been refreshed and is now GONE!! So I shrug and leave.

    In both instances, the advertisers lose and lose…and the Strib gains in the short-term by puffing up pageviews and impressions…until the reality slaps the the advertisers in the forehead and they immediately squeeze the Strib for compensation…or shrug and leave. Kinda reminds me of how they managed the dead tree version of the Strib, eh?

    As for the viewers, we already have noted this annoyance factor…I rarely click Strib links, only if I cannot find the same story elsewhere, and even when it is exclusive, I ponder a few seconds to determine if it really interests me or if I’m just curious and can just forget it. You might be amazed how many times I just say forget it.

  13. Back when I was a regular StarTribune reader, my workflow for reading stories was to scan the homepage or section pages, then click to open stories that interest me in new tabs. As each tab opens, the auto-refresh meter starts running. It may be from a few seconds to hours later that I get around to consuming the stories I’ve opened.

    So by consuming, say, 10 stories, I may have churned through 100 pages full or ads while never seeing 90% of them.

  14. That’s Odd Mike is another poster. 😉 It depends on the article as to how quickly I ponder it…I don’t tend to bother with the quick hits articles, ohh an American Idol review postings where in two minutes I’ve seen the latest agast look shot at Simon, more likely I’m there for hard news and commentary where I’m reading at a rate where I am rolling around my opinion of the matter.

    If that makes me odd, yep, that’s me. 🙂

  15. Ed – While the data is compelling, your argument essentially comes down to whether we concur with this one statement: “it’s quite common for people to keep the StarTribune open in one of their browser’s tabs overnight so they can check the weather first thing in the morning.”

    All the rest of your data is sourced and your math is dead on, but the conclusions you’ve drawn all depend upon that statement being true. I am not suggesting the statement is false, because I have no data on which to do so, but I’m surprised you are going so far out on a limb based on a single supposition. It’s not like you to engage in ‘focus group of one’ analysis.

    Also, I’d offer another piece of data that we all know is true: Nielsen numbers are based on an extrapolation from a survey of users and can vary massively based on the depth and breadth of the survey sample. The reliability of the Nielsen data is almost always proportionate to the size of any one site’s total audience. The data for USA Today and the Times and the Post are almost always more reliable than the data for the Ann Arbor News or the Anchorage Daily News. That doesn’t mean that the data for the Strib or the Chicago Tribune or Newsday is not accurate, but it means we need to consider the likely margin of error.

    Of course, none of that addresses whether site refresh is a good idea – either from a business model perspective or from a user experience perspective.

    For sites that feature a good amount of breaking news, I personally believe that limited site refreshes that account for activity – setting a timer variable to determine whether a browser window has the focus and whether the document body has scrolled and then setting the reload timer based on that variable – are a good idea on both accounts. But I can’t point to a big news site that is using such a reload script.

  16. @No Longer There, I outlined additional scenarios near the end of the post.

    Agreed on Nielsen data. Relative data is probably the most valuable perspective gleaned from it.

  17. Pingback: SEO Book.com
  18. Ed – you’ve spectacularly missed the point. Breaking news websites need to refresh their homepages so that new stories can be exposed.

    Under your own example the user who keeps “the StarTribune open in one of their browser’s tabs overnight so they can check the weather first thing in the morning” will be getting last nights news if the homepage isn’t refreshed.

    I’m not saying that news media publishers don’t milk this, but your assertion that there is “This provides no added value to visitors (in fact, it makes the sites less usable in many cases)” is wrong.

  19. Interesting point PTM. I’m willing to concede the homepage based on your example if you’re willing to agree that it’s ridiculous when applied to every page of a newspapers website beyond the homepage.

  20. Interesting article. Do you think this inflates traffic stats as well, if a page gets refreshed does it count as an additional pageview?

  21. @Dr Jen Arnold Fan, yes, the ads are tied to page views, so the refreshes are designed to increase page views, thus ad impressions.

Leave a Reply

Your email address will not be published. Required fields are marked *