Nieman Foundation at Harvard
Google’s economist-in-chief, Hal Varian, was the keynote speaker this morning at the Federal Trade Commission’s second round of hearings on the future of journalism. (The study is entitled “How will journalism survive the internet age?” Round 1 was held in December; transcripts and other material are linked here — scroll down. Not to be outdone, the Federal Communications Commission also has a project studying pretty much the same thing.)
Here’s the slide deck from Varian’s presentation, entitled “Newspaper Economics, Online and Offline”:
(Lab readers may recognize some of the slides and data as having appeared here previously. I provided some input to Varian as he prepared his talk. Varian also posted the presentation and a summary of his remarks at the Google Public Policy Blog.)
Varian took a leave of absence from academia a few years ago to take charge, among other things, of tweaking and explaining the workings of Google’s brilliantly clever auction pricing mechanisms for text ads. He’s also involved in analysis, finance, corporate strategy, and public policy.
Google has been in the crosshairs of the newspaper industry as newspapers struggle to hold onto print revenue in the face of digital onslaughts, including the text ads that bring Google the bulk of its $24 billion 2009 revenue (equivalent to about 85 percent of the entire newspaper industry’s ad sales). The industry’s biggest beef with the search giant is that it sees content aggregation on Google News as pilfering, without compensation, page views that ought to be going to newspaper sites; Google counters that it delivers a hefty share of total traffic at news sites (and that publishers can opt out of Google News if they really want to).
Varian offered no magic potion for newspapers, other than exhorting the industry to “experiment, experiment, experiment,” and to get better at analyzing and exploiting the information they can glean from their site visitor data. He began with a series of slides illustrating the dismal trend lines of the newspaper industry and its place in the media environment:
Twenty-six percent of all Americans (46 percent of those under 50) access news by mobile phone, and an astonishing 80 percent get news from e-mailed links, Varian said. But (as first analyzed right here at the Lab), only three percent of all consumption of newspaper-generated content happens online; 97 percent is still consumed in printed newspapers. This is true whether you measure pageviews in print and online, or time spent with printed newspapers versus newspaper web sites.
Still, news consumption ranks high among online activities of consumers, with 39 percent getting online news “yesterday” (according to the Pew Internet and American Life Project, 2008), ranking it the third-most-popular reported activity (after e-mailing at 56 percent and using search engines at 49 percent).
Using Google data, Varian showed that accessing news exceeds search on weekdays, but drops to a fraction of its weekday level on weekends, when search leads. From this he concludes that much access to news and newspaper sites happens in the workplace, when consumers have time only for quick checks of headlines, not for in-depth reading.
The challenge for newspapers, therefore, is to “increase involvement in the news by turning it back into a leisure-time activity.” He sees tablets and mobile phones as helping to do that.
ComScore data shows that search engines send 35 to 40 percent of traffic to major U.S. news sites — so, assuming that this traffic monetizes about the same as other traffic, search engines must be driving 35 to 40 percent of revenue, as well, Varian said. But he suggested newspapers could do a better job using the data that comes along with the search click (the keywords used in the search), and using it to categorize the reader’s interests and tailor content suggestions and advertising accordingly.
One problem with this, I’ve found, is that about half of visits that come via search engines tend to be generic — users type the name of the paper, the URL, or a variant thereof, into the search field, using it in lieu of their browser’s address bar. But of the non-generic search clicks, Varian pointed out that most are for categories like sports, news/current events, and local (Google’s categories), which are difficult to monetize, while few are for the more lucrative areas of travel, health, shopping, computers, and electronics. “So the news narrowly defined is pretty hard to monetize.”
What about charging the consumer for news online, then? Varian’s answer:
My view is, yes, I mean, you should try for sure. But there is this difficulty that you run into when you start thinking about the economics of it is that you can really only charge for thing ifs they’re differentiated. There are a lot of substitutes for a product then it’s hard to charge for it. Then you have this problem, what economists call Bertrand competition…You get this competing down to the lowest common denominator.
So you really have to have news that’s highly differentiated in order to support a charging model. One time I thought, well, local news, that’s highly differentiated. Local football scores, things like that. Then I realized all of the moms and dads are in the audience on twitter with the mobile phones, maybe the news isn’t so highly differentiated after all, they’ve got mostly specialized industry content, points of view, analyses are not easily imitated are also a case that they can differentiate news. I’m agnostic on whether the charging will work. I think it’s worth a try, but you can only try it for something that’s going to be unique content. It’s very hard to charge for, let’s say, the weather, or something of that sort.
Varian concluded with this exhortation to publishers: “The three things newspapers should do is experiment, experiment, experiment!” He cited a few options from Google, like Living Stories and starred stories that can be followed for updates during the day.
“I’m a big fan of the new devices,” he said. The iPad, Kindle and other tablets introduce a “completely different ergonomics for accessing the news…so what I believe they’ll see is a merger of the TV, magazine, radio, and newspaper experience. You’ll have a device which will access all of the different medias. Give you a deeper — potentially deeper involvement with the news…So I would like to see this — this area develop and we’re doing what we can to help that happen.”
Finally, Varian urged newspapers to better exploit the information they have:
You know, in many cases, the newspaper website is seen as — as something that for the techies or the person who’s managing the web log [stats] is doing it just to look at how performance is working. But it’s hugely valuable information in those web logs [stats] — both from an editorial point of view and from a marketing point of view. There’s lots of interesting things that you can do when you understand why people are coming to your site, where they’re spending the most time, what they’re coming back to. It’s just extremely valuable information. I think newspapers can spend more time on analyzing that information and end up with better ad effectiveness measuring better contextual targeting and editorial targeting.
Here’s a full transcript of Varian’s remarks, as recorded by the FTC’s transcription service. (Note: This is not fully cleaned up or compared with a recorded version. I’ve inserted the slide numbers at the appropriate points.)
[UPDATE, March 19: Commenters on this post questioned the original slide deck’s assertion that five percent of newspaper ad revenue came from online advertising. Our friend Robert Heath of Rough Numbers reran the numbers and discovered an error, to which we alerted Varian. The actual percentage, 8.2 percent, is now reflected in the slide deck, which is also corrected on the Google Public Policy Blog.]
Wow, thank you very much for that kind introduction. Happy to be here. [1] As you heard, we’re going to talk about on-line and off line economics of newspapers. [2] And basically this is going to be mostly a fact-based presentation, looking at revenue, costs, advertising level change, composition, and so on. Most of the talk is from the data from the newspaper association of America that’s put up a lot of trends on the website, a key foundation of some of the other sources and a little bit of Google data that’s also emerged with this report.
[3] So I want to start off with a little overview of what revenues and costs look like for newspapers. And basically the bottom line here is 80% of the revenue roughly comes from advertising, 20% from sales. If you break down the cost side of newspapers, turns out that about 50% of the costs are production and distribution, that is the physical production and distribution of the newspaper, obviously it’s attractive if you can reduce your costs by 50% for any business. So the promise to the internet is just to reduce costs. I understand we’re going to hear much more detail about that this afternoon.
[4] If you look at ad spend by medium in the United States, I pulled this data from the U.S. Statistical Abstract. Of course, the big gorilla in the room is TV. You look at broadcast and cable TV, you’ve got by far the largest expenditure on advertising on those two media. Surprising enough, the next biggest thing is direct mail. Then after direct mail comes the — comes the newspapers. You look at how things have changed over the years, broadcast TV has gone down a little bit. Cable TV has grown by quite a bit, almost a factor of three. The internet’s grown from nothing in 1995 to about 5% of ad expenditures in 2008. And newspapers, As you can see, have contracted from about 23% down to maybe 13% or so. So the big changes are apparent in this diagram. And I guess the next talk is going to be perhaps some more up-to-date figures on the advertising business and newspapers. Newspapers, of course, are still about three times as large in terms of ad revenue as the internet, so there’s still quite a major force in the advertising world.
[5] This is another chart showing pretty much the same thing. If you look at newspapers, that’s the blue line, they’ve been going down since basically 1950 in terms of media share. If you look at the yellow line, that’s TV and cable. That’s been going up quite dramatically over the same period. And way down there in the bottom right-hand corner, that light blue line, is the internet which came from pretty much nothing up until the — maybe late 1990 s started to become a force in — in advertising and other media stayed more or less the same.
[6] Now this is a plot of GDP which I just put there to have a general measure of economic activity and newspaper ad revenue. And I’ve adjusted it but the consumer price index that you can see what the changes will be in the real term. So basically we have real GDP and real newspaper ad revenue. And you can see, it’s pretty much pieced back in the late ’80 s, since then, more or less conference in the last couple of years where it took a big dropdown. By the way, the vertical grade bars are recessions. One thing to note is that typically during recessions, advertising expenditures are quite sensitive to cyclical conditions so you can see GDP dropping and advertising expenditures dropping as well. The last couple of years have been dropping outside and even more than the economy would indicate and we’ll see an echo of that in one of the — one of the later slides. The important point is that newspaper ad revenue pretty much Maxed out way before the internet came on the — on the scene.
[7] This is a picture of what ad revenue looks like by type, again, measured in constant dollars. So typically it’s broken down into four different categories, retail, which would tend to be local stores, national, which would be national brand advertising, classified, the blue segment there, and then on-line is the tiny little green segment that kind of popped up a few years ago. You can see what’s going on is retail advertising has been growing over this period. The brand advertising has been contracting and classified advertising stayed pretty much the same up until the last few years at which point it dropped fairly precipitously.
[8] This is the same chart only measured in shares so you can see the share and we’ll have a lot more clearly. I think the important point to note here is the on-line ad revenue is still — as of 2008, at least — is substantially less than 5%.
[9] What about circulation? If you look at circulation, the chart on the upper left-hand corner, the circulation stayed constant for a long period of time and drop in the last couple of years, but, of course, it’s a little bit misleading just to look at total circulation, what you’re most interested in, most likely, is circulation per house hold. So if you look at paid circulation per person, over on the right, you can see it was declining since the ’60 s and pretty much a steady manner. The interesting thing is, if you look at ad revenue per reader, or ad revenue per circulation, it actually was increasing since the late ’60 s with a few up s and downs in the recessionary periods and so on, but by in large increasing up until very recently in the last few years. But the ad revenue per circulation is going up even though ad revenue is going down because the circulation has been going down so much. So it’s the denominator that’s been causing this effect.
[10] And here’s another chart just showing circulation which, again, has been remarkably constant between say 55 million and 60 million copies.
[11] And here’s a chart of circulation per household, which is also been pretty stable in terms of its decline. Back in 1947, you were seeing a little over one newspaper per house hold, which I presume is morning and evening editions in many cases. But that’s gone down to something like 40 — .4 newspapers per household in today’s world.
[12] And this is the chart that — well, we just heard Susan refer to that now the internet has surpassed physical newspapers as the popular way of accessing information. I would say television is — got a pretty substantial lead on both of them. And, of course, most of the internet access is access to newspaper sites. So they aren’t, of course, the physical paper.
[13] In that same report, there were some interesting trends about getting news by phone. 26% of all Americans said that they actually access news on their phones and 43% of those under 50 — so this is yet another medium by which people can access news. But in many cases, given the interface that’s available, people are looking at weather or at current events because reading in depth on your phone may be somewhat inconvenient. I thought one of the more fascinating numbers that came out of the PEW report is that 8 80% of people get news by e-mailed links. That’s one of the more popular distribution mechanisms now. You see an interesting story, you send it to their friends. You go to the websites, you see the most mailed stories MRKS are accessed on people’s computers and now, increasingly, on handheld devices. And we shouldn’t think of a single medium per person. Half the population surveyed said they used four to six different media for accessing news. So it’s important to distinguish in these discussions between newspapers traditionally considered as the physical newspaper and, of course, all the other ways you can access news, on TV, on your phone, on your computer, your lap top, etc.
[14] Now, if you add it all up and you look at the difference between physical newspaper reading and on-line newspaper reading, you get this kind of amazing statistic that’s due to Martin Langeveld at Harvard[‘s Nieman Journalism Lab]. Only about 3% of total news comes on the computer. Most of it comes from looking at physical newspapers. You get nice numbers looking at the web data. This is data from the Newspaper Association Of America. People are spend 38 minutes per month on on-line news which works out about 70 seconds a day. Whereas a person who reads a physical newspaper tends to spend about 25 minutes a day. There’s also time use studies to back these numbers up. So even though accessing news on-line is a very popular thing to do, it’s actually the case that people are not spending nearly as much time on the newspaper on-line as those people are who are reading physical newspaper. Of course, they’re different populations, so you have to compare these carefully. But roughly speaking, about 3% of either page views or time accessing on-line news — sorry — 3% of the total access to newspapers is done on-line. On the other hand, it’s accessed quite often.
[15] This is from data from the U.S. statistical abstract. Also it came from Pew, that roughly 40% of adult internet users say they accessed news yesterday. And. In, if you look at those with household incomes of $75,000 or more, it’s about 53%. So it’s very popular to access that on-line news, it’s just that people aren’t spending a huge amount of time on it, at least compared to the people who are reading the physical newspaper.
[16] If you look, for example, at total number of hours per year where people are accessing newspapers or reading newspapers, it’s about — let’s see, in 2008, 168 hours per year. So roughly works out to 25 minutes a day. In terms of physical newspaper consumption — that’s the same order of magnitude as the time people spend on the internet.
[17] News — the third most popular activity on-line, sending a regular e-mail, using a search engine, getting news on-line. Those are, again, the three top things that people do on the internet, but they’re spending a lot more time, for example, reading e-mail than they are looking at the on-line news. Now this is a little bit of a paradox. Let me stop for a minute and show you the charts. The paradox is, it’s popular to access news on-line, but they don’t spend time doing it. Why is that? That’s the mystery. How much time they do it compared to physically reading the newspaper.
[18] So I pulled some Google data and I looked at the time use pattern of access to Google news. So what you got down there on the bottom are the hours in the over a couple of weeks. The two little small bumps are the weekend access. And the — the five bumps between them are the daily access. So the red line is search activities. This is how many people are searching Google for things. And the blue line is the news activity. So I plotted both of these charts from the area of — each graph is normalized to be one, so it’s measured in percentage terms.
So what’s the first thing you see in the blue line is a lot further up than the red line. What that says is that people are accessing the news during the day a lot more frequently than they’re doing searches. And if you go over to look at the weekend, you can see the searches dramatically exceed the news, people are doing searches more on the weekend than they’re accessing the news. What that suggests to me is that people are accessing on-line news a lot during business hours. It’s not so surprising that they’re not spending a whole lot of time on it because offline news reading is a leisure-time activity. You do it over a cup of coffee, you do it in the evening, maybe. Whereas on-line news reading, that’s a labor time activity. People snatch a few minutes out of the day to check the sports scores or the headlines or something of that sort. So if that’s true, people are spending much less time looking at on-line news than they traditionally spent reading on-line news because they’re doing it during working hours, much less during leisure ours. During leisure hours, you might sit and watch TV, as a matter of fact, it would be a common thing to do.
So the challenge, I think, that’s facing the newspaper industry is to try to turn that on-line newspaper access which is much more attractive way to reach a broader audience is to increase involvement of the news by turning it back to a leisure-time activity.
[19] If you look at the value of clicks sent to newspapers, according to COMSCORE, it’s 35% to 40% of the traffic to news sites. That monetizes about as well as other traffic, that means that search engines are driving about 35% to 40% of traffic of revenues, on-line news sites. Which is a substantial amount. However I have to remind you that the on-line news revenue is about 5% of the total. So even though they’re driving a substantial fracture of the on-line revenue that’s still a relatively small amount of the total revenue.
[20] One thing that’s interesting to do is if you look at a search click that goes to the newspaper site, the newspaper is sent a query — or any site, not just the newspaper site, the site is sent a query that generated that search click. And that means that the site that received the search click could direct the user to the appropriate section of the site. So you can take those queries that people are issuing when they click on news sites and ask, what are the categories? What are people looking for when they go to these on-line news sites?
And I’ve done that. It turns out that the kinds of things that people are looking for when they’re going to these on-line news sites are sports, news and current events, and local — those are the top-level categories that we use at Google categorize search clicks. But there’s relatively the same in travel, shopping, so on. And roughly the same in entertainment, computers, and electronics. I’m comparing searches that go to newspapers to just searches in general that go to sites that aren’t specifically classified as newspapers. I say newspapers, I mean sites indexed by Google news.
Now the bad thing — or maybe not the bad thing, just a fact is, that if you look at the money in on-line advertising, the money is in categories like travel, health, shopping, and consumer electronics. But if you look at the revenue that’s going to newspapers. That’s in sports, news, and current events and local. And believe me, it’s very, very hard to monetize those categories because there isn’t as much consumer dollars spent in those areas as there are in areas like travel, health, and shopping.
[21] So the news narrowly defined is pretty hard to monetize. Despite the fact that it’s popular and frequently accessed, there’s a relatively low level of involvement because of the time constraints that people face, and it’s typically not a highly commercial activity. In fact, newspapers have never made money from news. You look at where the revenue came from, they made money from the business page, the automotive page, home and garden, travel and technology, all those parts of the newspaper that wasn’t the raw news, not the newspapers. Why? You can target ads, not surprising that people who read the automotive page are interested in buying cars or people who look at the travel section might be interested in taking trips.You can see targeted ads in the physical newspaper but tied to the sections. Then it’s the revenue generated from those sections which are used to cross subsidize the actual production of news.
And what’s happened is, this has been a problem with this intermediation that now people can go directly to finance sites, to auto sites, to consumer electronics, books, to travel sites, real estate sites, and so on, so people go directly to seeking those specific sources of information, they tend to bypass the traditional sections of the newspaper and so the cross subsiization model that’s worked for many years has not really work ed now. It’s very hard to do conceptual targeting of the news. If you’re reading the travel section and you see a story about Hawaii, you wouldn’t be surprised to see ads for travel to Hawaii next to that story. If you read the news section and you see bombing in Baghdad, you’re not likely to see travel ads or anything else particularly relevant to that story. So it’s very, very difficult to do the same kind of cross subsidization we’ve seen work in the past.
[22] If you go look at advertising verticals for newspapers, you can see 20% is general merchandise, 14% financial. That would tend to be in the business section of the paper, home supplies, furniture and so on. So you look at the breakdown of where the money is coming from, then it tends to be somewhat different from the kinds of things that people are making money on on search engines and general internet advertising. Of course, all this doesn’t mean that newspapers aren’t valuable. You heard earlier — I would absolutely second that is critical both from the individuals and societal point of view. People find it valuable because people are going to look at news on-line. We see half of internet users read news on-line at some time or another. They just don’t spend a whole lot of time on it.
[23] I’ve seen this big debate on whether you can charge for news, replace the advertising model. My view is, yes, I mean, you should try for sure. But there is this difficulty that you run into when you start thinking about the economics of it is that you can really only charge for thing ifs they’re differentiated. There are a lot of substitutes for a product then it’s hard to charge for it. Then you have this problem, what economists call Bertrand competition — one seller sets it price here, one could sell it down. You get this competing down to the lowest common denominator.
So you really have to have news that’s highly differentiated in order to support a charging model. One time I thought, well, local news, that’s highly differentiated. Local football scores, things like that. Then I realized all of the moms and dads are in the audience on twitter with the mobile phones, maybe the news isn’t so highly differentiated after all, they’ve got mostly specialized industry content, points of view, analyses are not easily imitated are also a case that they can differentiate news. I’m agnostic on whether the charging will work. I think it’s worth a try, but you can only try it for something that’s going to be unique content. It’s very hard to charge for, let’s say, the weather, or something of that sort.
[24] So, in summary, if you go through and look at all of this, newspaper ad revenue is pretty much cost adjusted for dollars. The circulation per capita is going down since 1947. The really big increase of advertising revenues come from cable TV and that’s way before the internet. You do have this problem with on-line news that people are using it differently than they’ve used offline news. They tend to access it more episodically, and the challenge that the newspapers face is how can they use that to — how can they turn that deeper access to the news to the kind of deeper involvement that they would like to have? Maybe what you need, everyone said this is maybe not the news, but engagement. You need to increase the engagement with news.
[25] And the three things newspapers should do is experiment, experiment, experiment. Google has been working on doing some of the experimentations, I think a promising avenue is try to link news access during the day so you use this rather brief occasional access to stories, to a much bigger engagement, partially by shifting some of the access to leisure time.
So we’ve done things like living stories where you work with major newspapers to try to string together all of the items about a particular story as the newspaper developed through the day. Got this capability called star stories, you can look at a story and star it and then you can follow what happens in that story. Maybe look at it later when you have some free time, and other things like that. I’m a big fan of the new devices. I think that things like the ipad or the kindle and this whole group of tab let computing is going to potentially make a big difference because it gives you completely different ergonomics for accessing the news. If people are accessing on-line news at their workstation, computer, or their laptop during the day and they have a lot of things going on, when you come home, probably you don’t want to go sit in front of your laptop or your workstation at home to do the same thing. What you might want to do is sit in your easy chair and look at your tablet where you can follow some of the stories that you might have seen accessed originally at work.
Of course, this isn’t going to be a flat textural description, it’s going to be multimedia in those devices, and so what I believe they’ll see is a merger of the TV, magazine, radio, and newspaper experience. You’ll have a device which will access all of the different medias. Give you a deeper — potentially deeper involvement with the news. Because what happens with TV is you get this emotional experience from the visual side, but in many cases, it’s frustrating because you can’t go deeper in to the story because the newspaper, the physical newspaper with textural material you can go deeper in the story but maybe don’t have the same emotional involvement, get them both together, then potentially you can have a very positive and interesting and worthwhile experience. So I would like to see this — this area develop and we’re doing what we can to help that happen.
Finally, the last point is newspapers should better exploit the information they have. You know, in many cases, the newspaper website is seen as — as something that for the techies or the person who’s managing the web blog is doing it just to look at how performance is working. But it’s hugely valuable information in those web logs — both from an editorial point of view and from a marketing point of view. There’s lots of interesting things that you can do when you understand why people are coming to your site, where they’re spending the most time, what they’re coming back to. It’s just extremely valuable information. I think newspapers can spend more time on analyzing that information and end up with better ad effectiveness measuring better contextual targeting and editorial targeting. I think I’ll end there. And thank you very much for your attention.
