Skippy Records

Instead of blog entries on this topic lately, I have been drafting an article summarizing story to date. You can download an early draft of Online Newspaper Revenues: Albatross, Lifeboat or New World?.  Please send your comments and crticisms.  Just leave a comment or email me directly at scott[at]drskippy.net.

The short summary is that newspaper online revenues are meaningless for smaller newspapers (~40,000 circulation) and are unlikely to save the day for larger newspapers. 

Posted by DrSkippy27 at 09:52 AM | Permalink | Comments (0)

We hear a lot about how newspaper circulation continues to decline a few percent per year. Now that many newspapers have very active Web sites, how are online and offline audiences correlated? Based on current trends (decreasing circulation, increasing Web traffic), what can we say about the future of the newspaper business?

Some data will help build intuition and point us toward a model.

The figure below shows Newspaper Daily Ciruclation (from Info Please) vs. Unique Web Site Visitors (stats from Nielson) for about 3/4 of the Top 100 newspapers in the US.  The ciruclation numbers are daily circulation, while the traffic is given in unique Web site visitors per month. Download the combined data as tab-delimited text file.

This correlation gives a rule of thumb for comparing the relative success of newspapers in attracting online readers (see the formula in the plot). As circulation decreases, will online readership adauqately compensate?

The next step is to compare revenue generation per circulation and unique web visitor. Stay tuned...

Posted by DrSkippy27 at 01:31 PM | Permalink | Comments (0)

I have blogged previously about the economic interplay of the news/story content creation, advertising (classified and display) and circulation (printing and distribution) silos of the newspaper business.  A Wall Street Journal article has some interesting data that seems consistent with what I have seen.  The difference between a newspaper's ability to generate revenues online and off goes a long way to explaining the newspapers efforts to keep a kung fu grip on tradition revenue streams.  This data often looks strange to people with more experience online than off (for example, see some of the comments at Freakonomics).  The geographic monopoly and passage of 100+ years of doing business have given this business time and space to learn to generate cash.  And the age of the business and its slowness to move to "cool" Web technologies has kept many of us from looking closely at the newspaper business.  The magnitude of the advertising revenues, the margins on advertising, the overall profitability of newspapers, average salaries of employees, etc. surprise many of us in the online world.

Posted by DrSkippy27 at 04:21 PM | Permalink | Comments (0)

A few days ago, I posted Craigslist listing data for selected cities. To take the analysis a little further, today I looked at the data that best fit exponential growth and calculated doubling periods for each site (doubling periods are easier to think about than exponential growth rates).

Here are the doubling periods in days and the DMA sizes associated with each Craigslist site. DMA numbers are the number of households in the area.

XXX.craiglist.com	DMA Housholds	Doubling Time (Days)
tallahassee	260,194	247
minneapolis	1,739,407	320
louisville	614,940	236
boise	239,376	111
denver	1,550,960	322
miami	1,515,347	356
lasvegas	635,356	318
memphis	668,804	236
westpalmbeach	707,934	188
indianapolis	1,029,361	231
fortmyers	440,261	169
orlando	1,287,863	190
houston	1,859,586	215
dallas	2,201,625	203
nashville	922,435	176
sanantonio	753,076	165
tampa	1,663,780	188
jacksonville	614,068	169
Average		225

Below, the data are shown as a scatter plot. There doesn't seem to be much correlation between DMA size and growth rate. In fact, the fastest growing Craigslist site also has one of lowest number of (DMA) households in the data set. On the Boise, ID, Craigslist site, the number of live listings per day in the for sale, services, jobs and housing is doubling every 110 days.

This histogram shows the distribution of growth rates for the data set. The average doubling time for the number of live listings per day on Craigslist is 225 days for this data set.

Some Caveats:

• Ad duration: it appears that Craiglist may adjust the ad duration and/or scrub ads periodically. It may be that they control the number of live ads for a new site by increasing the ad during when sites are young and then cutting it down later to ensure fresher content. This means that Craigslist may target a specific growth rate.
• Categories: The categories of for sale, services, jobs and housing were chosen to match traditional newspaper categories. The volume of traffic and content in other Craigslist categories may attract online visitors and effect the overall growth rate of the listings counted in this data set.
• Life cycle: I don't have data showing the start date for various sites so this data glosses over any differences in the life cycle of a new Craigslist site and how that may be affected by DMA size.
• "Fits exponential Growth" means that I performed linear regressions on Ln(lisings) vs time and chose all the sites with a correlation coefficient > 0.9.

Posted by DrSkippy27 at 04:45 PM | Permalink | Comments (0)

These graphs show total live listings per day for for sale, jobs, services and housing categories.  The cities are grouped by size with the largest Craigslist sites in the first graph and some smaller sites in the second.  The categories are chosen to closely mirror categories of classified ads newspapers offer.

The data (tab-delimited windows text file) are harvested from the Craigslist home-page-reported listings for each property by a Python script.

 

 

 

 

 

Posted by DrSkippy27 at 12:23 AM | Permalink | Comments (0)

Why is my local newspaper classifieds site so hard to use? Why do they have different searches for different types of classifieds? Why are their multiple logins? Why can't I set up an RSS feed of the newspaper classifieds search? Why are newspaper classifieds $10s or $100s while Craigslist ads are nearly all free?

The answers I often hear are the cost of acquiring user generated content is very low; the cost of the infrastructure for online classifieds is low; users are attracted to the marketplace with lots of buyers and lots of sellers--'they [newspapers] just don't get it.'

Not so fast. Back-of-the envelope estimates show newspaper behavior is coolly rational.

Newspapers have operated with a local monopoly on attention. The "value" of the newspaper to advertisers is audience--circulation and demographics. The traditional way to realize the value of this audience has been through an elaborate revenue-cost relationship between the large costs of news reporting, printing and distribution and the very high-margin businesses of collecting advertiser-generated content for publication.

There are three main sources of newspaper revenue: Classifies (30-40%), Circulation (10-20%) and Advertising the remainder.  For a quick calculation, assume the Classifies are 35% of a given newspaper's revenues. Expenses are driven by these operations, but also include news reporting, printing and distribution, these two are large fixed costs.  It is clear that traditionally, newspapers have realized huge margins on Classifieds.

Looking at the larger picture, though, its part of an overall system that used to work great.  This is how the newspaper business has worked in the past--advertising paid the bills, reporting and community drew the audience.

Why not free online classifieds? Moving to free classifieds overnight throws things out of balance.  Look for the newspaper industry to try something more incremental.

 

[Housekeeping: Technorati Profile] 

Posted by DrSkippy27 at 11:33 AM | Permalink | Comments (0)