By John Gruber
DuckDuckGo Search + Safari’s Intelligent Tracking Prevention together solve the top three private browsing misconceptions.
There have been a bunch of encouraging inquiries regarding the new sponsorship/advertising initiative, but everyone wants to know about readership numbers and traffic — which of course is perfectly reasonable.
So, here’s the gist of the traffic this site generates, and how many impressions a sponsor can reasonably expect in a month.
According to my stats from Google AdSense, I averaged a little over 6,000 impressions per day for the last two months. This number has been slowly inching higher over the course of the last year. Traffic is generally lower on weekends (especially Sundays) and higher on weekdays.
I think Google’s estimate here is very accurate, and comes very close to the number I come up with analyzing my server logs. (No less than Tim Bray agrees regarding the accuracy of Google’s “impressions” count.) However, this only takes into account my home page and my individual article archives — which were the only pages where I displayed Google’s ads.
My projects section — for Markdown, SmartyPants, etc. — averaged between 700 to 800 page views per day for the months of August and September to-date. Unlike the Google ads, my sponsorship ads are running on these pages too. (They’re running on almost every page of the site in fact.)
So my honest estimate is that my sponsorship ads will average around 7,000 total impressions per day. Assuming I sell out all five slots, each 4-week sponsor should receive about 39,000 impressions:
((7000 pages/day) * (28 days)) / 5 ad slots = 39,200 total impressions
In terms of CPM, that works out to $500/39K, which is roughly $12.75 CPM. For a weekly $200 sponsorship, it’s roughly a nice even $20 CPM. When there are fewer than five active sponsors, those CPM rates come down.
To be clear, the appeal of Daring Fireball as an advertising forum is not the raw page view counts — there are scads of sites, including many weblogs, with significantly more traffic. Daring Fireball’s format — longer articles posted around twice a week — generates many fewer page views than a weblog consisting of multiple blurbs posted several times per day. There are other factors to consider as well, such as the fact that I don’t offer comments — a feature that encourages regular readers to reload pages several times per day.
How many pages I serve I can reasonably estimate. How many readers that translates into — well, that involves some guesswork. Extrapolating web server logs into readership numbers is right up there with political election polling in terms of exemplifying the old “lies, damned lies, and statistics” aphorism. [Update: I had originally attributed this quote to Mark Twain, but in fact it’s from Benjamin Disraeli, whom Twain in turn quoted in his own autobiography. Thanks to Paul Beard for the correction.]
In a typical 48-hour period after I publish a new article, I serve pages to around 10,000 unique IP addresses: roughly 6,000 for my home page, and 4,000 for individual articles in my archive. This is far from an exact count of unique readers. Some readers access the site from multiple locations (work and home, for example); in other cases, multiple readers are hidden behind a single IP address. But I think it’s a fair, conservative estimate of Daring Fireball’s regular, loyal readership. If anything, I honestly suspect the true number is higher than 10,000 — if for no reason other than that many readers might only check the site once a week or so.
Another way to look at it is to count unique IP addresses accessing my
public RSS feed. In the same 48-hour period after I publish a new
article, I average a hair over 10,000 unique IP addresses accessing my
/index.xml resource. That doesn’t count the 1,272 subscribers reading
my feed via Bloglines — a deservedly popular web-based
aggregator that only accesses my feed once and shares the data with
every Bloglines user who is subscribed to my feed.
Unique IP addresses are easy to count, but again, there are all sorts of reasons I can’t state with certainty how many people are subscribed to my feed. Some people might subscribe both from multiple locations (e.g. Bloglines at work, NetNewsWire at home). And because Daring Fireball is in the default subscription list for popular Mac aggregators such as NetNewsWire and PulpFiction, there are surely some subscribers to my feed who aren’t actually regular readers of the site — they simply never bothered to unsubscribe.
The most interesting metric would be how popular Daring Fireball is relative to other web sites. But if estimating readership is guesswork, estimating relative popularity compared to other sites — without access to verifiably-accurate copies of other sites’ server logs — is well-nigh impossible.
But let’s take a stab at it anyway.
Bloglines maintains a list of “most popular blogs”, ranked in
order of how many Bloglines users are subscribed to each feed. Because
Bloglines is web-based and thus cross-platform, I suspect their rankings
are pretty accurate. At this writing, Daring Fireball is ranked 99th.
That’s based solely on Bloglines users subscribed to my main
/index.xml RSS feed. If you added the 140+ Bloglines users who
are subscribed to my members-only feeds, the ranking would jump
another 10 spots or so.
Now that’s certainly not to claim that Daring Fireball is one of the 100 most popular web sites in the world. I mean, come on. But it strikes me as a fairly accurate gist of where Daring Fireball stands in relation to other weblogs — and perhaps more importantly, how popular Daring Fireball is amongst the sort of people who know what a “syndicated XML feed” is.
If humility didn’t stand in the way, I might point out that very few of the sites in Bloglines’ rankings listed ahead of Daring Fireball would be appropriate venues for advertisers targeting Daring Fireball’s savvy Mac-enthusiast audience.