Neale, you need to read the entire post Neale Talbot has responded at warblogger watch to my post below on tracking unique visitors, and once again, his math is suspect.
Let the festivities begin!
First, yes, I used a projection for this month, because the trend shows that traffic at Instapundit is growing, and that Instapundit has a monstrous retention rate. As proof, consider the following:
At the time of Neale's post, the average daily unique visitor count was 14560. Right now, at 7PM EDT, his daily average is 14773. These averages are computed over 31 days. Since the average is increasing, that means that he is getting more uniques per day than the previous average. Further, if Neale had looked past the averages to the daily uniques for the month of June, he would have found, as I did, that Instapundit's traffic has remained fairly constant at about 14,000 to 18,000 uniques per day during the week, and 7,000 to 9,000 uniques per day on the weekend. Unless you assume he is making a one shot splash on a daily basis, you must conclude that he is retaining a large portion of his increase in regular readers, which means that the total for this month is liable to be retained in large part through the next several months. It is not unreasonable to expect that Glenn will keep 70% of his increase this month over the long haul.
Second, Neale's good friend Jeff's analysis is right on. In fact, I addressed that in my post, assuming that approximately 80-90% of his daily total represent daily readers, not new readers, which is why my estimate stayed around 12,000 uniques PER DAY.
Next, Neale compares two numbers with different origins.
Using Richard's maths, Insnayapundit gets on average 225691 hits on average per month. So 225691/30 should equal 7523 hits a day. Why doesn't that match the average uniques? Why is it less? Isn't that a sheer impossibility?
Next, Neale questions my math, revealing his own weakness in the area.
7523 multiplied by 8% gives 602 for AOL users. Applying the AOL factor of 3, we can assume that 1805 (Richard multiplied wrong on his page, multiplying by 2). 7523-1805 = 5718. At the upper AOL level of 14% that leaves 4363.
From my post:
Now for more calculating. 13,000 daily hits multiplied by 8% gives 1040 hits from floating IP's. Applying our hit factor of 3, we can assume that 2080 hits represent repeat page views (1040 AOL users visiting 3 times each, count only the first visit, discard the remaining 2 visits),
I multiplied by two because, as I showed above, I still counted one unique visit out of the three. 3-1=2. Get it this time, Neale? If you subtract all three visits from the typical AOL user, you are now undercounting by a third.
Further from my post:
Also from eXTReMe, we find out that AOL traffic represents less than 4% of Instapundit's total traffic, which indicates that our first estimate was too conservative.
Of course, Neale continues to use the 8-14% estimate, rather than looking at the actual report, because it bolsters his total. Looking at eXTReMe, we see that AOL surfers represent about 0.19% of Instapundit traffic. However, I include half the Netscape 6 users, who may be AOL users. This is a complete guess, and could vary significantly in either direction. So we have a total of 3%
So, 7523 multiplied by 3% equals 14 visitors multiplied by (3 minus 1) equals a correction of 451, leaving 7072 uniques per day, within Neale's range, but at the upper limit, and certainly higher than his 3000 lower limit. However, taking into account the retention rate discussed above, we get the following:
14,799 uniques per day (current 30 day average) subtract losses of 30% (estimated at 4400) gives 10,399 uniques per day on average. Applying the AOL correction of 3% times 2 ( approx 623) leaves 9776 daily unique visitors.
If he maintains his traffic, as seems likely by looking at his daily tracking numbers (they aren't falling off) the total will exceed my original estimate of 12,000.
14,799 uniques per day. Applying the AOL correction factor of 3% times 2 (approximately 888) leaves 13,911.
From Neale:
Prove me wrong, children, prove me wrong!
Consider it done, Neale! Only the most conservative estimate manages to fall within your range, and that just barely.
Posted by Rich at June 20, 2002 7:53 PM