Re: Hits, pages, counting, NPD, IPRO and the whole mess

Sean Pfister (seanp_at_cnet.com)
Fri, 19 Jul 1996 10:29:05 -0700 (PDT)

• Messages sorted by: [ date ][ thread ][ subject ][ author ]
• Archive Home Page
• Search
• Next message: xerxes: "Reworked surveys, Hoffman, et al."
• Previous message: Tonyandpip_at_aol.com: "Fwd: Getting the "big guys" to lower their prices"
• Maybe in reply to: Bob Browning: "Hits, pages, counting, NPD, IPRO and the whole mess"
• Next in thread: Bob Poulsen: "Re: Hits, pages, counting, NPD, IPRO and the whole mess"

I'd like to make a couple general comments about measuring site traffic
and terminology.

Hits are a pretty useless measure of traffic volume because, as someone
pointed out, total hits is the sum of ALL client requests to the server. It
therefore includes all requests for jpegs, gifs, gif89s, text and so on.
For an advertiser, the only meaningful measurement (so far) is pages or
page views or some equivalent. Hits are probably a good measure of server
load but they don't map well to what media planners request.

Visits and unique visitors are also problematic terms, it seems to me. The
assumption is that any IP string, like foo_at_foobar.com, over any arbitrarily
selected session interval, say 20 minutes, belongs to 1 individual only.
This assumption is untested. Proxy servers and local caching will hide
portions of IP strings from the server log access files. The session
interval is entirely abitrary. I/PRO, for example, has defined a session as
something like "any user activity belonging to a unique IP string with a
maximum timeout interval of 30 minutes." In other words, if foobar.com is
idle for 29 minutes (there is a log file at 12:00 pm and then the next
occurs at 12:29 pm), all subsequent requests from the server are counted as
the SAME USER. For proxy servers, I/PRO ascribes the data from the IP
strings that they can identify.

I am not saying that these procedures are bad or unwarranted. In many ways
they are reasonable assumptions. There's a certain ambiguity in the data
though, but because we use a term like visitors or unique visits, we tend
to assume that it means something exact when it doesn't. And it isn't
always the case that one site defines visits in the same way as another
even though they use the same term.

A couple other comments...

> "Ever-popular Playboy and Penthouse, report significantly lower
> hits-to-visits ratios of 4-5% compared to other sites (average:
>13%)".

For the reasons above, I think hits-to-visits ratios don't offer a means of
comparing traffic among sites. They probably only contrast page
composition, ie which sites have a lot of gifs on their pages.

>
> [CyberAtlas] This very useful and informative page also defines
>hits, visits, and
>page views:

Agreed, it's a great resource. IPRO also has extensive definitions at their
site, I believe.

> Another popular way to state Web exposure is page views. NPD
>Research Group reports that each Web active individual visited 21.7 different
>domains......(snip)

Keep in mind that NPD meters are installed ONLY in homes, not in
businesses. "Web active individuals" should include this qualifier. There
are also some legitimate questions about the extent to which the NPD sample
is projectable to the Web-using population. They do intend to increase
their panel size from 4K to about 10K, however which may improve
projectability and decrease tolerance ranges of the data.

Sorry for the length. Your comments welcome.
Best regards,

sean pfister
director, research and analysis
CNET: The Computer Network
seanp_at_cnet.com
415-395-7805 x 1529
Fax: 415.395.9205

w w w . c n e t . c o m /
w w w . s h a r e w a r e . c o m /
w w w . s e a r c h . c o m /

• Archive Home Page
• Search
• Next message: xerxes: "Reworked surveys, Hoffman, et al."
• Previous message: Tonyandpip_at_aol.com: "Fwd: Getting the "big guys" to lower their prices"
• Maybe in reply to: Bob Browning: "Hits, pages, counting, NPD, IPRO and the whole mess"
• Next in thread: Bob Poulsen: "Re: Hits, pages, counting, NPD, IPRO and the whole mess"

With an archive of more than 14,000 postings, since 1996 the Online Advertising Discussion List has been the Internet's leading forum focused on professional discussion of online advertising and online media buying and selling strategies, results, studies, tools, and media coverage. If you wish to join the discussion list, please use this link to sign up on the home page of the Online Advertising Discussion List.