NONE: ONLINE-ADS>> More on Click-Thru %s
ONLINE-ADS>> More on Click-Thru %s
Rick Bruner (rick_at_bruner.net)
Sat, 31 May 1997 20:47:09 -0700
While several folks have already made a number of good points as to why
quoted click-through ratios are deceptive, I have two more I don't think
have been mentioned.
1) Caching:
While someone on the list pointed out that the caching that takes place by
browsers and proxy servers accounts for more actual impressions than what
sites charge advertisers for, caching inversely lowers the click/impression
ratio. Most browsers create a local cache of files (graphics, html, *ad
banners*, etc.) from Web pages you visit. If you go back to the same page,
the browser tries to download only changed files, while pulling the files
it thinks haven't changed from your local cache to conserve bandwidth.
Companies and online services like AOL use proxy servers similarly. When
the 1st user in a given period visits a page, all the page's files are
copied to a cache on the proxy server, so the next person to hit the same
page is served from the proxy's cache to conserve the bandwidth.
Hence, a lot of ads are being served out of local and proxy caches and not
getting counted by the site. Meanwhile, the actual number of click-thrus is
a known figure. That's because in most cases the URL to match a given
banner (telling where to go if clicked) is not set in the page's source
code, but rather it's served separately by the publication site. So even
when someone seeing the ad from a cache clicks on it, the request goes back
to the pub site in order to reference the advertiser's site, so the
publication does accurately count all click-thrus. But because the number
of impressions is higher than what the publication counts (due to caches),
the ratio of one to the other that they're quoting is too high.
For example, lets say the pub quotes you the "average" click-thru rate of
2%. That means for every 1,000 impressions they counted, you've gotten 20
clicks. But let's say really another 350 impressions of your ad are being
viewed from caches on AOL and behind corporte firewalls. Your real
click-thru ratio is 1.48%
I wrote a story for Advertising Age about an initiative by proxy makers to
try and help ( http://adage.com/interactive/articles/19970407/article1.html
).
2) Vague Terminolgy:
The other issue is that because "impressions" are poorly defined,
click-thru ratios again vary. Some sites and ad management software count
impressions as the number of times the actual ad file was served, while
others count just the number of times the page's source code contains
instructions for the ad to be served. Because again of caching, as well as
surfers with graphics turned off and other issues, the latter definition is
going to be a much higher figure. With different definitions of
"impressions" you're going to get different click-thru ratios based on a
forumal of clicks/impressions.
The net-net is don't trust click-thru rates, or at least grill the site
intensely on these and the other problems other folks mentioned with these
rates.
Rick E. Bruner rick_at_bruner.net
Bruner Communications Web site RSN ;^)
