NONE: Re: Refining Definations
Re: Refining Definations
Cliff Kurtzman (cliff.kurtzman_at_tenagra.com)
Wed, 21 Aug 1996 22:02:22 -0500
Maxine Most <maxmost_at_mpath.com> wrote:
> How about addressing the issue of Ads delivered versus Ads actually viewed?
> Just because the page is downloaded and viewed does not mean the ad
> downloaded before the user clicked off the page or srolled down beyond it or
> never scrolled down to where the Ad is located. As the WWW is taken more and
> more seriously by the Advertising Industry this is a measure that will be
> crucial.
And Donna Dolezal Zelzer (djz_at_efn.org) replied:
>Is there really a way this can be measured? Once the page and the ads on it
>are sent, how can we tell what the viewer does, unless they actually click
>on one of the ads? <snip>
I think you could construct a fairly simple test where you varied the
placement of a single banner on a long page. Place the banner at the top,
and see what the click through rate is. Then place the banner at the
bottom several screens down and see what the click through rate is.
I would think that it is probably a fair _first order_ assumption that the
banner will have a similar ratio of clicks per real person-views ratio no
matter where it is placed. It is true that people may be slightly more
inclined to click a banner at the bottom of a page because then they are
done reading the content, but I would guess this effect is small.
If the ratio of clicks per real person-views is constant, you will probably
see a continued reduction in actual click throughs as you put the banner
lower on the page. You can be reasonably assurred that most impressions
will be viewed by a person if the banner is at the top of the page, but not
if it is lower down. This is because some people will not scroll down
enough to see the banner or will click on a link they see above the banner.
In such cases, an impression will still register on the server, but a
human may never see the banner.
So lets plug some example numbers in and see how this works. Lets look at
a banner on a page that is viewed 100,000 times.
Case 1 the banner is placed at the top of the page and loads say 80,000
times (some people won't be loading graphics). You get 1,600 click
throughs on the banner (telling you that 2% of banner viewers are clicking
on your banner. The number of banner clicks can be measured through use of
a cgi script or by making the banner an image map).
Case 2 the same banner is placed at the bottom of the page (several screens
down) and loads 75,000 times (less than 80,000 because some people will hit
the stop button before the whole page loads). Now suppose you measure 750
click throughs on the banner. But you know, from Case 1, that 2% of the
people seeing your banner are clicking on it. From this, you can infer
that 37,500 graphic loading people are making it to the bottom of your page
out of the 75,000 graphic loading people that completely loaded the page.
This in turn tells you that, to first order, 50% of your page viewers are
making it to the bottom.
By "to first order" I mean that there are certainly a number of factors
that can introduce error into this analysis, but in most cases I would
expect that these factors would be relatively small in effect compared to
the primary factor you are measuring, namely how many people are really
making it to the bottom of your page.
--Cliff
Clifford R. Kurtzman
The Tenagra Corporation
http://arganet.tenagra.com/
713/480-6300
-------------------------------------------------------------------------
This week's Online Advertising Discussion List sponsor:
Let Web Surfers Know It's Time To Visit Your Site
Web Watches from Lynx Marketing, http://www.logotime.com/web.html
Earn $$$ for publicizing your own web site
--------------------------------------------------------------------------
