Re: Unverifiable Transactions / Cookie draft

Dwight Merriman:
>
>I'd like to bring up one another (small) point.
>
>Virtually all banner advertising on the web relies on redirects for
>click-throughs so that click rates can be measured.  This means that the
>redirection to the advertiser's web site will always be an unverifiable
>transaction.

The redirection won't be unverifiable if the script which does the counting
and redirection to the page is on the target site already.

>So, when a user visits an advertiser's site directly, cookie assignment on
>the home/jump page is possible, whereas when the user visits the page via
>an advertisement, it will not be possible.

Yes, if the script is on the site which served the banner.  But I don't see
this as a big problem.

>Designers of web sites (at least the large percentage who will advertise on
>the web) will have to take into account that cookie assignments on their
>home page may fail a large percentage of the time.  If they wish to measure
>number of unique visitors to their site, they will get a highly inaccurate
>reading since often multiple cookies will be assignied to a single user
>before one "sticks".

Some careful design can work around this: suppress assignment of cookie if
the referer field shows that the user comes from the click-through script.

Note however that higly inaccurare readings are almost guaranteed anyway
because many people disable cookies, or have browsers which do not support
them.

>
>This is not a major flaw, but it is inelegant, and I just want to make sure
>everyone has considered this.
>
>Dwight

Koen.