NONE: RE: ONLINE-ADS>> Value of Test Banners?
RE: ONLINE-ADS>> Value of Test Banners?
Mark J. Welch, Esq. (markwelch_at_ca-probate.com)
Fri, 10 Oct 1997 09:57:45 -0700
Glen wrote:
>> I've explained that it would be extremely helpful
>> to run a "statistically significant" test of our two to three top
>> performing creatives...
And then Mike wrote:
> You're right! From my perspective, it makes total sense to offer test
> banners to potential advertisers. It allows me to prove that our audience,
> technology, and service is as good as we say it is. Our competitors have
> bigger names than us, and we sometimes think that advertisers would rather
> rush like lemmings to them, rather than taking the time to evaluate
> something slightly smaller, but of demonstrably higher value. A test run is
> a great way to go.
I will jump in and disagree. Glen suggested testing a "statistically
significant" number of banners with 2 or 3 different creatives. At
Web Advertising '97, Ali Partovi (of LinkExchange) suggested
that measuring the response rate for a particular banner and
comparing it to other creatives isn't reliable until about 100,000
adviews for each banner (and he gave some very specific
examples of results of some campaigns). Putting these two
together, it appears that Glen is asking for 200,000 to 300,000
"test" adviews, apparently for free. If those are targetted
adviews, that may represent a huge chunk of available
inventory. On the other hand, it may take fewer total
impressions to generate statistically significant results
on a single site, as contrasted with an ad network like
LinkExchange.
If the "test" campaign will be based on a much smaller number
of adviews (such as 5,000 or 10,000), I think such small samples
would make any comparison of CTRs for two creatives very
suspect (e.g. a single spurious click-through could skew the
results).
There are some alternatives. For example, the site could
agree that payment during a test campaign should be based
on click-throughs, or could run a test campaign at a fixed
price until some fixed number of click-throughs is generated
(e.g. 100 or 1,000). (This assumes that the advertiser will
measure the campaign's performance based on click-
through, which in my opinion is a very poor surrogate for
"success" in most ad campaigns).
Another alternative would be to require the advertiser to
pay for a campaign, and offer to give the advertiser some
extra exposure from unsold inventory. But if the vendor
pays for 50,000 adviews and gets 50,000 free, he will
expect it in the future also.
A very devious alternative, if the vendor's name and logo
do not appear in the ad, would be to test the creative
with links that don't jump to the vendor site, but to a
related-theme not-for-profit site or to a "coming soon"
announcement. Thus you can measure CTR but the
advertiser doesn't get free benefits.
I certainly don't think an advertiser should be giving away
adviews just to help a vendor test and compare several
different creatives. When I test and compare two different
direct mail campaigns, the post office doesn't give me free
postage while I do so. (On the other hand, if the web site
is suggesting some changes to creative to increase CTR,
or is suggesting that the advertiser expand a campaign
to some new locations, it MIGHT be good sales strategy
to offer a small, short test run to show that the results
will justify spending more ad dollars on the site.)
I'm not aware of ANY media that gives away free "test" ad
campaigns. Sometimes I am offered special low rates for
a test campaign in a local newspaper, and quite honestly
this just leads me to expect that rate for ALL advertising
that I do. And no newspapers or TV stations will even
offer to guarantee results (although some TV stations do
accept ads, usually late-night, paid on a per-inquiry or
per-sale basis).
On the other hand, as I've mentioned before, I am
unhappy with the minimum campaign size for some
vendors. For example, the New York Times (online)
requires a minimum ad buy of $12,500 (which equates
to about 175,000 impressions using their targetted
ad rates). When I pointed out that I wanted to use
their impressive targetting features to reach only
people over age 50 in about 20 California zip codes,
and that this would presumably not generate more
than a few thousand adviews per month, they simply
had no alternative to offer, and they lost a sale.
-- Mark J. Welch, Esq. (510) 462-8483 http://www.ca-probate.com/
-- This message is NOT legal advice, and is NOT confidential.
-- Web Site Banner Ads and Web Counters: http://www.markwelch.com/
-- Small Office / Home Office Consumer: http://www.sohoconsumer.com/
----------------------------------------------------------------------
This week's Online Advertising Discussion List sponsor:
The Internet Advertising Report provides daily columns on every aspect
of Web advertising and marketing. It is updated daily and available in
a weekly newsletter. Subscribe at: http://www.internetnews.com/IAR
-----------------------------------------------------------------------
Online Advertising Discussion List To Unsubscribe send UNSUBSCRIBE
http://www.o-a.com/ to online-ads-request_at_o-a.com
