(no subject)

From: <owner-online-ads_at_o-a.com>
Date: Fri, 28 May 1999 16:40:51 -0500 (CDT)

From: Tim Lee <TLee_at_webcmo.com>
Subject: Re: ONLINE-ADS>> New Product Marketing Woes

SPENCER QUINN WROTE:
>Wouldn't an advertiser rather reach 300,000 people they know
>fit the target demographic versus reaching 1,000,000 in a
>crap shoot?

If we know prices and CTRs of both targeted and untargeted
banners the becomes much easier. But we don't know. The
things we know are:
1) targeted banners (by demos) are more effective than
untargeted banners
2) targeted banners are not as effective as what we expect.
(Please read our recent report about online targeting at
http://www.WebCMO.com.)

More importantly, targeting is not as easy as we think. To
this specific targeting question, we need to ask two
questions:
1) How do we determine the target demographics?
2) Can we find a web site that EXACTLY fits the target demographics?

For most of us, we use statistics to determine the targeting
demographics. Suppose after conducting a survey about
people's interests in your product, you have found the age
and income distributions of those who are interested in your
product:
a. Income distribution

< $25,000 12%
$25,000-$34,999 20%
$35,000-$49,999 44%
$50,000-$74,999 12%
>$75,000 12%

b. Age distribution

< 18 11%
18-25 20%
26-35 45%
36-50 12%
> 50 12%

The statistical result shows that about half of the people
who are interested in your product have incomes between
$35,000 and $50,000 or ages between 26 and 35.

The question is: can you conclude that your customers are
those who are between 26 and 35 and have income between
$35,000 and $50,000.

Suppose the joint distribution of the people who are
interested in your product is as follows:
<18 18-25 26-35 36-50 >50 Sum
< $25,000 100 200 500 100 100 11%
$25,000-$34,999 100 100 1500 100 100 20%
$35,000-$49,999 700 1400 (700) 700 700 45%
$50,000-$74,999 100 100 700 100 100 12%
>$75,000 100 100 700 100 100 12%
Sum 12% 20% 44% 12% 12% 100%

The result shows that if you select the web site (that
exactly fits your target demos ) to target your customers,
you can only reach about 8% of your total customers and that
combined age and income group is even not the most desirable
targeting group.

There are three serious drawbacks in our current targeting
practice:
1) we use predetermined demographic groups to determine the
target
2) we use bivariate analysis to determine the target
3) we determine the target only based on "statistically
significant".

As I always said, bivariate analysis is far from enough to
address segmentation and targeting issues. The above example
is an excellent demonstration. If we use the predetermined
demo groups to determine the target, we won't be able to
target our customers. We can only target to a group of
people who are more likely to be our customers.

More importantly, "statistically significant" does not mean
we are able to target our customers. My research experiences
have shown that even we use multivariate analysis (usually
Logistic Discriminant Analysis), we divide the population
into two groups--customer vs. non-customer (instead of
predetermined demo groups) and we find "significances" for
many demo variables used in targeting, we are still unable
to improve the targeting accuracy significantly.

So I have raised an important and practical question: What
is the TARGETABLE CONDITION?

For example, if the penetration is 20% and we are able to
increase the targeting accuracy to 40% by using a
sophisticated analysis, can we say that our customers are
targetable?

If we don't target, what shall we do?

Regards,


Tim Lee
Editor
Journal of Web Marketing Research
http://www.WebCMO.com
Innovations in web marketing research & strategies
Participate in "Email Marketing Strategy" survey to receive your EXCLUSIVE
report
http://www.WebCMO.com/mis/survey.htm

========================================================================
------------------------------------------------------------------------
        This week's Online Advertising Discussion List sponsors:
               Iconocast, Flycast, and Streammedia

*** We're saying "HASTA LA VISTA BABY" to Boring Conferences! ***
Register now for Web Attack! -- The Ultimate Internet Marketing Event
June 17-18, San Francisco http://www.iconocast.com/webattack/tenagra

                                ---

WEB ADVERTISING THAT MAKES YOU A HERO. Our Web experts optimize your
buy across 800 sites. All for less than what you'd normally pay for
placement on one. Call it superhuman. WWW.FLYCAST.COM/OA

                                ---

    Sure, you can continue to rub two sticks together.But that's
      not gonna make the Advertising Gods real happy. Check out
    http://www.radicalmail.com for advertising that really works
------------------------------------------------------------------------
========================================================================



Received on Fri May 28 1999 - 20:35:55 CDT