We have recently started an online advertisement campaign and we are tracking the response rate to each advertisement posted within our website and our electronic newsletters.  I am hoping to get some suggestions about the best way to calculate the response rate from a specific ad in our newsletter.  Since we have started the campaign, we have sent two newsletters that include this advertisement.  The first one included the ad as part of the regular bi-weekly distribution (which includes multiple topics) and the second newsletter was sent about a week later to further promote the advertisement (this was the only topic in the newsletter).    We have about 13,000 newsletter subscribers -- since the advertisement was included in two newsletters, would it be more accurate to calculate the response rate using a total of 26,000 ad views, or would it be better to use 13,000 since we sent the ad to the same people for each distribution?

Any feedback would be greatly appreciated! 

On 18:46:20 6 May 2005 TRI_CMN wrote:
[...]
>We have about 13,000
>newsletter subscribers -- since the advertisement was
>included in two newsletters, would it be more accurate to
>calculate the response rate using a total of 26,000 ad
>views, or would it be better to use 13,000 since we sent
>the ad to the same people for each distribution?
>
>Any feedback would be greatly appreciated! 

26,000.

(I'm resisting the temptation to say, "of course")

If you were paying per OTS (opportunity to see), or sending direct mail (where each additional recipient would incur additional cost), that's the way you'd plan it.

Hope that makes sense.

Consider looking at linear programming (see: http://en.wikipedia.org/wiki/Linear_programming, and http://carbon.cudenver.edu/~hgreenbe/courseware/LPshort/intro.html) for ways to model the effectiveness of your campaign -- although this may be a sledgehammer/walnut solution.

All good suggestions - but the factors influencing your response rates are too variable to be of any use when you're dealing with two separate mailings (even though the recipient list is the same). Creative and timing have already been mentioned by Mark. I'd ask myself 'what do I want to measure - and why?'
Online marketeers get very excited by the 'transpareny' or accountability of the campaign results they can deliver. In most cases response this data is meaningless unless you tie it to a commercial metric - such as you cost per customer or lead/sale figure. Only then can you make an ROI justification to do more of same and grab a larger slice of the offline media spend!
Chris
On 14:30:33 12 May 2005 MarkQuinn-Newall wrote:
>It would be appropriate to monitor each of the two
>responses separately, even though the ad is the same and
>potentially the audience is the same (see below for
>caveat)
>
>The data received on the effectiveness of the first ad
>being placed amongst a general newsletter with multiple
>topics should show a very different response profile to
>that of the second Ad display occurring within a single
>dedicated related topic email.
>
>If you mix the total response into one data point, you
>will lose the potential benefit of understanding the
>potentially very different levels of response within each
>type of email.
>
>Another issue when considering the response rates
>statistics would be to have a clear understanding of just
>how many of the 13k newsletter emails are being opened,
>and over what time period.
>
>You will probably have a general understanding of what
>kind of opening rates are occurring, however this general
>understanding can be skewed greatly by the issue of a user
>response to the subject line content.
>
>As I understand it one email was a general newsletter
>email with multiple topics which may be opened by a wide
>variety of potential readers, however the second email
>which is a single topic would presumably have lower appeal
>to potential viewers.
>
>Another issue will be on what day / time of day you are
>sending out the email, which will have a potential effect
>on the response rates.
>
>In short (not the above) analyze both 13k mailings
>separately.