Parking Service Empirical Comparison

I'm new here and hope you won't consider my request for technical advice too wacky....anyway, here goes:

Background:

I have dormant domain names parked at services like SEDO.com who create and host advertising filled webpages to monetize the traffic from the domains' legacy links and type-ins. Traffic is delivered to parking providers by pointing the domain registrar's DNS to the parking provider's Nameserver.

Parking providers have proprietary templates and keyword driven algorithms to create webpages and match the advertising displayed to the domain name and its legacy traffic. Advertising displayed may be varied during the course of the day or week depending on what offers are available.

These algorithms are secret and very smart - they will experiment and learn from experience to optimise revenue performance over time. The variety of the algorithms and advertising sources means that it is practically impossible to compare competing parking site providers other than empirically.

Objective:
I wish to statistically compare performance of alternative parking providers by "rotating" or randomly selecting the parking site that my domain points to. Sort of as if the DNS was load balancing between the parking sites.

Because the advertising available varies over time, rotation would ideally be effected transaction by transaction. If this is impractical, then batching of transactions for rotation would be still acceptable although the larger the batch the less statistically accurate the comparison (the coarser the time slice, the less likely the adverising on offer remains identical).

Provided transactions are randomly rotated between the competing providers, the providers' own revenue reporting is sufficient for me to compare their performance.

Methods
If nameserver rotation is impractical for any reason then alternatives could be considered such as redirecting traffic for "mydomain.com" to a rotation of different derivative domain names registered for the purpose : eg "mydomain-01.com", "mydomain-02.com", "mydomain-03.com"
mapping respectively to provider01, provider02 and provider03. Because the domain name itself is a parameter for the advertising algorithm, it is important to minimise any bias introduced.

Strategy:
What approach is recommended?
Are there any substantive implementation issues or concerns?

Phew! Any ideas?
Steve