If you mean by "ebb and flow" the natural tendence of web response times to vary in a regular way with the time of day, you're right. The delivered end-user performance will go up and go down in the same proportion.
What we've observed, interestingly enough, is that the intra-event timings in a typical Rich Internet Application (RIA) test playback tend to remain relatively constant. It seems as if the "startup" time is what varies the most. Once the connection is made (assuming the waits in the script are not so long as to exceed the "keep alive" time of the HTTP connection) then the only thing that variances internal to the application shows is variation due to response time at the server.
That's why we make such a big thing out of having the abiity to threshold inter-RIA-test timings: this is the way to separate the normal ebb and flow of web performance from the abnormal behavior that may be exhibited by your server.
Of course, THAT behavior is where the payoff is in RIA monitoring.
eValid Monitoring Team