Some things I've consider from a strategy point of view in various situations:
Could you move to a 'test based on change' approach? Could you monitor the industry for software upgrades (of for example browsers) and new screens (with different dimensions and characteristics) and thus test in response to change as opposed to periodic testing? Clearly you do monitor change to a degree because you have to upgrade tests to reflect these changes.
How closely do you monitor the system to try and discover issues as they occur. Are all web server and application logs automatically monitored? You can see indications of issues, users taking too long to perform a task, transactions abandoned, etc.
Could you move the testing requirement on to your clients? How easy could you make it for your clients to tell you about issues and problems? And how do you thank them? Entry into competition or a credit perhaps?
Do you actively try to measure the impact the issues you have discovered in the past have had? For example characterising issues by:
- percentage clients impacted
- length of time existed
- loss of revenue
- estimated type of issue: outage, limited access methods, loss of
functionality, lack of usability, etc.
It can be interesting and knowing the estimated cost of issues might be useful as part of your review.
Clearly every situation is unique and I know nothing of what you do so just some thoughts that I hope may stimulate further thoughts.