1: Before you start generating the scripts first find out the top 3 to 5 scenarios that will be used to test the site. I formulate these into what I refer to as click-paths.
* Load home page
* Enter username and password and click Login (Will need a set of usernames/passwords)
* Enter text in search field and click search (will need items to search for in data file)
* Click on random returned selection
2: Next: Create the scripts in the tool of choice. You'll also need to find out what requirements are available for the application so that you can code what think times or wait times you'll need for the scripts. Knowing this determines how you'll run the tests, if you're management truly wants ‘Stress Testing’ then this equates to running the virtual users simultaneously (user without wait times). Concurrent users = users on the site but have wait times coded in the scripts.
Here are some other common questions that you can ask:
1- What are the goals of the tests:
a. To see where the bottlenecks of the site exist?
b. To see how many users the AUT will support?
c. To see how many transactions (scenarios per/hr the site can handle)
d. To explore or determine at what level (load tests with realistic users) does the site’s performance begin to degrade. i.e. page response times goes above 10 seconds, etc…
2- How many users, given with the top 5 scenarios, will be on your site in an hours’ time frame?
3- Also determine if you’re load tool supports caching. I’ve seen instances where improperly configured load tests or scripts over utilize the bandwidth and image servers simply because the virtual users didn’t cache the site. On every page the virtual users were requesting every single image or object and the test results as you can imagine would be way off base.
IMHO before you do any of this make sure you have a solid understanding of basic performance testing methodologies and you have them implemented. Then you can follow any of the listed suggestions.
In theory and with a lot of these tools anyone can record and playback a script, being able to explain what you did and why you did it is just as important to the test as running the test itself (basically the results can be meaningless unless you can explain or justify what you did, why you did it and how you did it).
Just my 2 cents (can I get some change?)
But in the end doesn't it all come down to BEER? Beer is the ultimate answer to all questions in the universe so yes the answer to your question is BEER.