Formulas for test estimation?

Question

I have 60 pages which i need to test for compatibility in 112 devices and OS's, are there any basic formulas to estimate these kind of Testing based projects.

How should I give an estimation for this? are there any best practices?

Answer 1

Just like you would give an estimation for anything else, either it is something you can time or it is something that has high uncertainty and is inestimable.

If all the pages are relative the same you could maybe calculate it like this:

• PageTime = Time (in minutes or seconds) the testing compatibility of one average page
• DeviceTime = Time the setup of an average device
• TotalTime = (PageTime * 60 * 112 ) + (DeviceTime * 112)

Now on average someone works 5-6 hours per day, as you need slack and toilet breaks. Since you cannot test compatibility 8 hours straight, during testing you cannot focus you need short breaks here and there.

Now you should be able to give a man-day estimation. Now during execution, I would log the time per page per device or something to see if I am on track. If not update the estimation :)

Also, I wonder if you cannot skip some devices that have a low risk because they are not used by a large audience. Start with the high-risk devices and pages first.

Answer 2

I think the best formula is to avoid estimates whenever possible.

Never give an estimate based on insufficient information, which is basically nearly always the case. Certainly in this case it is true.

Also estimate what? Time? Cost? People?

Sometimes when pushed and it is short term and no unknown I use the "formula":

• Pick 2 from above 3 of time, cost, people
• Triple what seems reasonable