# Thread: How many testers does it take?

1. ## How many testers does it take?

I'm curious to know if anyone can tell me the recommended method of determining the number of testers for a product? and any sources for this info. I need to present a case for a QA department to our board of directors

------------------

2. ## Re: How many testers does it take?

It depends on the size and complexity of the application under test, the level/types of testing desired, and the qualifications/experience of the testing staff. There is no set method to come up with a test team size. I mostly do it through past experience.

------------------

3. ## Re: How many testers does it take?

I agree with Testy. There is not really a discriminant axis that you can use in all cases to determine the number of testers you will need. Likewise, there is not one discriminant axis for all products because it can depend on the experience of your testers and the overall complexity of the product as well as any time constraints you might be laboring under. Your better bet is talking about the labor rate for testing that will be required given the size and complexity of the product relative to any time-to-market considerations you are dealing with. If you can work out the labor rate needed (particularly if you have historical data to fall back on), you can at least justify a certain number without it being something that is just a pure (and uneducated) guess.

Now, just as an exercise in ticking some people off here who hate when I use "mathematical" approaches, at one company I was at, it was critical for the nature of the product that I had a certain staff level. I needed to present this to management and this particular management was very keen on having numbers and formulas to back up an argument. So what I did was show that my team size that would be necessary to complete the test cycles would be based at the rate at which requests arrived. Now, by "requests" I mean "requests for test" (i.e., new builds) and things like "defects fixed that need to be checked". In other words, anything that added to the labor rate and thus the time. What you can probably see is that I just used the very common method of queuing theory to show how, under reasonable assumptions for defect find rates and the number of builds we would have to test, I would need a staff of a certain level. (This was all somewhat independent of the complexity of the product, although, granted, the defect find rate did depend on that.)

Now, I bring this up only because (a) this went over really well, and (b) my particular management in that particular company required a strictly quantified approach and would not accept a "guess" - even if it was based on experience, and (c) I knew the team size I ideally needed to meet the time-to-market deadline relative to the test cycles I knew had to be done. So if (1) that kind of approach is demanded of you, or (2) you absolutely know the team size you need, or are pretty darn sure of it, and you need to justify it, this appoarch might be one thing to consider.

------------------

4. ## Re: How many testers does it take?

While there is certainly no rule or formula that works everywhere, I have found, like Jeff, that I get MUCH further if I come up with a statisical basis for my requests. In my original staffing, I estimated # of test sets to be written, chose a fairly conservative time multiplier for that, estimated # of major releases & multiplied by # test sets and again a very conservative time multiplier, estimated # of patch releases & time to test etc. Add in meetings, defect administration etc. I used MS Project to document those estimates and attached a doc that clearly defined all of my assumptions. It went over very well with the then-management, and I got what I asked for.

Since then, I collect actual hours to complete these tasks. My job is pretty easy now because I have a pretty good basis of information to pull from. For example, a "major" release averages around 600 man-hours to system test, a "patch" release around 200, etc. I have good estimates based on actuals for test planning for projects of various sizes. We do a yearly project and release schedule, and my resource planning is pretty proven.

What I am really getting at ---- your estimates right now are going to be pretty shaky, but keep track of your actuals, and next time it won't be too hard.

------------------

5. ## Re: How many testers does it take?

When I managed a Technical Writing group, I helped justify the size of my team by doing a quick phone survey of technical writing groups in other companies. Managers of the other groups were helpful when I explained why I was doing the research, and it wasn't confidential information. I wrote up the justification including how our ratio compared (favorably) to other companies.

It would be interesting to know the ratio of testers to developers in our different organizations. I'll start with an estimate (new group, and I'm not sure how many developers we're supporting yet):
SW/FW 8 testers to 40 developers
HW 4 testers to 20 developers
(do I see a pattern?)
Heather

------------------

6. ## Re: How many testers does it take?

Hi,
I too agree with the Testy.
It depends upon the experience on specific Domain, past experience.

May be some pointers can be taken from Estimation. How complex the product is.

Total No. of test cases to be tested etc.

cheers.

------------------
Dinesh

7. ## Re: How many testers does it take?

I recently posted this question and received quite a few replies, but nothing really definitive until I posed the question to a Microsoft Project Manager. Now I realize some people think of MS as the evil empire, but...here is the reply:

The rule of thumb that is used here is as follows:

Product groups: 2/1
IT Group (development of internal software tools): 1/1

------------------

8. ## Re: How many testers does it take?

<BLOCKQUOTE><font size="1" face="Verdana, Arial, Helvetica">quote:</font><HR>Originally posted by fredorison:
...nothing really definitive until I posed the question to a Microsoft Project Manager. Now I realize some people think of MS as the evil empire, ...<HR></BLOCKQUOTE>

I do not think Microsoft is the Evil Empire, so no bias here on that score, but keep in mind that these figures were (I would assume) given in the context of one company: Microsoft. That may work for them. It may even work across all projects for them. But that does not, a priori, mean that it will work for other companies or other project types.

Just something to consider. Also in your post you asked for a "method" for determining the number. What you were told (unless you are leaving out some details) is not the "method" by which that "rule of thumb" was derived. You were just given the results. Again, just something to consider. Also consider that what you were given is a ratio (apparently) and you did not ask for a ratio in your original post; you just asked for the number of testers. A good (but very simplistic) article on this can be found here.

Personally I find using ratios instead of work product numbers can be more misleading than helpful, but that is just my experience.

------------------

9. ## Re: How many testers does it take?

I think ratio should be different in different phases of project. Something like this Dev/Tester.

Requierment Analysis 10/1
Design 10/1
Development 10/1
Testing 5/10
Implementation 5/5

------------------

10. ## Re: How many testers does it take?

"Personally I find using ratios instead of work product numbers can be more misleading than helpful, but that is just my experience."

I agree, even though I started the discussion on ratios. When planning, you look at resources needed. However, when you have a well-functioning group and you are asked to justify the number of people (or lose some), it sometimes helps to show that your group follows or betters the norm in other companies. Or, if your group is struggling and you need additional people, it may help to show that you are understaffed compared to other companies. If you find that your group has a much higher ratio of testers to developers than other companies, maybe it is an indicator that processes could be improved. Just another way of looking a things.
Heather

------------------

Page 1 of 3 123 Last

#### Posting Permissions

• You may not post new threads
• You may not post replies
• You may not post attachments
• You may not edit your posts
•

vBulletin Optimisation provided by vB Optimise v2.6.0 Beta 4 (Pro) - vBulletin Mods & Addons Copyright © 2016 DragonByte Technologies Ltd.