Hoping this is the right place to ask this... I'm working with a group who wants to conduct survey research to evaluate projects. They're thinking about having an initial, overall question that would be required and making all of the other questions optional. This seems like a bad idea to me because if everyone answered only the initial question, they wouldn't have any usable/actionable data. Also, those who do choose to continue would likely be only those who had very positive or very negative experiences. I can't find any instance where this was used in practice. I'd love to get some input...
It will all depend on the kind of survey. It's not the same to use a survey to find out reasons for abandonment of a product/service than a survey looking to research on a product to be launched. It's not the same an online survey than an "on location" survey (for example, a survey ran on an UX lab, or hallmark testing).
However, as different as surveys can be, they share something in common: you want to extract some knowledge off them. Thus, you need to define what do you want to know. This is the most basic and yet the most problematic question, specially when you have many stakeholders. Anyways, let's suppose everybody agrees on the exact knowledge you want to have. Once defined, you need to define the minimum amount of questions needed to answer the problem / knowledge you want to acquire.
Take this example:
- Why did you stop using Service X?
- Is there something we can do in order to make you use the service again?
- Your Age
- Your Phone
- Your marital status?
Nice, huh? Short, easy to understand. Of course, I'm sure you'll see how wrong it is at first glimpse. Yet, this is more or less a REAL survey I was asked to fill (and obviously didn't).
In this particular case, you would only need the first question. And I'm deeply sorry if whoever did this can't figure how to extract useful information from such important event (abandonment or service), but I can't imagine why should I teach them how to improve a service they clearly don't want to improve, how is my age relevant and more important, if I'm married or not. Unless they want to ask me for a date!
So, as you can see, if you can make good questions, 1 and only one question would suffice! (yet, 1 question is a highly unlikely scenario unless your needs are really limited and specific)
The Optional Scenario
And then we come to the optional questions. While one or two optional questions could be useful, you need to ask yourself: do I need the knowledge these optional questions will provide? Yes or No? If Yes... why are they optional? If no... why am I alienating my users?
Again, something to remember: it's not the same to ask your user why s/he is leaving our service (thus, she's probably pissed enough not to want to fill anything, let alone optional and unneeded bits of information) than to make a survey about customer satisfaction and how to improve it. Or online vs face-to-face surveys.
Just for reference, take a look to the image below from 5 Common Mistakes Made When Writing User Surveys
The best path to follow is... keep the questions to the minimum, no optional questions. Either you need the knowledge or not. Then act accordingly.