Ideally, we invite a nationally representative sample and these people will be self-selected within your screening questions. For example, they might have to be buyers of a certain product category or accepting of a specific brand. It is usually best not to constrain your demographics from the start and allow self-selection unless we know from previous research and client data (or just plain old common sense) that target audience likely fits a particular demographic. Nappy buyers, for example, are probably going to be the parents of young children.

However, it is best not to make assumptions sometimes. You might assume that buyers of Jimmy Choo shoes are predominately wealthy, so may want to target higher end of the income scale. However, a young female on a modest salary living with parents may have more disposable income than a family with a mortgage, kids and associated expenses. So you may miss a chunk of your potential market. If we allow an erroneous assumption to drive sample design, the results will inevitably be skewed.

To help further, ask yourself some simple questions:

1. Do you want to find out what proportion of people in the general population use your product?
   In which case, we will need to invite a nationally representative (nat rep) sample to your survey with no quotas. In the survey, the sample will then reflect your target audience (its composition is likely to be different to UK nat rep population). We can even do nat rep within a subgroup of the population, for example among 25-70 year olds.
   
   Example: Invite a nat rep sample of a 50-50% gender split to find buyers of power tools.
   Completed interviews may be 80% male and 20% female.
2. Is it a niche product that you just want research opinions for rather than market statistics?
   You may need to target known users and age groups etc. For example, people who have used a London taxi in the last 6 months.
3. Need both? Or need set quotas (e.g. a minimum number of parents with kids under 10)
   In which case, begin with nat rep as stage 1 sampling and get your penetration from this before targeting or quotas kick in.
   
   Note: Nat rep is normally age and gender based. It is not advisable to interlock more parameters so always discuss this with your researcher at the beginning.

Just be clear about your sample design from the start, as it is difficult to rebalance it later.