I was wondering if someone can help.
For a university assignment I have been asked to propose a marketing research method & survey etc for a company.
I have been set an online takeaway website to do. We are asked in the survey to find out about food habits and the takeaway market. Survey has various likert scale questions to use in subsequent segmentation done through cluster analysis. We do not have to collect the data and do the analysis, just explain HOW it would be done and how its results can help the clients.

We have been asked for a sample of n=2000 of UK population and part of this must be users of the takeaway service in order to "augment the research"

We have been told to use an "online panel" for the population and do 1600 from here. This is not an accurate sampling frame so QUOTA sampling will be used. (Quota sampling using age and gender to represent the proportions in the UK population)

The 400 customers technically have a sampling frame as the company will have a database with all the customers so technically I could use a probability method such as stratified. But maybe its best to use quota in order for the two samples to use the same method? Either way i will be sampling the respondents to represent the UK population in terms of proportions of age and gender.

I am meant to then use these results to segment the data and create different clusters. i am happy and understand cluster analysis - that part is fine. But is it OK two combine the two samples (1600 from panel and 400 from the company) to make a sample n=2000. Technically both are sub groups of total uk population and will be asked the exact same survey.

Any thoughts? Can i just combine and then do my cluster analysis? Or is there something more complex needed ????