Qualitative analysis really begins in data collection. Survey researchers often notice that multiple respondents use the same language to answer questions. At the same time that responses are being recorded verbatim --- whether by electronic audio, video or in writing --- data collectors can speed analysis by taking note of these consistencies, though at this point they need not connect them to individual respondents. The same process may be helpful as recordings or pencil-and-paper verbatims are transcribed into files for analysis.
The first step in direct analysis of responses is to identify phrases and concepts that are meaningful to your research question. Working with hardcopy transcriptions, this can be as easy as marking phrases with a highlighter. In computer files, you may be able to perform searches through larger numbers of respondent records in a database or throughout an aggregated text file. This process is similar to the memos made during data collection and transcription, and may be informed by those notes, but should be left in place in the record for correlation with other variables such as respondent age, race or marital status.
To prepare for correlation and counting, assign codes to similar segments of the record and group them into categories. You may have created some codes and categories in advance, based on your research question, and others may develop as you analyze responses. For instance, if you're studying the secondary school experience of students at small liberal arts colleges, you may have already set up categories of college prep, arts and vocational courses. When a respondent reports having done a year of self-directed independent study, however, you'll have to create a new code to cover it. Create a separate set of codes for the characteristics across your sample with which you will correlate your findings.
Go back through the data a fourth time to count the occurrences of each code and clusters of codes. For instance, what majors have the students who did independent study in secondary school chosen in college, or are they clustered in the "undecided" category for college majors?
Qualitative research is especially often used for pilot studies, to identify the categories in which more directly quantitative research can be conducted with larger samples. Even when the final study is to be qualitative, it can be made more efficient by starting with a smaller sample to develop the survey instrument, identify variables and perhaps even have data collectors precode responses rather than keep memos.