In devising your analytical approach, a key issue to consider is the stage at which you should produce your initial template. You could do this after carrying out preliminary coding on just one transcript, or you could wait until you had done every transcript. In most studies, it is produced at a point between these two extremes. The danger of producing the initial template early is that you may find it interferes with your attempts to approach fresh transcripts with an open mind. You can become over-sensitised to material that ‘fits’ relatively neatly into the template and neglect those which are difficult for the template to encompass. However the advantage of doing this early is that the initial template can help you to focus on areas of greatest relevance to your research question(s) and avoid becoming bogged down in what may be redundant or repetitious coding.
Your decision about how soon you produce the initial template needs to take two things into account. Firstly, and crucially, you need to think about your overall methodological position. If your approach strongly emphasises the need to remain open to the data and avoid pre-suppositions, for example, in phenomenological studies, then you should hold back from creating the initial template until you have carried out preliminary coding on all transcripts, or at least a substantial proportion of them. If your methodology is driven by the need to answer quite specific theoretical or practical questions, as in many realist-oriented evaluation studies, then you are probably best advised to produce the initial template at a fairly early stage. A good rule of thumb is that you have probably reached an appropriate point when your preliminary coding is no longer producing many new themes that are distinctly different from those identified previously.
Secondly, you need to consider the practical constraints on your project in terms of time and resources. If you have a large number of interviews and a tight deadline to meet, then moving quickly to the initial template stage may be the only way to get it done. Of course, such practical concerns should not override the methodological integrity of your research. However, this is most likely to be a problem where you have not thought through your design carefully enough, for example, recruiting too many participants for a phenomenological study where you have very limited time available.
This is largely a matter of personal preference and the needs of particular projects, but it is often best not to attempt to produce too comprehensive a template initially. The danger is that if you invest too much time and effort in creating an elaborate initial template, you may be reluctant to make substantial changes to it later, for example, seeking excuses to avoid making changes to coding that would ‘mess up’ a nicely organised initial template.
Some of the computer packages available to assist with qualitative analysis can help you avoid this tendency, because they make it very easy to re-arrange the structure of the template as you go along. The "node explorer" feature in NVivo is a good example of this.
When producing your initial template, it is important to cover the main thematic areas emerging from your preliminary analysis, but not to be too concerned with fine distinctions at third, fourth or even lower levels of the coding hierarchy. The exception is where you feel that a very narrow and specific theme is especially important in helping you make sense of the data. In such a case, you should include a more detailed thematic structure for the relevant part of the initial template.