When you implement your evaluation depends on the evaluation questions/issues that you’re trying to answer. If you know nothing about your audience, a front-end evaluation and literature review should be the first thing you do as you plan your project. If you’re interested in making changes to an ongoing program, you want to collect and analyze evaluation data in a timely manner so that it can inform your decision-making. For more about the timing of evaluation during the project development process, see Evaluation Stages & Types under the section Some Basics.
What about pre/post, time series or experimental evaluation designs? If it is important to know how your audience changes as a result of your program, you need to develop a quasi-experimental or experimental design whereby you collect data from your audience multiple times over the course of your project and, in the case of experimental design, collect data simultaneously from a control or other treatment group. In those cases you will need to schedule (preferably at the beginning of the planning process) times for collecting data at regular intervals and then keep to that schedule.