It is hard to produce a successful product if there isn't data to inform its development process or to back up its effectiveness. However, collecting the data is only a part of the data analysis. Data must be processed, cleaned, and interpreted properly to be of use.
The first step of data analysis is defining what kind of data needs to be collected. Does it involve people? Is it going to be more quantitative or qualitative? Are there restrictions that need to be considered during collection? Ensuring the requirements of a process are clearly defined before beginning will save time and resources regardless of the specific methods used during data analysis.
The second step is, of course, gathering the data. This can be done in a myriad of ways including surveys, interviews, using sensors, usability testing and so on. What matters most is that the data is not influenced during collection. Bias must be avoided as much as possible, otherwise the data may become unusable or inaccurate. The specifics of how to avoid bias during research are outlined in other articles.
After collecting the data, you process it. Raw data can be difficult to use due to sheer quantity, unfortunate collection circumstances, or simply misunderstanding hand scrawled notes. Organizing data into a table or sorting it into distinct categories can go a long way in making data more functional. Again, the method of processing depends on the data collected.
Even after data is processed, it is often necessary to tidy it up to prevent later problems. Incomplete entries, errors, or duplicate entries should be dealt with during this stage, in addition to noticing possible or accidental bias and complete misinterpretation. Data cleaning is done by various methods such as comparing against acceptable thresholds, using spell checkers to standardize words, and double checking all research steps and research results.
Once data is cleaned, data interpretation can begin. Look for patterns and relationships between the various data points, and determine what they mean. Remember that there is a difference between what the data is actually saying and what you want the data to say. If given enough time and enough denial, data can be misinterpreted or manipulated. Thus, it is imperative that an objective perspective is maintained. It is also likely that more data will be requested, so do not get too attached to a particular interpretation.
Additional Resources:
“Data Analysis.” Wikipedia, Wikimedia Foundation, 12 July 2018, en.wikipedia.org/wiki/Data_analysis.