Data normalization is a term often associated with databases. In this sense, the database designer is looking to create a cohesive set of data that reduces data redundancy and increases data integrity. Sometimes your data might be presented to the client in a database, so this makes perfect sense. If you are instead creating a document in a text format or other non-database format, you still want to apply those same principles of reducing redundancy and increasing integrity. For example, data from disparate sources that describes the same event can be consolidated to reduce confusion.

Also, the report is likely to be read by a variety of audiences. This might include board members, end users, and technical administrators. They all need to be able to read through and understand your findings and recommendations. So, you need to target your reports to account for these differences. There might be sections for executive summaries for those who only need a high-level understanding. There might be links to more technical information that end users might find too confusing. Essentially, you want to normalize data in the report to make it as clear to the target audience as possible, all while minimizing extraneous information that just contributes to the noise.