Note that the redundancies we (and the Rules of Normalization) talk about are NOT data redundancies, but rather table/field redundancies.
For example, let's say that you had separate tables, one for each client, and they all had the exact same table fields.
That would be a "de-Normalized" structure. It instead should be one table, with a table field to indicate the client.
Or, let's say that you have one table, by similarly repeating type fields. Let's say that you had 12 revenue tables, one for each month, all with the same data fields.
That also would be a "de-Normalized" structure. It instead should be one table, with a table field to indicate each month.
One dead giveaway that you have a de-normalized structure is if you have to constantly add more data tables just to accommodate new data, when nothing basic has changed about the database, i.e. you need a new data table because you have a new client. If your database it designed well, you would be able to add new data (i.e. clients or months) without having to create new data tables which closely resemble the structure of existing tables that you already have now. In a well-designed database, you seldom have to create/add new data tables, unless something fundamentally changes about what your database is supposed to be doing. You should be able to create all the data tables you need right at the beginning in development, and not have need to constantly add new tables.