What are your top tips for collaborating with database developers and stakeholders?

The business analysis (and requirements gathering) phase of any project owes it's success heavily to a team of people who are effectively empathizing with their users and applying that persona(s) toward objective user experience workflows as well as product development. One of the most effective ways to ensure that your project is successful is to involve the project's stakeholders early in the product development lifecycle with iterative user feedback sessions and opportunities for the client to refine their expectations as the project progresses.

Involving your end-users or stakeholders in database-heavy projects is especially beneficial when incorporating this process into a CMA (content management application). These types of applications usually evolve into web sites or applications which involve heavy user engagement especially as it pertains to user-generated content. For example, Instagram is a very popular and simple CMA as the app has grown to see it's user adoption in a very niche use-case (in terms of content) -- most users use the platform for simple image and/or video sharing as part of their personal stories or business marketing themes. However, complex CMAs are often built with CMS's (systems) which can vary to the extent that their modeled closely to the DB design.

Normalization is very important and especially stressed in formal texts when approaching database development from a traditional, holistic perspecitve. In practice, when storing user-data for a web application (or CMA), having a strict reliance on normalization past the 3rd normal form can prove unnecessarily complex and potentially costly when requirements change in the future as your user-based application's needs evolve. If, however, you're developing a new schema or DB model for more traditional usage (ex. data warehouse, data lake, long-term user data retention, etc), adherence to best practices with an Oracle or PostgreSQL RDBMS is usually best. NoSQL DBs should be utilized mainly for web apps which will require little data analysis.

My comments are mostly drawn from my experience building Vignette-based CMAs throughout the 2000s for companies such as NY Life, AIG, Prudential, FBI, BET, Time, and People(.com) Magazine. While the Vignette platform follows a recommended workflow for CMA and database development, there is a great deal of flexibility in creating the web applications and workflows that arose from the these projects... all being different from the last (other than my fairly consistent use of the shell scripting language, TCL, as it is the native language for the platform template development on most Vignette CMS projects).