You have just joined a company. The system is called “DataFlow” and the documentation is three years out of date. You are asked to propose a caching strategy in your first week. You have no context.
The available context: the system is described as a “data aggregation and insights platform.” That tells you the archetype. A data aggregation platform is Data Intelligence (Archetype 6). Data Intelligence systems are read-heavy for analytics, batch-oriented for processing, eventually consistent at the aggregate level, and heavily dependent on the relationship between the batch and streaming layers. The most common failure modes in Data Intelligence systems are schema drift between the pipeline stages and latency amplification in the query tier.
You now have a starting point. The likely infrastructure patterns, dominant tradeoffs, and first failure modes to investigate are now identifiable. You have not read a single line of code.
Archetypes are not precise — no real system fits cleanly into one category. But they are useful precisely because they are inexact: an approximate map is better than no map when you are navigating unfamiliar territory.