Why severity rankers?
A workflow produces findings; a severity ranker decides which ones matter most. Rankers are how you encode your notion of priority so the top of the list is the part worth reading first.
The problem they solve
Two findings are rarely equal. An unauthenticated RCE and a verbose error message might both be "findings," but only one is worth waking someone up for. Left unordered, real bugs drown in noise.
A ranker gives the model your rules for triage, so it can order findings by expected impact - the way an experienced bug-bounty triager would.
How ranking fits in
After the workflow finishes and findings are de-duplicated, the ranker runs over the canonical findings and assigns each a priority. It can do a full rerank of everything, or append newly found findings into an already-ranked list. The result is the ordered findings list you see on the scan.
Why write your own
Off-the-shelf severity scales don't know your program. A ranker lets you say things like "auth bypasses on money-moving endpoints outrank everything else" or "deprioritize anything requiring local access," so the ordering reflects real payout potential for your targets.
Next: how to create a ranker.