The reasoning graph recalculates every day.
Approximately 30% of underlying signals refresh daily; the remainder refresh on longer cycles (closer to quarterly for slow-moving signals like income estimates). The output customers query is regenerated daily, incorporating all newly arrived data.
What makes Watt's data valuable if competitors can buy the same raw data?
The moat is in the joining, not the sourcing — unifying disparate data categories inside a single reasoning graph.
How does Watt handle conflicting data from different sources?
Each signal is assigned a confidence weight. When sources conflict, the highest-confidence source wins.