What we think
Sales Hub's new automation is only as good as the definition work you did before you turned it on.
Turn any one of these on before you've defined what you're actually optimizing for, and you get expensive noise instead of leverage. Casey's read on Rule Insights for Building Scores was to question whether a rule should be scoring at all, not just how many points it deserves — and her advice on Discover and Score High-Impact Events runs the same instinct the other direction: decide your sales handoff threshold first, then judge whether HubSpot's suggested point values make sense against it, never the reverse. Chris's warning on Buying Groups applies the same discipline to a deal's buying committee instead of a score: let the AI build the group from job titles — "the least reliable field in the CRM" — and you've based your Account Map on the one field in your portal you can't trust.
The prospecting agent take makes the payoff of doing that work first concrete: a client who got sales and marketing in the same room to define buying signals before switching the agent on sent ten targeted companies a signal and got four responses back, a rate most outbound teams would take at a hundred. And Casey's line on the Lead Scoring Public API — "what pushed them over" — names the same gap from the reporting side: even once you can see the score, you usually can't see the moment, and the honest read is that the last action rarely tells the real story anyway.
Across scoring, prospecting and account structure alike, it is one instinct wearing several product names: the automation gets good exactly when someone does the definition work before flipping the switch, not after.
One honest gap in this argument: forecasting itself — the pipeline number, deal-stage math — has not landed as its own Sales Hub feature take yet. When it does, it belongs here.
