The sharpest feature signal is not adoption of what they have, it is repeated reach for what they do not. Five or more attempts to use a premium, gated feature inside thirty days is a customer telling you, in their own clicks, exactly what they want to buy (Saber). An upgrade-gate click is the highest-intent signal in the entire taxonomy because the customer initiated it; there is no guesswork about fit, no inference about intent, just a person trying to do a thing your plan does not let them do yet.
The numbers behind this are the most persuasive in the whole field. A customer reaching for a gated feature is, in effect, a product-qualified lead, and a free trial that surfaces PQLs converts at 2.8x the rate of one that does not, because a PQL has actually used the product rather than downloaded an ebook (UserGuiding). The contrast with marketing-sourced leads is brutal: only about thirteen percent of MQLs become sales-qualified, and only six percent of those convert to paying, whereas one well-run PQL programme defined its trigger as trial users who finished onboarding, used the product on five-plus days, and had three-plus team members active, and closed those at 35 percent versus 4 percent on marketing leads (Gainsight). Intent expressed against the ceiling is simply worth more.
So instrument the gate itself. Every time a user hits a tier boundary, that is a logged event with a name, a count, and a timestamp, the same way a saturation alert is. Treat reached for an enterprise-only feature as a first-class signal alongside seats at ninety percent, and weight it heavily. Premium-feature pursuit and seat saturation are the two signals you instrument first; everything else is supporting evidence. For how this reach shapes the rest of the funnel, see lead activation and the automation of trigger detection in upsell and cross-sell triggers run by AI agents.