For most deep-tech companies, timing decides whether an outbound message gets a reply at all. A robotics or applied AI company selling into a market with almost no search volume and no established category name can't rely on a buyer typing a query and finding them. The reliable opening is contacting the right account in the week something changed for them, before the moment passes and the account goes quiet again.
When a market has a name, such as warehouse management software or CRM, buyers search for it, comparison content exists, and demand can be captured as it appears. A physical AI company selling a perception system for forklifts, or a sensing hardware maker selling into industrial inspection, often sells into a category the buyer has no word for, so search volume sits near zero and the work is noticing demand as it forms. A newly hired VP of Operations, a facility expansion, a safety incident reported publicly, a competitor's funding round: each is a dated, public fact that changes what an account is likely to evaluate this month, and each one closes. Reach the account inside that window and the message lands as relevant. Reach it three months later and the same message reads as a cold pitch to someone who has already moved on or already decided.
The trigger events that hold up in deep-tech markets are often ones a software sales playbook never lists: a building permit for a new site, a published tender, a design freeze announced at a conference, or a regulatory authorisation with a start date. Each one carries a date and a public record, which is what separates a trigger event from intent data, the inference of interest from page visits and downloads that a recipient can neither see nor check.
You can test this on your own list. Send the same sequence to two groups of accounts, one filtered to a dated trigger in the last thirty days and one without, and compare reply rates after a few weeks. Because the copy is identical in both groups, what you're measuring is whether the account had a reason to be listening that week.
A signal is only usable if it is real and checkable. A vague line like “we spotted the right moment to reach out” without naming what happened, when, and where it can be verified reads as a guess dressed up as a signal, and this audience, often technical people who have built systems like this themselves, notices immediately. A trigger worth acting on has a dated public source: a press release, a job posting, a filing, a permit, something a sceptical recipient could check in thirty seconds. Anything short of that is an assumption, and a program built on assumptions wastes as much as one built on the wrong market size.
For a revenue team, this changes what gets built first. Before writing a sequence, build the watch, which is what our Method page describes: a defined list of signal types worth tracking for this specific account list, a dated source for each one, and a rule for how fast the team moves once a signal fires, because a trigger's value decays fast and a signal caught on day one and acted on in week six has already lost most of what made it useful. The words in the email matter less than most teams assume, and whether the email arrives inside the window a real event created matters more.
An account that hasn't changed has no reason to answer, and one that just changed has a reason for a few weeks at most, so the number that matters is the share of your sends that land inside a real window. List the three most reliable dated, public signal types for your own category, such as a hire, a funding event, an expansion or a regulatory change, and check how many of your active accounts have fired one of them in the last thirty days. That count is closer to your real near-term pipeline than anything sitting in a static account list.