Contents
Signal-based selling means timing outreach to events that suggest an account is ready to buy, rather than working through a static list. A funding round, a new senior hire, a change in technology or a public filing can each be a signal. The GTM engineering work is building the system that spots those events, attaches them to the right accounts and people, and triggers the right play.
This guide uses one detailed public example, the setup at security company depthfirst, to show how the parts fit together.
What counts as a buying signal?
Clay chief executive Kareem Amin describes signals, in a post from 24 June 2026, as job changes, funding rounds and other time-sensitive triggers that say when to act. They sit alongside two other kinds of data: identity data, such as firmographics and contact details, and engagement data, such as CRM records and call transcripts.
The useful test is simple. A signal should change what you do this week. If it does not change who you contact, when or with what message, it is a data point, not a signal.
How does depthfirst build outbound on signals?
Mark Hardy is Head of Revenue Operations at depthfirst, an AI security company, and its entire RevOps team. Clay profiled his setup on 14 September 2026.
He starts with two target groups: companies still running older application security tools such as Snyk and Veracode, and companies whose engineers use AI code editors such as Cursor, on the theory that they produce more code than their security teams can review.
On top of that he tracks signals including:
1/ Annual 10-K filings, where public companies disclose cyber risk.
2/ Security incidents in the last six to twelve months.
3/ Breaches and their financial impact.
Hardy's method has four steps: find a signal, structure it, push it to the CRM, then build a segment of the people who care about it. Ideas for new signals often come from sales call transcripts, where customers say what worries them.
Why not just email when the signal fires?
Because some signals make blunt outreach backfire. Emailing a security chief right after a breach to suggest they buy a security product reads as insensitive, as Clay's write-up puts it.
Hardy's answer is to warm the segment first. depthfirst's researchers publish findings on vulnerabilities in widely used open-source libraries, and that research runs as LinkedIn ads aimed at the target segments. When a security leader clicks or follows, that engagement is recorded as a new signal on the account. Only then does outbound begin, and the message speaks to a problem every security team has, not the buyer's own incident.
Ads and email then run together. The people entering an email sequence are pushed to LinkedIn as an ad audience, refreshed weekly, so they see the ads while the emails land. Hardy has also found LinkedIn outreach works best coming from the founders rather than a sales rep.
What does signal timing do to the numbers?
An illustrative comparison. A team has 2,000 target accounts and capacity for 300 personalised outreaches a month. Working the list in order, it reaches every account once every seven months or so, mostly at random moments.
If signals flag 150 accounts in a given month, the team can spend half its capacity on accounts where something has just changed and the other half on the rest. The volume of work is the same. What changes is that half of it lands when the account has a reason to listen.
What breaks signal-based systems?
The signals themselves are rarely the problem. The plumbing is.
Hardy's own example: when he used Claude Code to compare Clay's enriched data with Salesforce, it found a field that should have been receiving data was populated on only 10% of records. The sync for that column was broken. After he fixed it, the field was filled on 89% of records.
That pattern is common. A signal that never reaches the CRM, or reaches the wrong record, triggers nothing. Three checks catch most of it:
1/ Audit field fill rates regularly. A sudden drop usually means a broken sync, not a quiet market.
2/ Match signals to the right entity. A funding round belongs to a company. A job change belongs to a person. Mixing them up sends the right message to the wrong place.
3/ Retire signals that do not convert. Track which signals lead to meetings, and stop paying for the ones that do not.
How do you start?
Hardy's advice is not about tools at all. "The first step is you've got to download something or get a license. You've got to try it," he told Clay.
In practice, start small: pick one signal that clearly matters to your buyers, one segment, and one play. Measure meetings, not activity. Add the second signal only when the first one is working.
A note on Peach Data
Peach Data builds go-to-market engineering systems for sales and marketing teams: the data foundation, the workflows and the agents behind outbound and inbound. We started as a verified data business, so every build starts with the data layer. peachdata.co.uk
