Signal-Based Outbound: What It Is and How We Run It

Most outbound fails before the first email is written. The list came from a static filter, the copy was written for a persona instead of a moment, and the sequence went to 5,000 people who had no reason to answer this week. Signal-based outbound fixes the order of operations: find the companies where something just changed, and reach them while the change still matters.

This is the method we run for B2B SaaS and AI companies and for B2B service businesses such as MSPs, IT and telecom providers, and staffing firms. Here is what it means in practice, which signals are worth acting on, and how the pieces fit together.

What "signal-based" actually means

A signal is a dated, observable event that makes a company more likely to buy right now: a new sales hire, a funding round, a leadership change, a tool added or dropped, a new office. The point is not that these companies are a good fit in the abstract. It is that they have a reason to care today.

A static ICP list answers "who could buy." Signals answer "who is in motion." You still need the ICP filter, but you apply it to the companies in motion instead of the other way around. That inverts the list: instead of 5,000 contacts you touch once a quarter, you work 150 to 400 contacts a week whose first line can reference something real.

The signals we score

Not every signal deserves a sequence. We weight each one by how directly it connects to the problem the client solves and by how fast it decays.

• Hiring signals. Job posts for roles the client's product replaces or supports. A company hiring its first SDR is a strong signal for sales tooling. A company hiring a second IT admin is a strong signal for an MSP.
• Funding and leadership changes. New budget and new mandates. These decay in roughly 60 to 90 days, so they go to the front of the queue.
• Tech installs and removals. Tracked through enrichment providers. A company that just dropped a competitor, or added a tool yours integrates with, is worth a specific opener.
• Growth and location signals. A new office, a new region, a new service line on the website, or headcount that moved noticeably in six months.
• Engagement signals. Company-level website visits, content downloads, event attendance. Weak on their own for cold outreach, strong when stacked on one of the above.
• Verification signals. For service companies we create our own. An AI calling agent dials the main line, confirms who owns the decision, and asks what they use today. That turns a guess into a fact before the first email goes out.

Each signal gets a score and a freshness window. Stacked signals move a company to the top of the week's list. Stale ones drop it out.

How the system fits together

Signal-based outbound is not a Clay table. It is a set of layers that hand off to each other, and the order matters.

1. Signal capture. Clay and Apollo workflows pull the raw events on a schedule, deduplicate them, and enrich the account and the contacts.
2. Scoring and routing. Each account gets a score. Above the threshold it moves into the campaign that matches its signal, with the right titles attached.
3. Infrastructure. A pool of sending domains and mailboxes, warmed and monitored, so volume never rides on the client's primary domain. We run more than 70 sending domains across client systems and treat deliverability as a weekly job, not a setup task. The details are on our cold email infrastructure page.
4. Copy. One opener per signal, not one per persona. The first line names the event, the second connects it to a problem, the third asks a question. Follow-ups add proof instead of repeating the ask.
5. AI calling. For business lines and service companies, an AI calling agent confirms the decision-maker, asks two qualifying questions, and books the meeting or routes a callback.
6. CRM handoff. Replies, meetings and dispositions land in HubSpot or GoHighLevel with the signal attached, so the client's team can see why the conversation exists.

Teams that start with copy and bolt signals on later end up with great emails sent in the wrong week.

Where teams get it wrong

• Too many signals at launch. Every signal needs its own opener and its own scoring rule. Start with two or three.
• No freshness window. A funding round from 2024 is not a signal. Set an expiry for every event type.
• Signals without a list filter. In motion is not the same as in market. Keep the ICP filter on company size, business model and geography.
• Measuring sends. Volume is deliberately lower than a spray campaign because relevance is doing the work. Measure qualified meetings held, not emails sent.

What to expect in the first 90 days

• Weeks 1 to 3. Audit, ICP and signal definition, domain and mailbox build, warmup, list and scoring workflows, CRM wiring. No sends yet.
• Weeks 4 to 8. First campaigns on the two or three strongest signals, daily volume ramped as warmup allows, copy variants tested per signal rather than per subject line.
• Weeks 9 to 12. Scoring thresholds adjusted from reply data, weak signals dropped, AI calling added where main-line verification lifts qualification.

What we do on client builds

We build and run the whole system as a managed engagement: the signal workflows, the infrastructure, the copy, the calling layer and the CRM routing, with weekly reporting. Clients see the signals, the sends, the replies and the meetings in their own CRM. The engagement models are in our pricing section, and the list layer is described in more depth on our list building page.

FAQ

How is signal-based outbound different from intent data?

Intent data is one category of signal, usually third-party topic research scores. Signal-based outbound uses any dated event you can observe and act on, including hiring, funding, tech changes and verification calls you run yourself. The method is about ordering outreach by timing, whatever the source.

How many signals should a campaign use?

Start with two or three strong ones tied directly to the problem you solve. Every extra signal adds a copy variant and a scoring rule to maintain. Add one only when the current signals are producing replies and you can write a distinct opener for the new one.

Does this work for service companies without big tech stacks?

Yes, with different signals. Hiring, new locations, licensing changes and verification calls to the main line matter more than tech installs. The calling layer is usually what makes it work, because you confirm the decision-maker before the sequence starts.