Most teams that try signal-based prospecting follow the same arc. They subscribe to a funding feed, set up job-change alerts, maybe buy an intent data package. Within a month reps are getting forty alerts a day, half of them about companies that were never going to buy, and the channel quietly goes back to being ignored.

The signals weren't the problem. What was missing was a way to decide which signals matter for your deals, how much each one is worth, how fast it goes stale, and what a rep is supposed to do when one fires. This post walks through that framework: inventory, scoring with decay, compound signals, response workflows, and a quarterly review that makes the whole thing more accurate over time.

What counts as a signal

A signal is any observable change that makes an account more likely to buy now than it was last month. Firmographics tell you whether a company could buy (size, industry, region). Signals tell you whether the timing is right. You need both: a perfect signal at a company outside your ICP is still a bad lead.

Signals fall into a handful of families. They differ a lot in where the data comes from and how long they stay useful:

The freshness column is a starting assumption, not a rule. Your own data will tell you the real numbers (step 6).

Step 1: Take inventory

List every signal that could plausibly precede a purchase of your product. Don't filter yet. Useful inputs:

• Your last 30 to 50 closed-won deals. For each, look at what was happening at the account in the 90 days before the opportunity was created.

• Interviews with recent customers. Ask "what changed that made you start looking?" and write down the literal answer.

• Closed-lost deals that died on timing ("not a priority this year"). These tell you which signals were missing.

• Your reps. They usually know that a new CFO means a vendor review. They just haven't written it down.

You'll end up with 15 to 30 candidate signals. Most of them won't survive the next step, and that's the point.

Step 2: Score signals against your own wins

A signal is only useful if it shows up more often before wins than it does across your target market in general. So compare two rates for each signal:

• Win rate: the share of closed-won accounts that had the signal in the 90 days before the opportunity opened.

• Base rate: the share of all accounts on your target list that had the same signal in a comparable 90-day window.

Divide the first by the second and you get lift. Here's a worked example with made-up numbers for a sales tool:

In this example, "uses HubSpot" shows up in more than half of wins, which looks impressive until you notice it's common across the whole list. It's a fit filter, not a timing signal. Office expansion has no lift at all and should be dropped. The new VP signal is rarer but far more predictive.

With a small number of deals these figures are noisy, so treat anything under roughly 20 closed-won examples as a hypothesis. Also watch the base rate. It has to come from the same data source you'll use for monitoring, or you're comparing different things.

Step 3: Turn lift into points, and let the points decay

Give each surviving signal a base score roughly in line with its lift, then set a decay period so old signals fade out. A simple version:

Linear decay is easy to explain to reps: current_score = base_score × (1 - days_since_signal / decay_period), floored at zero. A new VP who started 30 days ago is worth 30 × (1 - 30/90) = 20 points today, and nothing after day 90.

Then apply ICP fit. Multiply the summed signal score by a fit factor (for example 1.0 for a strong fit, 0.5 for a partial fit, 0 for outside ICP). That stops a flood of signals from a company you can't sell to from reaching the top of anyone's list.

Step 4: Prioritize compound signals

Single signals are weak on their own. A funding round could mean anything. A funding round, plus three SDR roles posted, plus a new VP Sales, describes a company that's about to build an outbound team and is going to buy tools to do it. When signals pointing at the same need show up together, they tell you far more than any one of them.

The additive score from step 3 already rewards stacking, but it helps to spell out tiers so reps don't have to interpret numbers:

• Watchlist: one mid-strength signal at a good-fit account. Add to a light nurture or marketing audience. No rep time.

• Priority: two related signals within their freshness windows, or one strong people-change signal. A rep reaches out this week.

• Act now: any first-party hand-raise (demo request, repeat pricing visits from a known account), or three stacked signals. Same-day response.

Pay attention to whether signals are related. "New VP Sales" and "hiring SDRs" point at the same project. "New VP Sales" and "new office in Denver" probably don't, and shouldn't count as a compound.

Step 5: Build a response workflow for each tier

A signal nobody acts on is just a notification. Each tier needs a defined path from detection to outreach:

1. Detect. Scheduled checks against your target account list, daily for fast signals, weekly for slow ones.

2. Enrich. When an account crosses a threshold, find the right contacts automatically: the persona the signal points to, with a verified email and ideally a phone number.

3. Route. Create or update the record in your CRM with the signal, its date and the score attached, and assign an owner.

4. Reach out. Enroll in a sequence written for that signal, within the SLA for the tier.

5. Log. Store which signal triggered the outreach on the contact or opportunity, so you can measure it later.

The message should make sense given the signal without making the prospect feel watched. A few examples:

• New leader: "Congrats on the new role. Most sales leaders I talk to spend their first quarter figuring out which parts of the outbound machine are actually working. Happy to share what we see across teams your size."

• Hiring cluster: "Saw you're adding several SDRs. The first month for new reps usually comes down to whether their lists are ready on day one."

• Tech adoption: "Noticed you're running Outreach now. Teams that pair it with [category] usually do it to fix [specific problem]."

For first-party and third-party intent specifically (website visits, category research), the rules are different: you usually shouldn't mention the signal at all. We cover that in detail in our guide to using intent data in B2B sales, including response-time targets for inbound hand-raisers.

Step 6: Review every quarter

This step is what separates a framework from a pile of alerts. Once a quarter, pull every opportunity created in the period and tag it with the signal that triggered outreach (or "none"). Then look at:

• Meetings per signal fired. How often each signal led to a conversation.

• Opportunities and win rate by signal. A signal that books meetings but never closes is worse than it looks.

• Time from signal to opportunity. This is how you replace the guessed decay periods with real ones.

• Ignored alerts. If reps skip a signal type consistently, either it's noise or the playbook for it is bad.

Recalculate lift, adjust base scores and decay periods, drop what isn't working and test one or two new signals. The first quarter's model is a rough guess. After a few reviews it's fitted to your own pipeline, which a competitor can't copy by buying the same data feed.

The infrastructure behind it

You need three pieces, and you can build them from separate tools or combine them.

Signal sources

Funding and company events (Crunchbase, PredictLeads), job postings and hiring patterns (PredictLeads, job boards, careers pages), technographics (BuiltWith, HG Insights), people changes (LinkedIn Sales Navigator and people data providers), third-party intent (Bombora, 6sense, Demandbase, G2), and your own first-party data from the website, forms and product.

Detection and scoring

Something has to check those sources against your account list on a schedule, calculate scores and decide when a threshold is crossed. Many teams do this in a warehouse or spreadsheet. In Databar you can keep your target accounts in a table, add enrichment columns for the signals you track from 160+ data providers, run them on a schedule, and use the AI Researcher for signals that don't live in a structured feed (for example "has this company announced an expansion into Europe in the last 60 days?"). From there, the score itself can be calculated wherever your team already keeps account scores: a spreadsheet, your warehouse, or scoring properties in your CRM. See our walkthrough of sales trigger alert systems for more on the alerting side.

Response automation

When an account crosses a threshold: find contacts with a waterfall enrichment for verified emails, sync the record to HubSpot or Salesforce, send a webhook to post the alert where reps work, and enroll contacts in the right sequence. Flows let you chain those steps visually so they run the same way every time.

Mistakes that sink signal programs

• Monitoring everything at once. Start with three to five signals that showed real lift. Add more after your first quarterly review.

• No fit filter. Signals from companies outside your ICP waste the most rep time because they look urgent.

• Scores that never decay. A funding round from last year keeps an account at the top of the list long after it matters.

• Generic outreach. If the email would read the same without the signal, you've paid for data and gotten nothing for it.

• Quitting after a month. The first version will be mediocre. The review cycle is how it gets good.

Common questions

What's the difference between signal-based prospecting and intent data?

Intent data is one family of signals: behavioral evidence that an account is researching a topic, on your site (first-party) or across other sites (third-party). Signal-based prospecting also includes company events, hiring, people changes and technology changes. Most strong programs combine them, with intent data usually carrying the timing and the other signals carrying the reason to reach out.

How many signals should I start with?

Three to five, picked from the lift analysis in step 2. That's enough to fill a pipeline and few enough that reps learn the playbook for each one.

How long before this shows results?

Replies to signal-based outreach show up within the first few weeks. Whether the signals actually lead to revenue takes at least one full sales cycle to judge, which is why the quarterly review matters more than the first month's numbers. If you want a scoring model that also covers fit and segmentation, our post on lead scoring and account segmentation is a good companion.

Start with one signal

Pick the signal your reps already swear by, check it against your last 30 wins, and build the full workflow for that one signal before adding a second. If you want to run the monitoring and enrichment in one place, Databar paid plans start at $99/month, you only pay for results, and there's a 14-day trial with the full product.