Intent-based marketing works when signals inform audience decisions through a deliberate planning step, not when raw signals flow straight into an ad platform. The signal tells you something changed at an account. It does not tell you the account fits your offer, that the timing window is still open, that no exclusion applies, or that the record is even eligible for the ad platform you plan to use. LinkedIn's own framing of intent-based marketing describes intent as an input to account and audience decisions, not a guarantee of matching or advertising results (LinkedIn). The bridge between the two is audience planning, and it is a human-owned process supported by data infrastructure, not an automation you switch on.
This article lays out that bridge as a repeatable five-gate workflow, walks through an illustrative example with exact numbers, and covers the eligibility and latency realities that determine whether a signal-informed audience is worth activating in your existing ad tool.
What a signal tells you, and what it does not
Research and engagement signals may indicate interest; business events add context. For example, a company may advertise engineering roles, announce funding or appear in relevant news. Those events can inform a targeting hypothesis without showing that it is researching your product. Useful, but narrow. A signal does not carry fit information. A 12-person agency and a 4,000-person enterprise can emit the identical hiring signal while only one of them belongs anywhere near your audience.
Signals also decay. A funding event from eleven months ago describes a company that may have already deployed much of that money. If your thesis is "recently funded companies invest in tooling," the signal window is part of the definition, not an afterthought. For a deeper treatment of signal types and how they map to buying behavior, see our guide to buying intent and business signals and the intent data glossary entry.
The practical consequence: a signal feed is a candidate stream, and an audience is the output of filtering that stream through conditions you own.
The audience planning bridge: five gates
Every signal-sourced record should pass five gates before it reaches an ad platform. Skip one and you risk wasting spend on poor-fit accounts or creating a compliance problem.
| Gate | Question it answers | Typical failure if skipped |
|---|---|---|
| 1. Fit conditions | Does the account match ICP criteria (size, industry, geography, technology)? | Budget spent reaching companies that could never buy |
| 2. Signal window | Is the signal recent enough for your thesis (for example, 90 days)? | Messaging built on stale context |
| 3. Exclusions | Is it a current customer, open opportunity, competitor, or do-not-target account? | Ads shown to existing customers or active deals |
| 4. Owner review | Has a named person approved the candidate set? | Nobody accountable when the audience underperforms |
| 5. Platform eligibility | Does the data meet the ad platform's matching and data-use rules? | Rejected uploads or policy violations |
Treat this as a worksheet, not a philosophy. Each gate has an owner, a written condition, and a pass rate you can inspect. If nearly all candidates pass fit, check how the source list was built. That can be appropriate for a prequalified account list; the pass rate alone does not prove the rules are too loose.
An illustrative worked example
The numbers below are an illustrative example of the funnel shape, not a benchmark or a claim about typical results.
A RevOps team monitors hiring and funding signals for mid-market software companies. Over 30 days, signals surface on 200 companies.
- Gate 1, fit: 80 of the 200 meet ICP conditions on employee count, industry and geography. 120 fall out immediately. In this example the majority of the volume exits at the fit gate, which is exactly the work fit conditions exist to do.
- Gate 2, window: All 80 signals fall within the 90-day window, so none are removed here.
- Gate 3, exclusions: 20 of the 80 are excluded as current customers, open opportunities or accounts on a suppression list. 80 minus 20 leaves 60 candidates.
- Gate 4, review: The demand gen owner reviews the 60, spot-checks provenance on a sample, and approves the set.
- Gate 5, eligibility: The team confirms the data source and identifiers meet the target platform's rules before export.
Keep account counts separate from audience-member counts. The 60 approved companies may map to zero, one or many eligible people depending on the destination and audience type. Track company resolution, eligible identifiers and matched members separately; use the platform's actual size estimate to assess delivery eligibility.
Platform eligibility is a real gate, not paperwork
Google Customer Match illustrates why gate 5 exists. Google requires qualifying first-party customer information for Customer Match; a purchased contact list does not meet that collection rule (Google Ads policies). Separate account research from upload eligibility. Buying or enriching a contact record does not make it first-party information collected directly from that customer. Confirm account eligibility and the other current requirements before activation.
This is where data provenance stops being an abstract governance topic. If you cannot answer "where did this record come from and what are we permitted to do with it," the record should not go into a platform upload, however strong the signal looks.
Latency, tradeoffs and failure handling
Three tradeoffs shape how tight to run the workflow.
Freshness versus review. Owner review adds latency. A weekly review cadence means a signal can be six days old before the audience updates. For many B2B ad motions that is acceptable, because display and social campaigns run over weeks, not hours. If your thesis depends on same-day reaction, reconsider whether paid media is the right channel for that signal at all.
Precision versus audience size. Tighter fit conditions shrink the candidate pool, and after match-rate shrinkage the audience may fall below the platform's minimum size. When that happens, widen the signal window before you loosen fit conditions. A 120-day window with strict fit will often be a better trade than a 90-day window with sloppy fit, because fit errors waste spend on every impression.
Automation versus accountability. Autonomous agents can handle discovery, enrichment, scoring and exclusion tagging continuously, which is exactly what a data layer like AstroFabric does: agents watch signals, enrich candidate records, apply scoring against your fit conditions, and stage platform-ready audience data with suppression applied. What stays with the team is approval and activation. AstroFabric delivers matched or custom audience data into your existing ad tooling; it does not launch or optimize campaigns, and the campaign itself continues to live where your media team already works.
Failure handling matters as much as the happy path. When an upload is rejected for eligibility reasons, quarantine the batch rather than retrying with the same data. When match rates drop sharply between refreshes, audit identifier quality before blaming the platform. When an excluded account slips through, trace the exclusion list sync and fix the source, because a one-off manual removal hides the systemic gap.
FAQs
Does an intent signal mean an account is ready to buy? No. A signal is an observation that can prompt a relevance hypothesis; its predictive value needs validation. Fit conditions, the signal window and exclusions still decide whether the account belongs in an audience, and even then the audience is a targeting hypothesis, not a verdict on buying stage.
Can I upload a purchased contact list to Google Customer Match? No. Purchased contacts do not satisfy Google's first-party collection requirement. Review the Customer Match policy and the other eligibility requirements before uploading customer information.
Why is my matched audience smaller than my candidate list? For a list of eligible individual identifiers, unmatched records can reduce the matched count. A company list is different: one company can correspond to many people. Compare counts at the same level, check the platform's size estimates and investigate unmatched identifiers before changing targeting rules.
Put the bridge in place
If your signals currently live in a spreadsheet and your audiences are rebuilt by hand each month, the gap is infrastructure, not effort. AstroFabric's agents can run the discovery, enrichment, scoring and suppression steps continuously and deliver platform-ready audience data into the tools you already use, with your team keeping approval and activation. Start with the data infrastructure for prospecting overview, or sign up and define your first fit conditions and exclusion list this week.
Sources
Every playbook on this blog ships as a runnable mission.
Open a workspace and the playbook library is waiting - describe the outcome and the agents carry it end to end, on your plan's monthly credits.