Targeting is the allocation decision underneath every go-to-market motion: which accounts get a rep's hour, which get an ad impression, which get a nurture track, which get left alone. Sales calls it account selection, marketing calls it audience definition, and both are the same question asked of the same data. Data infrastructure for targeting is the set of filled, verified, scored records that lets the question be answered consistently across channels, plus the delivery that gets the answer into the CRM views, sequencers and ad accounts where the allocation is actually made.
The term is having a moment because the channels have converged on the same input. Ad platforms accept lists of companies and contacts and match them to members; sequencers accept the same lists; CRM routing runs on the same fields. A team that can produce one well-targeted list can point it at every channel at once. A team that cannot ends up with a sales target list, a marketing audience and a CRM tier that disagree about who matters. The product page for this job is Data Infrastructure for Targeting; the ad-side specifics are in AI agents for audiences.
What data infrastructure for targeting means
Good targeting has three components on every row: a fit judgment (does this account match the customers we win and keep), a timing judgment (is something happening that makes now better than next quarter), and an exclusion judgment (is this a customer, a competitor, an open opportunity or someone who asked not to be contacted). The infrastructure produces all three as fields with provenance, and keeps them current.
That sounds like scoring, and scoring is the visible part. The invisible part is the data the scores run on. A fit model that weights employee count, industry and technology is worthless on a database where those fields are empty for a third of accounts, and misleading where they are stale. Infrastructure for targeting therefore starts with enrichment and verification and ends with delivery, with the scoring in between. The executable ICP guide covers how the model is written; this guide covers what it runs on and where its output goes.
The data jobs inside targeting
Identify. Establish the universe to target within: the accounts already in the CRM plus the market-wide set of companies that match the broad shape of your customers, including lookalikes of the accounts you have already won. Enrich. Fill the fields the ICP evaluates, through a waterfall of licensed sources: firmographics, technographics, funding stage, geography, and the buying committee, since targeting a company without knowing who to reach there is half a decision. Verify. Check the fields that move an account across a threshold, resolve parent and subsidiary confusion, verify the contacts that will be matched into audiences, and apply the exclusions. Score. Evaluate the ICP on every row to produce a fit tier and a reason, attach the freshest signal for timing, and combine the two into a priority the channels can sort on. Deliver. Push the resulting tiers into the CRM as a field, into the sequencer as ordered batches, and into the ad accounts as matched audiences shaped the way each platform expects, with suppression lists alongside.
The data layers and the fields targeting runs on
| Layer | Fields that matter | Targeting use |
|---|---|---|
| Company data | Domain, industry, employee count and growth, revenue band, HQ country, technologies in use, funding stage, parent and subsidiaries, similarity to won accounts | ICP fit evaluation; company-list audiences on ad platforms |
| Person data | Title, seniority, department, verified work email, phone, location, profile URL | Buying-committee targeting; contact-list audiences and customer match uploads |
| Signals | Hiring by function, funding round and date, technology adopted or dropped, news, intent topics and recency, relationships | Timing and priority; in-motion audiences |
| Verification | Email status and verified date, field confidence and source, duplicate flag, customer, competitor and opportunity flags, consent and do-not-contact status | Match rate on platforms; suppression lists |
| Delivery | Fit tier and reason, priority score, CRM field values, audience IDs per platform, export file per platform format, refresh date | The same targeting in every channel |
Verified email deserves emphasis on the ad side. Matched and customer-match audiences are built by hashing contact fields and matching them to platform members, so the match rate - the share of uploaded contacts the platform recognizes - depends directly on the accuracy of the emails uploaded. A verified list matches more people for the same upload. The matched audience and customer match explainers cover the platform mechanics.
Autonomous agents versus targeting by hand
By hand, targeting happens three times in three tools. Sales builds a target-account list in the CRM from whatever fields it has. Marketing builds an audience in the ad platform from a CSV that was accurate last quarter. RevOps maintains a tiering formula in a spreadsheet. The three drift immediately. An autonomous AI agent runs the targeting once - enrich, verify, score, suppress - and delivers the same result into each channel in the shape that channel expects, then refreshes all of them on a schedule.
| Step | By hand | Run by an autonomous agent |
|---|---|---|
| Universe | The CRM as it stands | CRM plus market-wide discovery and lookalikes of won accounts, resolved to canonical domains |
| ICP evaluation | Run on partly empty fields; empties silently fail the threshold | Fields filled through a waterfall first; tier and reason stored per row |
| Timing | Rep intuition, or a news alert someone saw | Freshest signal attached per account, priority combines fit and timing |
| Suppression | A do-not-contact tab, applied unevenly | Suppression list built from CRM flags, refreshed weekly, pushed to every channel |
| Ad audiences | Quarterly CSV upload with unverified emails; low match rate | Verified contacts pushed as matched audiences to connected accounts, exports for the rest |
| Consistency | Three lists, three definitions, three refresh dates | One list, one definition, delivered everywhere, parked for approval |
The person keeps the decisions that carry judgment: the ICP thresholds, the weight given to timing, and the approval on every push. The agent keeps the decisions that carry repetition. That split is what makes the targeting consistent enough to trust with budget.
A useful discipline is to treat the ICP as a hypothesis the targeting infrastructure tests every week. When the tiers are stored on the row with their reasons, and the outcomes flow back from the CRM, the agent can report which fields actually separate winners from losers: perhaps employee growth predicts more than employee count, or a particular technology in the stack matters more than the industry code. That report is the input to the next revision of the ICP, and the revision propagates to every channel on the next scheduled push. Targeting stops being a workshop output and becomes a model that improves on evidence, which is the most durable advantage the infrastructure provides.
The metrics that show it is working
The figures below are illustrative examples for a mid-market B2B program running the same targeting across sales and paid.
71%match rate on a verified contact-list audience, up from 43% unverified (example)3.4xmeeting-rate gap between tier-one and tier-three accounts (example)86%of paid impressions landing on accounts inside the ICP (example)$1.90cost per qualified account after enrichment, verification and scoring (example)
Match rate is the platform-side proof that the contact data is real. Outcome gap between tiers is the proof that the fit model captures something; if tiers perform alike, the model or the fields need work. In-ICP share of spend tells you whether the audiences delivered to the ad accounts are the same accounts sales is working. Cost per qualified account keeps the whole thing honest, dividing credits by the accounts that survived enrichment, verification and the ICP evaluation.
How AstroFabric does it
AstroFabric runs targeting as a mission against one catalog and delivers the result to every channel. The universe builds from discover_companies for firmographic and technographic filters, similar_companies for lookalikes of the accounts you have won, companies_using_tech for displacement targets and buyer_intent_companies for accounts researching your category. Fields fill through list_enrich, drawing on company_lookup, tech_stack, funding_events and hiring_signals across licensed sources, and the buying committee comes from buying_committee with find_email and email_verify so the contacts will match on the platforms.
list_score evaluates your ICP on every row and stores the tier and the reason; list_hygiene builds the suppression set from customer, competitor and opportunity flags. Delivery fans out from the same list: crm_upsert_contacts writes the tier into the CRM, list_push loads ordered batches into the sequencer, audience_push creates matched audiences and suppression audiences on connected LinkedIn, Meta and Reddit accounts, and audience_export produces platform-ready files for Google, TikTok, X and Pinterest. Every push is parked for one approval, and create_schedule refreshes the whole set weekly. Plans start at $49 per month, a verified contact is a few credits, and finder misses are free. The landing page for this job is Data Infrastructure for Targeting.
Frequently asked questions
What is data infrastructure for targeting?
The filled, verified and scored account and contact data that lets a team decide who deserves attention, when, and who must be excluded, plus the delivery that pushes the same decision into the CRM, the sequencer and the ad accounts. It turns targeting from three drifting lists into one refreshed list.
How does targeting differ from segmentation?
Segmentation divides a universe into groups that get different treatment; targeting decides which groups and accounts get resources now. Targeting sits on top of segments, adding fit scores, timing signals and suppression, and delivering the prioritized result to the channels where budget and rep time are allocated.
Why does verification matter for ad targeting?
Matched and customer-match audiences work by matching hashed contact fields to platform members. Invalid or stale emails simply fail to match, so the audience is smaller than the upload and the spend reaches fewer of the intended people. Verifying contacts before the push raises the match rate directly.
Which ad platforms can receive a targeting list?
Matched and custom audiences, including suppression audiences, push directly to connected LinkedIn, Meta and Reddit accounts for approval. For Google, TikTok, X and Pinterest the agent produces platform-ready export files in the format each expects, which your team uploads in the platform interface.
How do I keep targeting consistent between sales and marketing?
Build it once as a persistent list with the ICP evaluated on filled fields, then deliver that list to every channel from the same source on the same schedule. When the CRM tier, the sequencer batch and the ad audience all derive from one list, the two teams are working the same accounts by construction.
Sources
- Google Ads Help - About Customer Match
- Meta Business Help Center - About Custom Audiences from customer lists
- LinkedIn Marketing Solutions - Matched Audiences
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.