The ABM data playbook: building the account universe ABM
"ABM data" is shorthand for four layers of data that work together to make account-based marketing operational: the account universe (discovery), firmographic and technographic enrichment, the contact and buying-committee graph, and intent and signal data. Most ABM frameworks teach you how to target, score, and engage named accounts. They assume the account universe and contact graph already exist. This playbook fills the missing first chapter: how to build the universe from scratch, layer the data sources that make scoring work, and maintain the foundation as accounts decay.
Most ABM playbooks assume an enterprise or mid-market SaaS account universe, where the standard data graph (LinkedIn plus corporate web, surfaced through Apollo, ZoomInfo, Clay, Demandbase, 6sense) covers the TAM well enough that the targeting and scoring models work. For teams running ABM into local businesses, trades, restaurants, franchise operators, healthcare, or other non-LinkedIn-native segments, the same playbook breaks at step one. The account universe itself is incomplete. Traditional providers cover decision-maker mobile at 10-20% in these segments against a discovery-first benchmark of 60%+. ABM into a non-LinkedIn-native ICP requires a discovery-first data layer underneath the standard ABM stack.
1. What ABM data actually refers to (four distinct layers)
1.1. Layer 1
The set of accounts that match your ICP and could become customers. The discovery layer. This is the layer most ABM frameworks assume is solved. For non-LinkedIn-native segments, it isn't. Discovery is upstream of enrichment. You can't enrich records that don't exist.
1.2. Layer 2
Industry, sub-industry, employee count, revenue band, tech stack, geography. This is what most enrichment vendors sell. Apollo, ZoomInfo, Clay, Cognism, and Lusha all play here. The layer that's most commoditized and most often confused with the full ABM data stack.
1.3. Layer 3
The named contacts at each account, their roles, and their contact information (verified email plus mobile). Where coverage gaps hit hardest. Decision-maker mobile coverage runs 10-20% on horizontal providers for non-LinkedIn-native segments and 60%+ on a discovery-first stack.
1.4. Layer 4
Behavioral signals indicating in-market state. 6sense, Bombora, G2 buyer intent, Demandbase. Useful when paired with a complete account universe. Misleading without one. Intent flags accounts as in-market that the rep can't reach because the contact layer underneath is empty.
2. Building the account universe
2.1. The universe-first mindset
Before scoring, before targeting, before campaigns: ensure your account universe contains the accounts you actually want to win. A provider's 300M+ contacts claim doesn't predict whether your 5,000 target accounts are in the data set with usable detail. Database size is a vanity metric. Coverage on your accounts is the only number that matters.
2.2. How to test your provider's coverage of your TAM
Pull a list of 100 accounts that fit your ICP from public sources (state registrations, license records, industry rosters). Run them against your provider's data set. Track three numbers: account match rate (does the account exist?), decision-maker presence rate (does the right contact exist?), and mobile direct-dial usability rate (does the contact have a usable phone?).
Honest benchmarks vary wildly by segment. LinkedIn-native enterprise: 70-90% account match, 60-80% contact, 50-70% mobile. Local-business, trades, franchise, and SMB: 40-60% account match, 20-40% contact, 10-20% mobile.
2.3. Where discovery has to come from non-LinkedIn sources
Trades and home services: contractor licensing data (805,000+ records nationally). Restaurants and multi-unit operators: POS and tech-stack detection plus franchise-hierarchy registries. Healthcare: NPI and credentialing data. Local SMB: state business registrations and secretary-of-state filings. No LinkedIn-dependent provider reliably pulls from these sources. Discovery requires a different data architecture.
2.4. Account universe maintenance and decay
Universes decay. Enterprise baseline is about 30% per year. Local-business segments decay structurally faster because of higher closure rates, ownership transitions, and phone-line turnover. Quarterly refresh cadence at minimum. Monthly for high-decay segments. Without refresh, the universe drifts and every metric calibrated against it drifts with it.
3. Tier modeling
3.1. Tier 1
Custom outreach, named campaigns, exec sponsorship. Data-depth requirements highest. Full buying committee mapped, signal data on each account, account-level research and personalization.
3.2. Tier 2
Vertical or segment campaigns. Programmatic personalization with template variation by sub-segment. Data depth: industry, role, company size, plus top-three stakeholders identified.
3.3. Tier 3
Programmatic ads plus automated nurture. Data depth: firmographic plus intent. Closer to demand gen with account-based reporting.
3.4. How coverage gaps distort tier modeling
If your data graph misses 50% of your TAM, your "Tier 1" accounts are the LinkedIn-visible slice, not the highest-fit slice. The bias propagates through the entire program: best resources spent on accounts the provider happened to cover, not accounts most likely to close.
4. Account scoring and propensity modeling
4.1. Firmographic fit score
Industry, size, and geography matched against ICP. Requires accurate firmographic enrichment in Layer 2. The most stable component of the composite score.
4.2. Technographic / tech-stack fit score
Tools the account already uses, predictive of fit. Useful for SaaS targeting where complementary tooling signals readiness. Weaker for non-tech ICPs.
4.3. Behavioral / intent score
Site visits, content engagement, third-party intent signals (6sense, Bombora, G2). Intent without a complete account universe is misleading because the platform reports on accounts you can't reach.
4.4. Engagement score
Marketing and sales touch quality and recency. Internal data only. The cleanest source.
4.5. Composite account fit-and-intent score
Weighted blend of the four components. Most ABM platforms compute this natively. The weighting is the team's call. Default weights of 40/20/30/10 for fit, technographic, intent, and engagement work as a starting point.
5. ABM platforms and the data they bring (or don't)
5.1. Demandbase
Account identification, intent, and ads orchestration. Strong on enterprise SaaS. Same architectural data layer as the rest of the category.
5.2. 6sense
Intent platform first. Predictive scoring on signal data. Pairs with Apollo or ZoomInfo for the contact graph. Architectural ceiling on contact data is the same as the others. The pattern teams notice on local-segment account uploads: load 1,000 accounts, get usable enrichment on roughly 100. The other 900 fall out at the resolution step.
5.3. Terminus and RollWorks
Mid-market ABM ad orchestration. Strong on programmatic. Data-layer constraints are the same.
5.4. HubSpot Breeze intelligence (formerly Clearbit)
Now part of HubSpot's Breeze suite. Company enrichment focus. No contact data for local businesses. Useful for enterprise and mid-market SaaS company-level enrichment. Not a discovery tool.
5.5. Where the data layer sits underneath all of them
Apollo, ZoomInfo, Clay, Cognism, Lusha provide the contact graph. Plus a discovery-first layer (DataLane) for non-LinkedIn-native segments where the horizontal providers don't cover the TAM. The platforms above orchestrate. The data layer feeds them.
6. ABM data across the lifecycle (new logo, expansion, retention)
6.1. New-logo ABM
Standard target-account, engage, close motion. The most-discussed ABM use case and the one most ABM platforms are built for.
6.2. Expansion ABM
Existing-customer multi-product, multi-location, multi-stakeholder expansion. Data requirement: full account hierarchy. Franchise hierarchy is the structural gap. A 22-location franchisee may transact at the location level even when the brand is on a master agreement, and most horizontal providers don't resolve this cleanly.
6.3. Retention ABM
At-risk account identification, win-back campaigns. Data requirement: usage and sentiment signals plus a complete contact graph so the team can engage the right stakeholder when a champion leaves. Champion-departure detection is one of the highest-use signals in retention motion.
7. How DataLane fits in an ABM data architecture
DataLane is a discovery-first data layer. It builds the account universe (Layer 1) from non-LinkedIn sources: contractor licensing (805,000+ records), franchise and multi-unit operating data, POS and tech-stack signals, regulatory filings, and state business registrations. 17M+ US local-business locations indexed from this graph.
It then enriches the universe with firmographic, technographic, and contact-graph detail (Layers 2 and 3). Mobile direct-dial coverage runs 60%+ in segments where Apollo, ZoomInfo, Clay, Cognism, and Lusha return 10-20%. The positioning is complement, not replacement. For the LinkedIn-native portion of your TAM, the horizontal providers still work. For the local-business, SMB, franchise, trades, restaurant, and multi-location ICPs, DataLane builds the foundation those tools architecturally can't.
Teams without a discovery layer pay the manual enrichment tax. About 45 minutes per account hand-doing license-record lookups, operator verification, and location resolution. About two minutes per account on the discovery-first stack. The vendor-churn pattern follows. A VP cycles through Apollo, ZoomInfo, and Clay annually, looking for the provider that finally returns coverage on their ICP. None of them does, because they share the same source graph.
8. Implementation
Days 0-30. ICP definition, TAM sizing, provider coverage test on a 100-account sample, gap analysis. Output: a clear picture of where your provider covers your TAM well and where it doesn't.
Days 31-60. Pick the discovery-and-enrichment stack. Populate the account universe. Run firmographic enrichment. Score for fit. Output: a scored, deduped account universe in the CRM with field-level source-of-truth assignments.
Days 61-90. Add the intent layer if relevant. Build the buying-committee graph for Tier 1 accounts. Instrument the scoring model in the marketing automation or ABM platform. Run the first cohort of campaigns. Output: a working ABM motion against the scored universe.
Quarterly: refresh cadence on the universe and decay management. The 90-day plan is the build. The quarterly refresh is the maintenance.
9. ABM data pitfalls to avoid
Trusting database-size claims. 300M+ contacts doesn't mean your TAM is covered. Skipping the account universe step. Most teams jump to scoring; the foundation isn't there. Using intent data as a substitute for contact data. 6sense flags hot accounts you can't reach. Letting decay erode the foundation silently. Quarterly refresh is the floor. Cycling vendors instead of fixing architecture. Same source graph, same ceiling. Treating ABM as a tool purchase rather than a data discipline. Tools amplify. They don't fix.
For local-business ICPs, cold calling the mobile is the highest-leverage outbound channel because it bypasses both the email-deliverability ceiling and the LinkedIn-presence gap that affect the same segment. The main-line dial routes through a gatekeeper (the hostess at the restaurant, the receptionist at the dental office, the foreman screening calls for the GC), and the conversation never reaches the decision-maker. The direct mobile skips that layer entirely.
Frequently asked questions
What is account-based marketing data?
ABM data is the four-layer foundation that makes account-based marketing operational: the account universe (discovery), firmographic and technographic enrichment, the contact and buying-committee graph, and intent and signal data. Most ABM frameworks assume the universe layer is already built. In non-LinkedIn-native segments, it usually isn't.
How do you build an account list for ABM?
Start with ICP definition (industry, size, geography, tech and behavioral fit). Pull a TAM sample from public sources (license records, registrations, industry rosters) and test your provider's coverage against it. Where coverage is weak, layer in discovery-first sources. Score for fit, then enrich for engagement.
What's the difference between intent data and ABM data?
Intent data is one layer of ABM data: behavioral signals indicating in-market state (6sense, Bombora, G2). It pairs with the other three layers (account universe, firmographic enrichment, contact graph). Intent data alone, without a complete account universe, surfaces accounts you can flag as in-market but can't reach.
Is ABM only for enterprise targets?
No. ABM works at any account size. Tier 1 (one-to-one, 50-200 accounts), Tier 2 (one-to-few, 200-1,000), Tier 3 (one-to-many, 1,000+). Data depth requirements vary by tier. The constraint isn't account size. It's whether the data graph covers your ICP.
How often should you refresh ABM data?
Quarterly at minimum for enterprise-style ICPs. Monthly for high-decay segments: local businesses, trades, restaurants, and franchise operators where ownership transitions, closures, and phone turnover compound faster than the enterprise baseline of about 30% per year.
What ABM platforms should I evaluate?
Demandbase, 6sense, Terminus, RollWorks for the orchestration layer. The choice depends on segment (Demandbase strongest in enterprise SaaS, Terminus in mid-market) and existing stack. The bigger evaluation question is the data layer underneath: whether the platform's data graph covers your ICP, and what to do where it doesn't.
How do I know if my ABM program has a data problem versus a process problem?
Sample 50 target accounts. Check whether the firmographic fields are accurate, whether the buying committee is mapped, and whether the contact mobiles are usable. If the answer is no on more than half, the program has a data problem. Process improvements can't compensate for an incomplete foundation.
ABM data is what most playbooks skip. Account universe and contact coverage are the upstream variables that determine whether the orchestration ever fires correctly. For LinkedIn-native ICPs, the standard data stack works. For local-business segments, the discovery problem is the binding constraint, not the targeting layer. For the intent-data layer ABM depends on.
