This ABM engine guide was originally shared by Kyle Poyar in his Growth Unhinged newsletter. You can read his original article here: How to build a modern ABM engine.
Account-based marketing (ABM) is a go-to-market strategy that works backwards from a list of target accounts.
It has historically been expensive and highly manual, but AI agents and data enrichment have made it way more accessible.
We've personally worked with 250+ companies at workflows.io, including dozens of ABM implementations. This is the guide we use, broken into the seven components every team needs to ship.
When ABM is the right call
We recommend ABM for any B2B company with two specific characteristics.
Annual contract value (ACV) above $50,000.
Total addressable market (TAM) below 20,000 qualified companies.
For companies with a TAM under 20,000, the relatively low conversion rates of other GTM approaches like automated outbound and LinkedIn content make them hard to rely on in silos.
If both apply, ABM is one of the highest-leverage GTM programs you can build.
The 7 Components of a Modern ABM Engine
Step 1: TAM and Stakeholder Mapping
The goal of step one is to create a target account list (TAL) that captures 90 percent or more of your TAM.
When your TAM is under 20,000 companies, every incremental increase in coverage has a real revenue impact. So this step is worth doing properly.
Start by building a data-supported ICP (ideal customer profile) model. Many GTM teams don't actually have their ICP written down in an agreed-upon document.
If you need help, our ICP modeling guide, Freckle's closed-won Claude skill, and Growth Unhinged's ICP sharpener skill are all good starting points.
Once your ICP is locked in, pull lists from multiple data sources to hit full TAM coverage. There are four main categories.
General prospecting databases. Like Apollo and ZoomInfo. Best for companies with a LinkedIn presence.
Lookalike databases. Like Ocean.io and Discolike. Good for hyper-specific searches that include companies missing from LinkedIn.
Specialized databases. Like Storeleads or influencers.club. Useful for niche verticals.
Web scraping. Using Apify or Claude Code. Necessary to map TAMs that public databases don't cover well.
We typically combine three or more sources to hit 90 percent coverage. After pulling lists, we merge and deduplicate them.
We used to do this in Clay, but we recently switched to Claude Code and find it much faster.
Once the company list is qualified, repeat the process for contacts. Map titles to decision-makers, champions, influencers, and end-users. Pull contacts from Apollo, AI Ark, and Clay. Verify emails through Findymail and BetterContact.
Step 2: Account Research
After the list is built, you collect the data points that personalize outreach and help reps research accounts without leaving the CRM.
For most clients, this works out to 30+ custom data points, which means 30+ custom CRM properties to configure.
We use a spreadsheet to map every data point with its official HubSpot property name and any overwrite rules.
Real examples from clients, in order of simple to complex:
Engineering headcount
Custom sub-industry classification
E-commerce hosting platform
Competitor tech usage
Closest coffee spot to the prospect's office
Recent clinical trials
Parent-child company relationships
Composite media-buying score
We use two main tools for research: Claygent (web research agents in Clay) and Clay's 150+ data providers.
A full TAM map with account research typically costs $2,000+ across data, Clay, and AI credits. It's worth it because the data lives in your CRM forever and powers every campaign that follows.
Step 3: CRM Cleanup and Enrichment
Once the static research is done, you turn it into a continuous workflow.
Every new record in HubSpot or Salesforce should automatically run through the same enrichment process.
The setup is straightforward:
Duplicate the TAM workflow.
Set the trigger to list enrollment (24-hour cadence) or webhooks (instant, requires HubSpot Data Hub Pro at $800 a month).
Change the final action to "Update record" instead of "Create record".
Enroll the backfill of companies and contacts.
We typically add an "Enriched by Clay" date field on company and contact records, which makes it easy to re-enrich on a 12-month freshness cycle.
Before pushing any new data, do a CRM cleanup. At a minimum: remove hard domain and email duplicates, and audit existing properties to prevent redundancies.
Three tools cover 99 percent of CRM hygiene use cases.
Clay + HubSpot integration or API
HubSpot Data Hub
Claude Code plus the HubSpot MCP
After cleanup, add quality-of-life improvements:
Customize company and contact views to surface the most important research
Create segments by tier and category
Add on-demand enrichment buttons inside HubSpot so reps can click "Find additional stakeholders" without leaving the CRM
Step 4: Signal Tracking
Signals are the part of an ABM engine that removes the silo between sales and marketing.
A signal is any data point that suggests buying intent. When marketing drives 50 ad impressions to an ICP account, that signal can be sent to the sales team, and both teams stay aligned on the only stat that matters: ICP pipeline progression.
Signals split into three categories.
1st-party signals. Your internal data from your own tools. CRM activity, website visitors (Warmly, RB2B, Vector), gated content, product usage. Usually free and the highest-intent.
2nd-party signals. Exclusive data from external tools. Social engagement (Clay, Jungler, Teamfluence), champion tracking (Clay, Champify, UserGems), ad engagement (ZenABM, Fibbler, Factors.ai).
3rd-party signals. Public data from external sources. News and fundraising (Clay, PredictLeads), job openings (LinkedIn Sales Nav, TheirStack), competitor tech usage (BuiltWith, HG Insights).
Capturing a signal is easy. Activating it is 90 percent of the work.
We use a 13-step process for every signal workflow:
Capture. Pull signals from all sources via webhooks, APIs, or native integrations.
Aggregate. Route everything into one orchestration layer.
Normalize. Standardize key fields like company domain, LinkedIn URL, and job title.
Enrich. Add basic enrichment for qualification.
CRM lookup. Check if the account already exists and pull the assigned owner.
Qualify. Use enrichment data to qualify net-new companies and add to the CRM if missing.
Score. Pull tier scores (Tier 1, 2, 3, Unqualified) from the CRM.
Segment. Group accounts by size, industry, location, or business type.
Route. Assign signals to the right rep using a live rep assignment table.
Sync to CRM. Push assigned signal data back into HubSpot or Salesforce.
Activate. Tier 1 gets a Slack alert plus manual outreach. Tier 2 gets retargeting plus automated outbound via Instantly and HeyReach. Tier 3 gets automated email.
Track. Roll up signals into awareness stages.
Enablement. Build custom sequences, call scripts, weekly digests, and dashboards per rep.
In our best cases, 20 to 40 percent of active pipeline is attributed to signal activation (using a 7-day deal creation window).
Step 5: Awareness Scoring
Awareness scoring is a strategy we learned from Growth Unhinged, particularly the case studies with Parabola and Emilia Korczynska. We then operationalized it inside HubSpot and Clay for our clients.
The five awareness stages:
Identified. Part of the qualified TAL, no engagement yet. Default state.
Aware. Showed surface-level engagement. One website visit, 50+ ad impressions.
Interested. Repeated or high-intent engagement. A positive outbound reply, event attendance.
Considering / Evaluating. Bottom-funnel stage just before the biggest conversion drop-off, usually right before the first meeting.
Selecting. In an active deal cycle with an opportunity in the CRM.
When an account moves stages, you have a few activation options:
Tasks and Slack notifications for high-intent signals
Lists that reps prospect from (e.g. Tier 1 + 2 in Interested)
AI signal summaries added to the task description every time the stage changes
Awareness stages solve a real shortcoming of lifecycle stages. Lifecycle stages start after an opt-in, which means you miss a big chunk of warm market activity that happens before the form fill.
Step 6: Demand Generation
By step six, the core ABM infrastructure is in place. Demand generation becomes the focus.
After implementing the earlier components, the same patterns show up across our clients:
Awareness score segments become ad retargeting audiences
SDRs stop list-building outside the CRM entirely
Sales naturally focuses on Interested and Aware accounts (which convert 3x higher)
Sales starts using marketing-generated signals to fill pipeline
A BDR at one of our clients booked four meetings after one day of cold calling out of an "Aware" list. Their reps had been booking 1-2 meetings a week on a $100k+ ACV product, so these numbers were significant.
ABM demand gen channels split into two groups.
1:1 demand generation. ABM gifting campaigns, warm intros, event invites, and manual outreach for dream accounts.
1:many demand generation. Automated outbound through Instantly and HeyReach, parallel dialing through Nooks or Orum, LinkedIn social content, on-site lead magnets, video outreach, targeted ads, public event campaigns, and connection request waves.
One ABM channel worth highlighting: targeted LinkedIn ads. Because your TAM is small, you can upload company and contact lists to LinkedIn ads and guarantee 100 percent of your spend lands on ICP accounts.
For LinkedIn content, the mistake most companies make is posting promotional or corporate content. What drives inbound is content the ICP actually wants to read, usually unrelated to your product.
Step 7: ICP Pipeline Progression Reporting
Reporting is straightforward, but the right six reports make or break an ABM program.
ICP pipeline created (month over month). The most accurate read on ABM program success. This should grow from baseline if the system is working.
Signal influence by category. Model each signal category against pipeline and closed-won, using 7-day and 30-day attribution windows.
Overdue signal tasks by rep. Tasks are the most effective way to assign work to reps, but only if they action them. Percentage of overdue tasks tells you who needs more enablement.
Awareness stage progression and regression by tier. Accounts moving forward predict future pipeline. Accounts moving backward expose leaks before they show up in revenue.
Accounts by tier broken down by awareness stage. The high-level market penetration view.
Tier 1 accounts with no activity in the last 30 days. Accountability metric. This number should be as close to zero as possible.
Conclusion
AI has lowered the manual work and costs of doing ABM, but you still need the right RevOps infrastructure to be effective.
This is far more than buying an intent dashboard and expecting your reps to check it. It's building out the data pipeline that powers all your GTM activities.
When done right, you'll see a flywheel effect: high-performing social content becomes ads, ads warm up your outbound lists, and outbound nurtures social engagers.
This is the only way to break into a limited market fast enough. We've seen it directly with companies like Medrio generating $1.3M in ABM-attributed pipeline in four months.
If you're looking to build something like this out, start by hiring a GTM engineer or RevOps specialist. Or work with us as an external partner. Book a strategy call or visit our ABM services page.
