We sat down with Pete Crowley on Adam Robinson’s show and spent an hour breaking down how we build ABM programs at Workflows, covering live clay tables, tiering criteria, and actual client results.
We just wrapped up one for a series A company selling six-figure contracts. 25% of their pipeline now comes from the ABM program.
They thought their TAM was 3,500 companies. We mapped it properly and found 7,000.
Reps booked four meetings on day ONE right after we turned on awareness scoring, by calling accounts that were already in the “interested” stage.
Key takeaways from the session (full recording is linked at the bottom):
1. ICP modeling from your CRM
Start with your closed-won and closed-lost data. Interview the AEs who closed those deals and the CSMs managing those accounts.
Build your model on three things: firmographics, technographics, and account fit signals.
One thing we kept coming back to: companies mix up account fit with intent signals in the same scoring model. Account fit stays stable. Intent changes daily.
They belong in separate systems.
2. TAM mapping across multiple sources
We never trust one database. Pull from multiple databases, aggregate in Clay, and then qualify with AI.
Industry tags in databases are wrong more often than people think. A healthcare SaaS might tag itself "healthcare" on LinkedIn. If you are filtering for software companies, you will probably miss them.
That's why we always go broad first and narrow down after qualifying in Clay.
3. Account tiering
We shared how we tier our own accounts at Workflows: dream accounts that we handpick, then tier 1, tier 2, tier 3, and disqualified.
The main thing here is to keep things simple.
Complex point scoring sounds good in theory but it falls apart once reps have to use it every day. If a rep can’t look at a tier and immediately know what kind of account they’re dealing with, the model is too complicated.
4. Enrichment in Clay
We showed a live Clay table with 40+ data points on a single client, things like GitHub activity, public repos, stargazers, languages, all uploaded as HubSpot properties so reps get the full picture without leaving the CRM.
One GTM engineer runs this whole operation.
5. Signals and awareness scoring
Five stages in HubSpot: Identified, Aware, Interested, Evaluating, Selecting.
They’re all auto-updated by workflows, and each stage triggers a different play so reps open their CRM and know exactly who to call and why.
This is the one RevOps system reps always ask for. We got into the full methodology in the video.
5. Clay vs Claude Code
We use both.
Clay for repeatable workflows where you need to see what happened to every record, and Claude Code for flexible AI work at scale. Our team logs 200+ hours a week in Claude Code across 20 people.
We took our SOPs from Notion and turned them into Claude Code skills that do the work instead of just explaining how.
That has now shifted how our whole team operates.
Watch the full session
We went way deeper on each of these in the recording. If you’re thinking about building an ABM program, it’s worth your time.
Want help building an ABM system for your team? Book a meeting here.
We'll walk you through it.