Target Account Pipeline
Simplified ABM with Clay: qualify 5,000 accounts and score them
A simplified ABM playbook: take 5,000 accounts through Clay for tech stack, news, hiring and org chart checks, score them, and work the top tier by hand.
7 min · Ayhan K. Isaacs · November 2026
- Signal cohorts
- CapturedRelationalObserved
- Difficulty
- Advanced
- Time to run
- About 3 weeks to the first Tier 1 outreach
- Tools
- Clay, HubSpot, Salesforce, LinkedIn Sales Navigator, Smartlead, Instantly, Apollo, BuiltWith, LinkedIn Ads
ABM usually fails at the list. Teams load 5,000 accounts into an intent platform and spread the same ads and sequences across all of them.
ABM with Clay starts by qualifying the 5,000 with checks anyone can read: what each company runs, what it announced, who it hired, and who sits in the buying team. People then work the few hundred accounts that score. Ads and light outreach cover the rest.
Who this is for
This playbook is for B2B SaaS companies with a market of a few thousand companies and deals large enough to justify research on each account.
It fits when marketing and sales agree on the ICP in words but not yet in a list. You need Clay, a CRM, and reps or BDRs who can work a short list of accounts with care.
Who this is not for
This is not for markets of tens of thousands of small buyers, where research on each account costs more than the deal is worth.
It is also not for teams that cannot describe their ICP in one paragraph. Start with closed-won analysis, then come back to the 5,000.
The signal it engineers or captures
Most checks in this ABM playbook are observed signals: tech install or removal, a funding round, a press mention, job postings for relevant roles and a new executive hire. Every vendor can see them, so they qualify an account and decide timing. They are not the reason to write.
The playbook adds relational signals (a past champion or a past conversation at the account) and captured signals on top: ad engagement from a named account, an identified website visit and an email reply. The captured signals decide which accounts get a person's time this week.
The workflow
01 Start from closed-won, not a database filter
Export the last 12 to 24 months of closed-won and closed-lost deals from HubSpot or Salesforce. Write down what the won accounts share: size band, industry, country, the tools they ran, and the title of the person who signed.
Those shared traits become the scoring criteria. If you have fewer than 20 wins, use your 10 best customers and interview their account owners.
02 Build the 5,000
Pull the account universe from Clay's company search, Apollo or LinkedIn Sales Navigator, using only the hard criteria: employee range, industry, country, and a website that resolves.
Deduplicate on domain against the CRM. Mark customers, open opportunities and recent closed-lost accounts so they are routed to their owners. Cap the table at 5,000 so the per-account checks stay affordable.
03 Run the qualification columns
Run the columns in this order, so accounts that fail early stop costing credits. (1) Business model, from a Claygent prompt on the homepage: "Does this company sell software to businesses? Answer yes, no or unclear." (2) Employee count and its growth over 12 months. (3) Tech stack from BuiltWith or a Claygent check for named tools: the systems you integrate with, the competitor you replace, and the tool that shows maturity.
(4) Funding stage and last round date. (5) News from the last 90 days from Clay's news enrichment: launches, acquisitions, new markets and leadership changes. (6) Open job postings for the roles your product helps or replaces. (7) Product direction: does the site mention AI features, an agent or a copilot? (8) A one-line summary written by an AI column from the columns above, for the rep.
04 Review the org chart for accounts that pass
For accounts that pass step 3, find the buying team in Clay or Sales Navigator: the economic buyer, the person who runs the process your product changes, and one or two users. Record each person's title, start date in the role, and whether they were a contact at one of your past customers.
An account with nobody in the role your product serves is a weaker fit, however good the rest looks. Do this step last, because it costs the most per account.
05 Score and tier
Use a formula column out of 100. Example weights to tune: firmographic fit 30, stack fit 20, timing (funding, news or hiring in the last 90 days) 20, buying team found and reachable 15, relationship (a past champion, a mutual connection or a past conversation in the CRM) 15.
Tier 1 is 75 and above: named accounts for reps, usually a few hundred at most. Tier 2 is 50 to 74: ads plus a light sequence. Tier 3 gets ads only, or nothing. Read the top 20 rows of each tier by eye before you trust the formula.
06 Work Tier 1 by hand
Each Tier 1 account gets an owner, a one-paragraph account note built from the Clay summary, and a first touch that uses one fact from the research: the news item, the hire or the stack.
Write to two or three people per account, by email, LinkedIn and phone. Smartlead or Instantly can send the follow-ups, but a person writes the first line for each account.
07 Run ads and capture engagement for every tier
Upload Tier 1 and Tier 2 to LinkedIn as a matched company audience. Track ad engagement and site visits by account in Clay or your visitor identification tool.
A Tier 2 account that clicks an ad and visits pricing moves to Tier 1 that week.
08 Refresh every month
Re-run the timing columns (news, hiring, funding and job changes) every month and rescore. Retire Tier 1 accounts that had a full sequence and no engagement for 60 days, and promote the next ones in line.
Write the tier and the score to the company record in the CRM, so sales and marketing work from the same list.
What it costs and how long it takes
Tools: Clay, a source for the account universe (Clay, Apollo or Sales Navigator), BuiltWith or Claygent for stack checks, a LinkedIn Ads budget, a sequencer and the CRM. The qualification columns are the main Clay cost, so order them from cheap to expensive and stop early.
Time: a week for the closed-won analysis and the scoring model, a week to build and run the 5,000, and outreach starts in week three. The monthly refresh takes a few hours.
The numbers
Two of our records show the qualification step at work. For Intelak, the Emirates startup accelerator, we sourced about 7,000 startups from Crunchbase and qualified them in Clay using company websites, founders' LinkedIn profiles and public news. 3,000 qualified, which produced 164 interested companies and 48 applications in the 2025 round.
For Brandpilot, we estimated each company's paid search spend with Clay and the SEMrush API to decide which accounts to target. The program produced 40 qualified leads and USD 600,000 in pipeline.
For your own run, measure accounts per tier, Tier 1 accounts engaged within 60 days, meetings by tier and pipeline by tier. If Tier 2 books as many meetings as Tier 1, the weights are wrong.
- Intelak 3,000 qualified from about 7,000 Startups from Crunchbase qualified in Clay using company websites, founder LinkedIn profiles and public news; 164 interested and 48 applications in the 2025 round.
- Brandpilot USD 600,000 pipeline Each company's paid search spend estimated with Clay and the SEMrush API; 40 qualified leads.
- Optix 124 interested prospects A Clay table that found and qualified co-working operators running up to 20 spaces, then 20 messaging angles tested; handed to their account executives over 8 months.
Where it breaks
Bad source data is the first thing that breaks. Company databases miscount employees and mislabel industries. The business-model check in step 3 catches most of it. Read 50 rows by hand before you run the full 5,000.
Credit burn is the second. Running every column on every account costs far more than running them in order with early stops.
The third is a score nobody trusts. If reps were not in the room when the weights were set, they go back to their own lists. Show the summary line and the score breakdown on the CRM record.
The fourth is a Tier 1 list that keeps growing. If Tier 1 holds more accounts than your reps can work with care, raise the threshold. A long Tier 1 list turns back into volume outbound.
Hand-off to sales
The rep gets the account in the CRM with the tier, the score breakdown, the summary line, the buying team with titles, and the fact to open with.
The first message uses that fact and asks a question about it. Example: "Saw you opened a Riyadh office in May. Is hiring the local sales team on your list this quarter, or is that already done?" A reply goes to the account owner the same day. An ad click or pricing visit from a Tier 2 account creates a task for the owner.
FAQ
Related
- Signal-based outbound: hiring, funding, new leaders and stack changes
Outreach timed to buying signals, like a new VP, a funding round or a job posting, instead of a bought list: the Clay setup, the scoring and the messages.
- Champion tracking: follow past buyers to their new company
How to track job changes among past buyers and users with Clay, score the new company, and reach out in the first 90 days with a reason to talk.
- Website visitor identification: from content to booked meetings
Publish one piece, promote it with outbound and LinkedIn ads, identify the companies and people who read it with RB2B, Warmly or Clay, then follow up.