ABM campaigns do not fail because of bad creative. They fail because the account list is garbage, the contact data is 18 months stale, and nobody agreed on what “Tier 1” actually means. We at WebCoreLab learned this the hard way on a 2024 campaign with a New York cyber firm. Here is the playbook we built after that mess.

Fair warning. This is 800 words of boring plumbing. But if you skip any step, you will spend Q3 wondering why your 23% open rate on “target accounts” converted to 0 meetings.

Step 1: Define the ICP with 5 hard filters, not 50 soft ones

Most ICPs we inherit from clients look like a Christmas list. “Mid-market, growing, innovative, tech-forward, budget available.” That is not an ICP. That is a mood board.
We force clients to pick 5 hard filters. Example from a DevOps SaaS client in San Francisco:

• Industry: software or fintech (SIC 7372, 6199)
• Size: 200-2,000 employees
• Tech stack: uses Kubernetes (detected via BuiltWith)
• Funding: Series B or later in the last 24 months
• Geography: US only, excluding California (they had a conflict)

5 filters. Black and white. Run it against a clean source and you get 1,200 accounts, not 48,000.

Step 2: Pick your 3 data sources, not your 7

Every ABM team wants 10 enrichment tools. You need 3. More than that and the deduplication becomes a second full-time job.

Our standard stack in 2026

• Apollo.io for firmographic base layer. $149/seat/month. Covers 85% of US B2B accounts.
• Clay for enrichment waterfalls and custom signals. $349/month for the agency plan.
• 6sense or Demandbase for intent data. Budget $2-4k/month. Skip if you are under $2M ARR.

That is it. Resist the urge to add ZoomInfo, LeadIQ, Lusha, and Cognism on top. You will spend 6 weeks writing dedup rules.

Step 3: Waterfall enrichment in this exact order

Run the enrichment in a waterfall. Hit the cheapest source first, cascade to the expensive one only when the cheap one fails. This is how to build data for ABM success without torching your budget.

1. Apollo lookup by domain. Cost: ~$0.01 per account. Covers 80%.
2. If Apollo blank or stale (>6 months), query Clay waterfall with LinkedIn + Hunter + Dropcontact. Cost: ~$0.08.
3. If still blank, kick to a VA for manual LinkedIn Sales Nav pull. Cost: ~$1.50 per record.

This waterfall costs us roughly $180 per 1,000 accounts enriched to 95% completeness. Buying the same data from a single “premium” vendor would run $1,400-$2,000.

Step 4: Build the account score with 4 variables max

Overcomplicated scoring models are dead on arrival. Sales will not read a 14-variable score. They will read a simple 0-100 with 4 inputs.

Our default score formula

• ICP fit (0-40 points): how well the firmographic profile matches
• Intent signal (0-30 points): from 6sense or Bombora surge data
• Engagement (0-20 points): website visits, content downloads, email opens in last 30 days
• Technographic fit (0-10 points): do they use a relevant tool in their stack

80+ = Tier 1, call this week. 60-79 = Tier 2, email sequence. Below 60 = nurture with content. A Boston analytics client using this scored 340 accounts, and their SDRs booked 47 meetings in the first 60 days — 4x their previous cadence.

Step 5: Refresh data every 90 days, no exceptions

Contact data decays at 2.5% per month. After a year, 30% of your list is wrong. We put all client ABM lists on a 90-day re-enrichment cycle. We rerun the full Apollo + Clay pass and flag movers (job changers are warm leads — hit them first).
The job-change signal is underrated. People who switched roles in the last 90 days are 4x more likely to buy, because they have budget and want to make an early win.

Step 6: Sync to the CRM with 3 custom fields

Your CRM needs these 3 custom fields to make ABM actually run:

• abm_tier (1, 2, 3, or null)
• abm_score (0-100, updated weekly)
• abm_last_signal (text: what triggered the score bump)

SDRs work the list sorted by tier, then by last_signal. No guessing. No stale queues.

Common mistakes we see every month

• Buying a $30k ZoomInfo seat and never running refresh cycles
• Letting marketing own the list without sales alignment on tiers
• Scoring every account and ignoring the intent layer
• Skipping the waterfall and paying $1/record from one vendor
• Treating ABM as a 1-off campaign instead of a permanent data ops function

Fix those 5 and your hit rate doubles. We have watched it happen on 9 accounts in a row. The short version of how to build data for ABM success is: tight ICP, 3 sources, waterfall enrichment, 4-variable score, 90-day refresh, CRM sync. Do that and the rest of ABM feels easy.