- 01
Score your current infrastructure
- Rate each checkpoint 0 (not in place), 1 (partial) or 2 (fully in place)
- Cover four pillars: market and data, lead prioritization, rep workflow, AI implementation
- Total out of 32: 0–8 means start at the data foundation, 25–32 means keep iterating
- Fix the lowest pillar first, not the one that feels most urgent
- 02
Map your obtainable market, not your TAM
- List the accounts you can win today, not the fundraising-deck TAM
- Source from a waterfall, not one database: lead finder API, Sales Navigator, lookalikes of your best 10 closed-won accounts
- Add event attendee lists, company followers and funding feeds where relevant
- Keep the list updating continuously, not once a quarter
- 03
Enrich for context, not just emails
- Run every account through Clay: description, tech stack, headcount growth, open roles, funding, review activity
- Run contact data as a sequential waterfall (IcyPeas, Prospeo, BetterContact, LeadMagic), not all at once
- Get verified mobile numbers, not only work emails
- Enrich specifically for the 2–3 signals your best reps say predict a close
- 04
Turn your best rep's instinct into a score
- Pull your last 50 closed-won accounts
- Ask your best closer what those companies had in common before the demo
- Extract 5–8 observable signals (headcount growth, relevant hiring, funding, new decision maker)
- Build a 0–10 weighted score in Clay, start with equal weights and adjust as data comes in
- Refresh scores weekly and swap low scorers out of live sequences
- 05
Tier and route every account
- Tier 1: highest fit and signal, manual personalized outreach
- Tier 2: strong fit, moderate signal, multi-channel automated sequences
- Tier 3: fit confirmed, no signal, low-touch nurture re-evaluated weekly
- Use a routing table (score, tier, signal type) to set channel, sequence and owner, so the system decides, not the rep
- 06
Automate pre-call research
- Run an AI prompt in Clay on the enriched account data
- Generate exactly three company-specific talking points, one sentence each
- Push them to a custom CRM property (for example a Pre-Call Brief field in HubSpot)
- Show that field in the default contact view so reps see it before they dial
- 07
Centralize the build with one owner
- One person owns the Clay table: enrichment, scoring, routing and prompts live there
- Reps never touch the table, they get outputs in the CRM and sequencer
- Deliver AI outputs inside tools reps already use, no new surfaces
- Spend 80% of build time on pipeline (list quality, scoring, pre-call research), 20% on reporting and hygiene
- 08
Validate manually, then scale and iterate
- Prove the signals and scoring by hand on 50 accounts before automating
- Aim for fewer, better accounts, not more volume
- Close the loop with reply categorization in the sequencer and deal source tracking in the CRM
- Review weights, prompt quality and waterfall hit rates every week
Tools mentioned
- Clay
- IcyPeas
- Prospeo
- BetterContact
- LeadMagic
- LinkedIn Sales Navigator
- HubSpot
- Smartlead
- Claude