Andrew G.

Most outbound fails because the ICP was a guess. I test competing ICP hypotheses against live reply data before you scale spend. Last framework: 300 companies, winning segment at a 4.17x reply-rate uplift and 48% positive responses. I work with B2B companies of 10 to 300 people where one person owns pipeline: a founder, CEO, or solo commercial lead. Usually they have tried outbound, heard silence, and cannot tell whether the list, the message, or the ICP is broken. Finding out is the work. How an engagement runs: Scoping call. We map your current motion and pick the one metric that decides success. ICP diagnostic. Your assumptions become 3 to 5 testable hypotheses, each with a scoring model behind it. Build. Data collection, enrichment, routing, messaging. A working system, not a slide deck. Test and read. Every hypothesis gets enough volume to produce a verdict. Winners scale. Losers stop before they burn budget. Results from real engagements: ICP testing framework for IT Software Company: multi-hypothesis LinkedIn outreach across 300 companies. Winning ICP delivered a 4.17x reply-rate uplift and 48% positive responses. Outbound diagnosis for a workforce management SaaS: audited 10,000 leads and 238 human responses. Isolated messaging as the failure, not the list, and rebuilt the hooks from reply data. First-mover intent engine for a data center software provider: land acquisitions, utility permits, and construction filings surfaced before any press release. Sales reached decision-makers before the RFP stage. CRM rescue for the same provider: 30,215 Salesforce contacts enriched, 24,949 validated (82%), 7,300 job-changers mapped to their current roles. Investor and partner screening for M&A Boutique: 500 targets scored down to 48 validated, high-intent investors at 87% AI accuracy. You own everything I build. The system runs in your accounts, your team gets trained on it, and nothing depends on me staying. No agency lock-in, no black box. Stack: Clay, Make, n8n, Instantly, Smartlead, Expandi, HubSpot, Claude API, Python. AI sits in the decision layer, replacing analyst judgment rather than just effort: cheap models for bulk enrichment, strong models for the calls that matter. Every data point traces back to its source. Based in Warsaw. Full EU coverage, and I overlap US Eastern from morning to end of business. Not a fit: commission-only projects, bulk spam sends, or "just get me leads" with no room to fix the targeting. If your outbound is quiet and you do not know why, message me with one line on what you sell and to whom. I will reply with the first hypothesis I would test.