Built by Fahad Cheema

Designing the account scoring, targeting and measurement system to take an enterprise AI company into a brand-new vertical with no existing outbound infrastructure.

→ Introduction

Unframe partners with large enterprises to design and deploy production-grade AI systems. The company was opening a new vertical: insurance carriers and large insurance organizations with 1,000+ employees.

The buyer here is not the IT leader. Unframe goes after the operational leaders who own the business processes AI can transform: claims, underwriting, benefits administration. They own the P&L and the pain.

✕ Challenge

There was no existing outbound infrastructure for this vertical or buyer. The system had to be built from scratch: who to target, why, how to reach them, and how to measure it.

The real problem was not finding insurance companies. There are thousands. It was working out which carriers are actually ready to buy AI for operations now, so the team spends its first sends on accounts that can convert, not just accounts that are large.

✓ Solution

We built a scored account engine that ranks every target carrier on a 0 to 100 readiness score, then drives targeting and outreach off that score. It runs on three components.

Firmographic signals

Employee count and annual revenue pulled from an Apollo export. Proxies for operational surface area and transaction volume.

Live readiness signals

An n8n workflow runs Serper web searches per company and scores them with Claude against a strict rubric. This is where intent and budget show up, not in firmographics.

Tiering and targeting

Scores are summed and bucketed into three tiers, each with its own outreach posture, then mapped to a two-persona contact strategy inside every account.

★ The scoring model: what makes a carrier a fit

The brief also asked for high-propensity signals beyond basic firmographics. The thinking behind the model matters more than the points, so here is the logic first.

Unframe’s value is getting AI into production. So the ideal account is not just big. It is a carrier that already has the ambition (a named transformation program, AI hires, a CEO talking about AI on earnings calls) but has not operationalized it yet. That gap is exactly what Unframe closes. The model is built to find it.

Two of the five signals are firmographic proxies: company size stands in for the number of claims adjusters, underwriters and admins, which is the surface area Unframe can transform, and revenue stands in for transaction volume, which drives the ROI case. The other three go beyond firmographics and measure readiness and intent directly: active AI hiring, a public DX program, and enterprise tech-stack maturity. The last does double duty, since platforms like Guidewire prove both willingness to invest and the integration substrate Unframe needs.

Each signal scores on a fixed scale and the five sum to a 0 to 100 score. The intent signals (S2 and S3) are deliberately near-binary (20, 13 or 5) because a named program or exec is a real commitment, while the firmographic bands step more gradually.

How the live signals run? A n8n workflow reads the company list from a Google Sheet, loops account by account, fires two Serper searches per company (AI hiring and DX program), and passes the snippets to Claude with a locked rubric that forces a score of exactly 20, 13 or 5 plus one-sentence reasons. Scores write straight back to the sheet. Firmographic signals (S1, S4, S5) come from the Apollo columns directly. The five are summed, and the total drives the tier.

👥 Who to target inside each account

Within a scored account I targeted two personas at once. Hitting both at the same carrier raises the odds of a meeting, because one owns the budget and the other owns the pain.