---
name: fit-scoring-engine
description: Score thousands of industries (or companies, or events) against a rubric to find where your product fits best, then find and score the events and podcasts inside the best-fit markets. Calibrate the rubric on 10, then 100, before scoring everything, and push the bulk work to cheap models. Use when you need a go-to-market target list you can defend.
---

# Fit Scoring Engine

> **Where this came from.** Distilled by Calvin (Colton Mulligan's Claude) from Taylor Thomas's demo of the go-to-market engine behind Sales Shark at AI Open House, Session 001, 13:34 in the recording: https://aiopenhouse.co/sessions/001/?t=814
> This is a reconstruction of the pattern Taylor explained on the call, not Taylor's own file. Taylor promised his list of data sources and APIs; it lands in the library when he shares it. Credit Taylor if you use it. Shared under CC BY 4.0.

## 1. Build the universe

Start from a complete list, not a brainstorm. Taylor started with about 8,000 industry verticals drawn from US Census and Bureau of Labor Statistics classifications.

## 2. Write the rubric

Five to ten criteria, each scored 1 to 5, each with a weight and a one-line definition of what a 5 looks like. Mark the must-haves as **gates**: fail a gate and the score stops there. Taylor's criteria included outbound phone-sales intensity (his product coaches live calls), a creator and community presence to sponsor, and ability to pay.

## 3. Calibrate before you scale (the step that matters)

- Score 10 by hand-checking every result. Adjust.
- Score 100. Read the top and bottom twenty. Adjust weights until the ranking matches what you know is true.
- Only then score everything.

Taylor's biggest lesson: his first rubric over-weighted phone intensity for industries that mostly take inbound calls, which isn't his fit. It only showed up at 10 and 100.

## 4. Score everything, cheaply

- Each result carries: score, a per-criterion breakdown, a confidence level, and a two-sentence rationale.
- Push the high-volume stages to cheaper models through a router like OpenRouter (Taylor uses DeepSeek for throughput), and use a search API for web lookups. Keep the expensive model for judgment calls.
- Taylor's cost lesson: scoring 100 verticals on a premium model ate 27 percent of a week's usage on his subscription.

## 5. Go one level deeper

For the top verticals, find where their people gather: conferences, events, and podcasts (podcast directories and booking sites where hosts ask for guests convert best). Score each against the vertical it serves. Record who to contact to speak or appear.

## 6. Run it nightly

Re-score as new events appear. Keep every rationale so you can click into any score and see why.

## Output

A ranked table you can hand to a salesperson: vertical, score, confidence, rationale, and the three best events or podcasts in it, with a contact for each.