# TAM Sizer

> Size the total addressable market for an ICP against ZoomInfo's company database. Iteratively refine the firmographic and technographic filter set until the account universe matches intent, then save the working filter for use in other skills.

**Source:** https://gtm.ai/marketplace/gtm-skills/zoominfo-tam-sizer

---

## Overview

TAM Sizer iteratively refines a company-level ICP filter set against ZoomInfo's company database. Each pass returns a headline count, a banded sizing read (too broad, sweet spot, too narrow), two labeled sample views, and concrete refinement options. The skill terminates when the user finalizes, and the output is both the count and a structured filter-set artifact that Build List, Score Accounts, and Find Similar can all consume.

The skill guards against common sizing failures. Industry terms get resolved against ZoomInfo's taxonomy via Lookup, and a missing match is surfaced as a taxonomy gap rather than silently falling back to a noisy keyword search. When funding or revenue filters collapse the count to near zero, the skill probes with and without the filter to distinguish a narrow ICP from a data-coverage gap. When the directional sample shows 20% or more off-target rows, the count gets noise-adjusted and re-banded.

Persona criteria are recorded but not applied to the count. Buyer personas describe who you sell into, not who the account is. Persona discovery happens later through Build List or Search Contacts once the filter set has settled.

## What It Does

- **Banded sizing read**: Classifies the count as too broad (>50k), healthy, sweet spot, tight, or too narrow, with operational implications for each.
- **Taxonomy-gap gate**: Industry terms resolved via Lookup, and a missing match surfaces explicitly rather than silently keyword-searching.
- **Data-sparsity probe**: When funding or revenue filters drop the count by 90%+, runs both filtered and unfiltered counts and treats the larger as operative TAM.
- **Two sample views**: Trophy view (default ranking, biggest logos) and Anchor view (sorted by headcount) for sanity-checking the filter set.
- **SAM hypothesis**: Optional layer that multiplies TAM by an addressable fraction and ARPA to produce a labeled SAM hypothesis.

## Use Cases

### Sharpening an ICP

When the ICP is fuzzy and the question is whether it's too broad or too narrow, TAM Sizer takes a natural-language description and returns a count plus a banded read in one pass. Refinement options name the specific dimension to tighten or loosen, with an estimated post-refinement count for each.

### Territory and Capacity Design

For territory design, run the skill with the territory filter set and the desired tier segmentation. The sample views surface whether the territory contains the expected logos, and the saved filter set drops into Build List for the actual rep-assigned account list.

### Investor-Ready Market Sizing

For an investor-ready sizing, the SAM hypothesis layer takes an addressable fraction and ARPA and labels the output as hypothesis, not forecast. The subsidiary-records caveat is surfaced explicitly, so the count reflects the relevant level of corporate hierarchy rather than double-counting parent and subsidiary.