Intro

Hi, I'm Adam. I'm an independent builder & industry analyst. My goal is to help GTM leaders make sharp decisions in uncertain markets.

This is my methodology for the GTM Index — the score, the rankings, and the category maps. Three principles guide this work:

1. Mindshare matters. We can gain an edge by understanding the conversations and perceptions within our market. My scores are based on share of voice (SOV).
2. Curation is judgement. My judgement is encoded into this research system by curating the panel and coverage. That's what makes this different from vanilla Claude queries on GTM tech.
3. Everything must update and evolve constantly. Things change fast in this market, so my research and data collection is continually running and refreshing.

Coverage

I cover products and companies with a central application for B2B go-to-market in the AI era. Most are sales- or marketing-focused (e.g. Gong, Clay), but I also cover broader solutions that are clearly GTM-relevant (e.g. Claude Code, Vercel, Exa).

On the edges my judgement determines what fits (and what doesn't). Like all things, this evolves as I learn and as the market changes.

SOV Score

Brand's share of panel conversation, normalized 0–99 against the field.

My scores are a share-of-voice (SOV) measure, not a rating of the product capabilities. I am not scoring whether a product is good or best. I'm measuring how much it shows up in the GTM conversation. This a useful concept to understand market trends, but isn't a buyer score alone.

Here's how it works:

1. Curated Panel. Sources I've hand-picked across social media, newsletters, blogs, and podcasts. Not a broad sample — the best minds in GTM: operators, authors, builders, executives.
2. Tracking Mentions with Context. I track substantive product mentions — and their context — across the panel. Link-only and content-locator references don't count as brand conversation.
3. Engagement Weighted. Each mention is weighted by engagement and relevance (the factors vary by mention type). A product's own posts don't count — the score is earned voice only.
4. Normalized 0–99. Every product sits on one scale, indexed to the loudest product in the window — that product scores 99 and everything else is placed against it. The index follows a power curve, not a straight line.

The two windows

The move is a percentage change in share of panel conversation — the brand's share over the last 60 days against its share over the 60 days before that. Both are recalculated nightly, against the panel as it stands that night. The move window ends two days back, because the most recent day or two of collection is still filling in. Scores you see today are as of 2026-08-16.

What the number is not

• It is not a percentage. A SOV Score of 40 does not mean 40% of anything. It's a position on a 0–99 curve where 99 is whoever the panel talked about most in the window.
• It is not linear. The curve compresses the top and spreads the bottom, so the gap between 90 and 95 represents far more conversation than the gap between 20 and 25. At the very bottom the scale is steep: roughly, SOV 9–11 is one earned mention over 90 days, 14–16 is two, 19–21 is five, 24–26 is nine.
• It is not a quality rating. A product can be excellent and quiet. That shows up here as a low score, and that is the measure working, not failing.

What counts as a voice

A voice is a source on the curated panel. It's a person or a publication whose GTM commentary I've chosen to track, across LinkedIn, X, newsletters, blogs and podcasts. It is a carefully curated list or experts and operators, not a crawl of the internet. I have three rules for reading the panel.

• SOV is earned only. A brand's own accounts, and posts by its own staff, are identified and excluded.
• Weighted by engagement. A post nobody read counts for less than one with a strong response.
• Context-checked. Noise like common-word brand names, and other irrelevant mentions are scrubbed when they aren't associated with a GTM conversation.

Markets, membership, and the ranking cutline

Brands sit inside Markets. These are my framework for grouping companies by the AI era "races" and don't necessarily align to traditional industry categories. Two different things can be true of a brand at once:

• Membership is permanent. Once a brand belongs to a Market it keeps that Market, whatever its score does. Membership is about what the company is.
• Ranking is dynamic. A brand is ranked once its SOV Score clears its Market's floor — currently SOV 15. Below that it keeps its Market and stays visible on the page, marked unranked.

Category Definitions

GTM Tech Types

My highest level organizing framework has a strong bend toward AI. I believe the traditional categorization is breaking down quickly. This is how I see the space at a high level:

GTM Tech Categories

Categories are generally organized around use cases, go-to-market motions, or the intersection of the two. That said, there is a balance of many factors in this framework and it ultimately comes down to my judgement.

Rules & Guidelines

Products can fit types and multiple categories, but there must be a strong case.

Agent Readiness Score

I score Agent Readiness as a secondary factor, but it matters: in an agent era, whether your systems can actually build on a product matters more than its UI. It's an analyst-judged 0–99 score (with an A–F grade) across four surfaces — API, MCP server, SDK, and CLI — plus the quality of the docs behind them. Multiple agents, each with a different persona, assess those surfaces and docs; the Surfaces column shows exactly which a product ships.

Topic Tracking

Topics measure what the GTM conversation is actually about, at two grains. A cluster is an editorial theme — a named argument I think is worth following. A topic is a measured term: a phrase and brand matcher I maintain, with a definition and a boundary. Clusters are built from topics, so both appear in one ranked list.

Rank is calculated within each grain and hidden in the combined view. A cluster and its member topics describe overlapping evidence, so a single mixed ranking would count the same conversation twice.

Where a cluster is named after one of its topics — GTM Engineering the theme contains GTM Engineering the phrase — that term is shown inside the cluster rather than twice in the list. I call it the cluster's head term, and the gap between the two numbers is worth reading on its own: it tells you whether a conversation has settled on a name yet. A high share means the phrase is the conversation. A low one means the argument is real but nobody has agreed what to call it.

Share is engagement-weighted across every maintained topic, so an item matching two topics counts once in a cluster and twice in the overall denominator. Clusters deduplicate their members' evidence before reporting.

Feedback & Corrections

Send any suggestions or issues to adam at adamgtm.com. I verify them, write the fix back to the system and publish.

If you're building something interesting in this space, but it's not covered yet, please suggest a product here.