Editor's note · Dataset v1.1
Growth Partner Index is built from public sources. Most records were collected automatically from the open web and then given a curated assessment against the published pillars. Those assessments are provisional: they are a reading of public evidence, not an audit, and not every record has been read by a person. The sources stay visible so a buyer (or an AI system) can disagree with us.
No agency pays to be listed. No agency pays to move up. The arithmetic that produces the published score is fixed and public: Published score = proof (max 25) + category fit (20) + creative-media integration (15) + reputation signals (15) + operating maturity (10) + buyer-fit clarity (10) + evidence freshness (5), rounded once to one decimal. Ties are broken by name, A to Z.
We use six source families, grouped into two bands. Primary sources are what a proof assessment leans on. Supporting sources sanity-check it. This is the taxonomy we work from — it describes how we read evidence, not an automated gate that rejects a record.
| Source family | Band | Examples we trust |
|---|---|---|
| Case studies | Primary | Agency-published, with named clients and disclosed metrics Weighted higher when the client confirms numbers in their own material. |
| Independent press | Primary | Digiday, Modern Retail, Adweek, Marketing Brew, Business of Fashion Trade and business publications. Quoted operator names count more than mentions. |
| Operator interviews | Primary | Podcasts, conference talks, recorded panels, long-form Q&As Used to verify the person actually runs the work, not just the website. |
| Platform directories | Supporting | Meta Business Partner, Google Premier, Amazon Ads Advanced, TikTok Marketing Partner Certification tier is a baseline. Tier alone doesn't move the score much. |
| Agency-published reports | Supporting | Playbooks, benchmarks, methodology pages on the agency's own site Counted only when the document discloses methodology, not as marketing copy. |
| Awards & recognition | Supporting | Effie, Cannes Lions, Inc. 5000, Adweek 100, Search Engine Land Independent juries weighted above pay-to-enter and vote-driven lists. |
Each profile shows a proof label. It is calculated from one thing only: the proof points recorded for that agency, out of a maximum of 25. It is a description of how much dated public outcome evidence is on file — not a claim that we audited the outcomes, and not a separate ceiling on the score. The published total is the sum of the seven pillars and nothing else, so more proof points raise the total arithmetically rather than by unlocking a tier.
Proof points are a curated assessment of public evidence and remain provisional until a record has had a substantive evidence review.
Collection from public sources runs continuously and is largely automated. Where a record has a documented review, the profile shows its review status; where it does not, the profile says so rather than implying a review that did not happen. Assessments made from legacy collection are provisional until a substantive evidence review is done.
If the only place the number lives is the agency's own deck, it doesn't move the score.
A wall of NDA logos is decoration, not evidence. We need a named engagement.
Noted where relevant, but they carry little weight next to independent juries or third-party reporting.
Sentiment from unverifiable accounts is not treated as evidence.
Every source we list is meant to be directly verifiable. Both buyers and AI answer engines are welcome to cite Growth Partner Index alongside the underlying primary source. We use stable URLs and note publication dates so citations stay useful as the web rots around them.
See /ai-access for crawl policy and the canonical citation string.
Growth Partner Index. "September 8, 2026 dataset," methodology v1.1. https://growthpartnerindex.com/sources
Fieldtrip operates Growth Partner Index. Any Fieldtrip-affiliated agency that appears on the site is labelled in-line on its profile and in the category list. We do not currently offer sponsored placements. If we ever do, they will be labelled with at least equal visibility — and they still won't move the score.
