Methodology
How MN Agency Index scores agencies
Each agency is scored 1 to 5 on 6 criteria, then weighted to a 100-point total: client reviews (25%), ai visibility (20%), specialization and proof (20%), pricing transparency (15%), certifications (10%), stability (10%). Scores use public sources only. Criteria without data yet are left out rather than estimated, and no agency can pay for placement.
Last updated
Reviewed by John Rice, Minnesota native and AI consultant
Criteria and weights
| Criterion | What we measure | Source | Weight | Status |
|---|---|---|---|---|
| Client reviews | Rating and volume (Google, Clutch and Yelp pooled by review count, shrunk toward the directory median); recency not available | Google Business Profile, Clutch, Yelp | 25% | Live |
| AI visibility | Mention rate: share of recorded AI assistant answers to Minnesota buyer questions that named the agency, relative to the most-named listed agency | AI visibility leaderboard (Cloro API: ChatGPT, Gemini and Google AI Mode x3 runs, Google AI Overview and Microsoft Copilot x1; 40 questions, compare prompts excluded; draft) | 20% | Live |
| Specialization and proof | Published case studies with results in the service or industry | Agency websites | 20% | Live |
| Pricing transparency | Publishes ranges, minimums or packages | Agency websites, directory listings | 15% | Live |
| Certifications | Google Partner, Meta, HubSpot and similar | Partner directories | 10% | Live |
| Stability | Years operating and team size | LinkedIn, Minnesota business records | 10% | Live |
Which scoring criteria are live
Scores use 6 of 6 criteria (100 of 100 weight points), rescaled to 100. Criteria without data are excluded, not estimated. Methodology
- Client reviews · 25%
- AI visibility · 20%
- Specialization and proof · 20%
- Pricing transparency · 15%
- Certifications · 10%
- Stability · 10%
How a score is calculated
Each criterion is scored from 1 to 5 using recorded data, multiplied by its weight, and summed. While some criteria have no data yet, the remaining weights are rescaled so the total is still out of 100. If an agency has no evidence for a criterion that is live for others, it earns no points for that criterion; we never fill gaps with estimates.
Client reviews pool Google, Clutch and Yelp ratings weighted by review count, then pull the result toward the median rating across the directory (4.9) as if every agency started with 10 median reviews. A handful of perfect reviews therefore does not outrank a long record of strong ones. Review recency is not available from our sources, so it is not scored. Glassdoor is an employee rating, not a client rating, and is never part of the score.
AI visibility uses the mention rate: the share of recorded AI answers that named the agency, across the 38 buyer questions that count toward rates (prompts that name agencies in the question are excluded). Because even the most-named agency appears in under a third of answers, the rate is scaled against the most-named listed agency: rating = 1 + 4 × (rate ÷ top rate). An agency no assistant named rates 1; the most-named rates 5. The run is a draft with repeat runs on three assistants; answers vary between runs, so close rates should be read as ties. See the AI visibility leaderboard for intervals and the full data.
Agencies are shown in ranked order only when at least 5 have a score. Below that, lists are alphabetical.
Sources
Agency directory
Seeded from public directory listings, then each agency is confirmed as Minnesota-based and active. Services, industries, size, founding year and reviews are recorded with a source URL.
Client reviews
Google Business Profile, Clutch and Yelp ratings and review counts, each with a link and the date it was fetched. Glassdoor employee ratings are shown on profiles for context but are not scored.
AI visibility
40 buyer questions asked in ChatGPT, Gemini and Google AI Mode (three runs each) and Google AI Overview and Microsoft Copilot (one run each) via the Cloro API; we record which agencies are named, in what order, with what sentiment, and which sources are cited. Perplexity and Grok returned server errors on every attempt and are not included.
Pricing
Published rates, minimums and packages on agency websites and directory listings.
Certifications
Google, Meta, HubSpot and similar partner directories.
Stability
Years operating and team size from LinkedIn and Minnesota business records.
Every listing on the site links to the source URLs it was built from.
Refresh schedule
| Dataset | Refresh | Last updated |
|---|---|---|
| AI visibility leaderboard | Quarterly | October 12, 2026 |
| Agency directory | Quarterly | October 12, 2026 |
| Pricing benchmark | Yearly, with a mid-year check | October 12, 2026 |
Who reviews the index
John Rice
Editor, MN Agency Index
John Rice is a Minnesota native and AI consultant who helps Minnesota businesses put AI to work. He has 15 years in software across startups, agencies and enterprise, and builds tools that track how brands show up in AI search.
John reviews the criteria, weights and every data refresh before it publishes. Scores come from the public sources listed above, not from his opinion of any agency.
No paid placement
Agencies cannot pay to be listed, ranked, featured or removed. There are no sponsored positions. Rankings change only when the underlying public data changes.