Minnesota Local SEO Statistics 2026: We Analyzed 1,058 Small-Business Websites

Most local SEO statistics you'll find are recycled national numbers from five years ago. So we built our own dataset.

In August 2026, we analyzed 1,058 Minnesota small-business websites — every member business listed in three Minnesota chamber-of-commerce directories — and then asked a second question almost nobody is measuring yet: when an AI assistant recommends a local business, whose website does it actually cite?

The short version: Minnesota's small businesses are in Google, but they're not in the answer. Here's the data.

The headline numbers

Finding	Number	Sample
Business websites with no LocalBusiness schema	87.7%	922 live sites
Sites scoring "poor" on mobile performance (Lighthouse < 50)	74%	94-site sample
AI citations that went to directories, not the business	55%	29 citations, 6 grounded queries
Member websites that are dead or unreachable	12.9%	1,058 sites
Live sites with no visible phone number	41.3%	922 live sites
Chamber directory links that are broken	9.1%	3,258 links

1 in 8 member websites is simply gone

Of the 1,058 member businesses listed in the three chamber directories, 12.9% had websites that were dead or unreachable — domains that no longer resolve, homepages returning errors, sites that have quietly disappeared while the listing lives on.

Related: 9.1% of the 3,258 outbound links in those chamber directories point at dead or broken destinations. Directory rot is real, it's measurable, and it silently strands customers every day. (If the web's overall decay is your interest, Pew Research found 38% of pages from 2013 were gone by 2023 — local business sites appear to churn even faster.)

87.7% are invisible to structured search — and to AI

This is the number that should worry every Minnesota business owner.

Of the 922 live websites we analyzed:

• 87.7% have no LocalBusiness structured data — the machine-readable block that tells Google (and increasingly, AI assistants) what the business is, where it is, and how to reach it.
• 34.4% have no JSON-LD structured data of any kind.
• 30.5% have no meta description, and 30.8% have a title tag under 30 characters — often just the business name.
• 27.7% don't have a single H1 heading.

None of this is exotic optimization. It's the basic markup layer that search engines and AI systems read before deciding who to show. Which connects directly to the next finding.

When AI recommends a local business, it usually cites everyone but the business

This is the measurement almost nobody is doing yet. We asked a live, web-grounded frontier model (via API) the questions real customers ask — "Who is the best plumber in Eagan, Minnesota?", "Best HVAC repair company in Rosemount MN", "Who should I call for tree removal in Apple Valley?" — and analyzed every source the AI cited in its answers.

Across 6 local hiring questions and 29 citations:

• 55% of citations went to third-party directories and platforms — BBB, Expertise.com, Angi, HomeAdvisor, booking platforms, Reddit.
• Only 41% pointed at a recommended business's own website. (The rest: manufacturer sites.)

Put plainly: the AI answered "who should I hire," and more than half the sources it leaned on weren't the businesses at all. The businesses that did get cited directly had one thing in common — websites with enough real, readable content about their service and location for the AI to ground on.

The mobile performance picture is worse than the SEO picture

We ran Google Lighthouse mobile audits across a random 94-site sample of the same population:

• Median mobile performance score: 35 / 100.
• 74% scored under 50 — Google's "poor" band.
• Only 7.2% of the full population lacked a mobile viewport tag — so these are "mobile-friendly" sites in the 2015 sense that are slow in the 2026 sense.

Page speed is a confirmed ranking input and a brutal conversion input. A median score of 35 means the typical Minnesota business website loses mobile visitors before it finishes loading.

The twist: these businesses aren't invisible in Google — they're unchosen

Here's the finding that surprised us, and we're reporting it straight because it cuts against the easy narrative.

Using ranking data for a random 138-site sample of the population:

• 0% ranked for zero keywords. Essentially every established business ranks for something.
• Only 7% ranked for fewer than 10 keywords.
• The median site ranked for 707 keywords — mostly brand terms and incidental long-tail phrases, counted anywhere in the top 100.

So "small businesses are invisible in search" is a myth — at least for established, chamber-member businesses. The real problem is sharper: they're indexed but not chosen. Showing up at position 40 for 707 scattered phrases wins nothing. The layers that decide who actually gets the call — the local pack, the top 3, and now the AI answer — run on exactly the signals this study shows are missing: structured data (87.7% lack it), performance (74% poor), and a citable, readable site.

What this means if you run a Minnesota business

1. 1

   Check whether you're in the 87.7%. View your homepage source and search for "LocalBusiness". If it's not there, adding schema is hours of work with an outsized payoff in the AI era. Our free schema generator does it.

2. 2

   Search yourself the way AI does. Ask an AI assistant "best [your service] in [your city]" and see who gets cited. If the answer routes through Angi or BBB instead of you, that's your gap — and your competitors' gap too, which means it's still winnable.

3. 3

   Take the mobile score seriously. Run PageSpeed Insights on your phone URL. If you're under 50, you're in the 74% — and speed is one of the few factors where fixing it helps rankings, conversions, and AI citability at once.

4. 4

   Make the basics machine-readable. Visible phone number (41.3% don't have one), real title tag, meta description, one H1. This is the floor, and a third of your local competitors aren't on it.

Methodology

Population. All member businesses listed in the online directories of three Minnesota chambers of commerce (a regional metro chamber, a statewide chamber, and a county regional chamber), collected August 2026. After excluding national brands, institutions, government/education entries, and non-website links (newsletters, documents, social profiles): 1,058 unique business websites.

On-page factors (n = 1,058). We fetched each site's homepage directly (standard desktop browser user-agent, following redirects) and parsed the served HTML for: reachability, HTTPS, JSON-LD presence, LocalBusiness-family schema types, title tag and length, meta description, mobile viewport tag, H1 presence, and a visible phone number pattern. Sites unreachable after retries were classed dead/unreachable (12.9%).

Directory link rot (n = 3,258). Every outbound link in the three directories was probed and classified (HTTP 404/410, soft-404 by content analysis, connection/DNS failure). 9.1% were dead or broken.

Mobile performance (n = 94). Google Lighthouse mobile audits (performance + SEO categories) via the DataForSEO API on a random 100-site sample of live sites; 94 returned complete scores.

Rankings (n = 138). Ranked-keyword totals (any position ≤ 100, US Google) via DataForSEO Labs on a random 150-site sample; 138 returned data.

AI citations (n = 6 queries, 29 citations). Live, web-grounded queries to a frontier model (GPT-5.5 with web search, via the DataForSEO AI Optimization API) using natural local hiring questions across Minnesota cities and service categories. Every URL cited in the answers was classified as the business's own site, a third-party directory/platform, or a manufacturer site. We label this a probe, not a survey: the sample is small and we'll expand it in future updates.

Limitations, stated plainly. This is a single-state population of chamber-member businesses — established enough to pay dues, which likely makes these numbers better than the true small-business average. Ranked-keyword totals count any top-100 position and therefore overstate meaningful visibility. The AI-citation probe is directional. We prefer honest, small, well-defined samples over impressive vague ones; we'll grow the dataset in future revisions.

Cite this research

Referencing these statistics? Link to this page as the source. For questions about the data or methodology, contact us. This page is updated as we re-run the research — the date at the top reflects the most recent revision.