Research · edition 2026-q3

The Front Range AI Visibility Index.

Of 66 owner-led Front Range businesses we audited, 33 — exactly half — are invisible, partial or compromised in AI answers.

Published 2026-09-09 · fieldwork 2026-06-14 to 2026-06-16 · ungated, no signup

What we did

Between 2026-06-14 and 2026-06-16 we audited 70 businesses across seven batches, four Colorado cities and five verticals. Most are owner-led; a handful were included as market benchmarks, one of them macro-owned, and Rule B below takes them out of the denominator. For each one we ran customer-style prompts for its city and category against AI answer engines, recorded which businesses were named, and graded the business on seven dimensions.

1. Website Quality
2. SEO Basics
3. Social Media Presence — except in the law batch, where it is Authority & Content (60 files vs 10)
4. Review Profile
5. GEO Visibility Check
6. Top 3 Gaps
7. Opportunity Estimate

Note the third dimension. It is Social Media Presence in 60 of the 70 scan files and Authority & Content in the ten law-office files. We are not going to smooth that over: the law batch measured something slightly different there.

How a verdict became a grade

The seven batches recorded their verdicts as 30 different free-text strings — "⚠️ Partial / fragile", "✅ Strong (site-throttled)", "❌ Invisible for Boulder". Any headline number therefore depends entirely on how those strings were collapsed onto one scale. So here is the rule, before the number.

A1. "Highly Visible" is Batch 1's own wording for the top of the sellable range and maps to Strong. "Visible", also Batch 1's, is kept as its own tier — it appears once and we do not silently promote it.
A2. A cell naming a trajectory ("Moderate→Strong") is coded at the CURRENT tier — the one named first, as written.
A3. A cell naming two tiers ("Partial/Invisible") is coded at the WORSE of the two. Where the text and the status emoji disagree, the text wins.
A4. Parenthetical qualifiers ("(niche)", "(broken booking)", "(site-throttled)") describe the cause, not the grade, and do not change the tier.

And the rule for who counts in the denominator. Some audited businesses were included as market benchmarks — the dominant name everyone else is measured against — rather than as potential clients. Counting them would flatter the result.

B1. A numbered row is excluded if its verdict cell or business name marks it as a benchmark, macro-owned, a chain, or explicitly not a target.
B2. A row explicitly marked "(independent)" is RETAINED even when graded Dominant — an independent business that already wins is a real result, not a benchmark.
B3. The three unnumbered comparison rows (New Belgium, Woodhouse Spa, and the Burnham/Zinda/Sloat group) were never targets and are excluded before counting. 73 glance rows − 3 unnumbered = 70 audited businesses.

Applying Rule B excludes 4 numbered rows: Avery Brewing, Frasca Food and Wine, Odell Brewing, Longmont Climbing Collective. That leaves 66 businesses we could actually have sold to, of which 33 (50.0%) graded Invisible, Partial or Compromised.

The recoding table

This is the table that makes the headline falsifiable. Every source string, the tier it was mapped to, and how many times it occurred. Disagree with a row and you can recompute the number yourself.

Source verdict string	Coded as	n
`⚠️ Partial`	Partial	12
`✅ Moderate`	Moderate	10
`❌ Invisible`	Invisible	8
`✅ Strong`	Strong	6
`🛑 Compromised`	Compromised	4
`🏆 Dominant (benchmark)`	Dominant	3
`🟡 Moderate`	Moderate	3
`❌ Invisible for Boulder`	Invisible	2
`⚠️ Compromised`	Compromised	1
`⚠️ Partial (German niche)`	Partial	1
`⚠️ Partial (enthusiast only)`	Partial	1
`⚠️ Partial (enthusiast)`	Partial	1
`⚠️ Partial / fragile`	Partial	1
`⚠️ Partial/Invisible`	Invisible	1
`✅ Highly Visible`	Strong	1
`✅ Moderate→Strong`	Moderate	1
`✅ Strong (#2)`	Strong	1
`✅ Strong (Yelp best-of)`	Strong	1
`✅ Strong (brand split)`	Strong	1
`✅ Strong (broken booking)`	Strong	1
`✅ Strong (broken site)`	Strong	1
`✅ Strong (contested)`	Strong	1
`✅ Strong (niche)`	Strong	1
`✅ Strong (site-throttled)`	Strong	1
`✅ Visible`	Visible	1
`❌ Partial/Invisible`	Invisible	1
`🏆 Dominant`	Dominant	1
`🏆 Dominant (independent)`	Dominant	1
`🏆 Dominant (macro-owned)`	Dominant	1
`🟡 Moderate (PI)`	Moderate	1

The distribution

Tier	Businesses
Partial	16
Strong	15
Moderate	15
Invisible	12
Dominant	6
Compromised	5
Visible	1

By vertical and by city

Sellable businesses only, after Rule B.

Vertical	Audited	Weak	Share
Breweries	28	15	54%
Fitness	9	5	56%
Law offices	10	5	50%
Restaurants	9	4	44%
Salons & spas	10	4	40%

City	Audited	Weak	Share
Boulder	38	17	45%
Fort Collins	9	6	67%
Golden	10	5	50%
Longmont	9	5	56%

Per-city and per-vertical counts are small — nine or ten businesses in most cells. Read them as texture, not as rankings between cities.

What this Index is not

This is a census of 70 businesses we chose, not a random sample. Businesses were selected as plausible clients in a target city and vertical, so the distribution describes the owner-led Front Range field we sell into — not all Front Range businesses.
Grades were assigned by a human analyst reading answer-engine output during a three-day window in June 2026, not by a repeatable automated harness. Answer engines are non-deterministic; a re-run would not reproduce every row.
The underlying scans record which businesses answer engines named. They do not record which URLs the engines cited. That is the more valuable question and we have not yet answered it — it needs a column the pipeline does not currently capture.
Review counts in the source scans lean on Yelp, Avvo, Birdeye and other proxies where Google counts were not retrievable, and are labelled as estimates in the batch files.
No market-size or opportunity figure appears anywhere in this Index. We have no measured supply ceiling for any of these markets, and the per-batch target of ten businesses is a pipeline default, not a finding.

We are not naming the businesses that scored badly

Every business in this Index is a real, named local business, and most of them are potential clients. Publishing a list of which Boulder salon or which Fort Collins brewery is invisible would be a cheap way to get attention at their expense. The distribution is the finding; the names are not ours to spend.

Cite this

Front Range AI Visibility Index, edition 2026-q3. NovaSapien Labs, 2026-09-09. https://novasapienlabs.com/research/front-range-ai-visibility-index

Questions

How many Front Range businesses are invisible in AI answers?

Of the 66 owner-led Front Range businesses we audited and could sell to, 33 — exactly half — graded Invisible, Partial or Compromised in AI answers. We audited 70 businesses in total across seven batches; four were excluded from that denominator as market benchmarks rather than potential clients, under a rule published on this page.

Which industries were audited?

Five: breweries, fitness, law offices, restaurants, salons & spas. Across four cities — Boulder, Fort Collins, Golden, Longmont — in seven batches of ten businesses each, between 2026-06-14 and 2026-06-16.

How were the grades assigned?

A human analyst ran a fixed set of customer-style prompts for each business's city and category, read which businesses the answer engines named, and recorded a verdict per business. Those verdicts used 30 different free-text strings across the seven batches, so they were recoded onto a single seven-tier scale using the rules published on this page. The full mapping from every source string to its tier is in the recoding table below.

Is this a random sample of Front Range businesses?

No, and the distinction matters. These are 70 businesses we selected as plausible clients in a target city and vertical. The result describes the owner-led field we sell into, not all Front Range businesses, and it should not be read as a population statistic.

Scan your own site How we score