In July 2026 we ran 120 United States managed IT providers through the same AI buyer questions their prospects ask, across ChatGPT with live web search and Perplexity. 112 of the 120, or 93 percent, were not named by either engine when a buyer asked for the best managed IT provider in their city. Their average score on our AI visibility audit was 45 out of 100, and not one of the 120 cleared 75, the score we treat as healthy. If your buyers are asking AI which IT company to hire, the odds today are that yours is not in the answer.
This is our own data, gathered from our own audit engine and our own capture runs. Below is what we found, how we measured it, and what it means if you run an MSP.
The headline number
We tested each provider with two buyer-intent questions, "best managed IT services providers in [city]" and "who should I hire for managed IT services in [city]", on two engines that read the live web. A provider counted as visible if either engine named it in either answer. Only 8 of 120 cleared that bar. The other 112 were invisible.
Per engine the picture is almost identical: ChatGPT with web search named the audited provider in 5 of 120 runs, and Perplexity named it in 5 of 120 runs. Both engines miss the local incumbent about 96 percent of the time and reach instead for a small set of the same well-cited names.
The winners AI names instead
When an engine will not name your company, it still names someone. Across the six metros, the recommendations concentrated on a short list of providers that have done the off-site work AI rewards. In Austin the answers were dominated by IT GOAT, TechProComp, and GCS Technologies. In Tampa it was BCA IT and Symmetric IT Group. In Dallas, Sagiss and Cloudavize. In Charlotte, AT-NET Services and Bit by Bit. In Phoenix, LayerCake Technologies and MyTek. Columbus was the most fragmented metro, with no single dominant local name, where IT GOAT, Realnets, Astute Technology Management, Sabre IT Services, and Arnet Technologies split the answers between them.
One finding stood out. IT GOAT appeared in the answers for 90 of the 120 audits, cited across all six metros, not just its home market. A single firm that invested in citable, structured, well-reviewed presence has become the default AI recommendation for managed IT nationwide. That is the opportunity and the warning in one number: AI answers reward whoever did the work, and right now almost nobody in this field has.
The per-metro breakdown
The invisibility was near-total everywhere we looked. Austin: 27 of 29 providers invisible. Tampa: 22 of 24. Dallas: 20 of 21. Charlotte: 19 of 21. Phoenix: 15 of 16. Columbus: 9 of 9. No metro had more than two providers who were named at all. These are real, staffed, revenue-generating managed IT firms, most of them 10 to 50 people, and to an AI assistant answering a buyer they effectively do not exist.
Why they are invisible: the audit scores
The audit explains the silence. We score six factors that decide whether an AI engine can find, read, understand, trust, and quote a site: crawler access, content legibility, structured data, citation worthiness, live AI presence, and technical hygiene. The 120 providers averaged 45 out of 100. The scores clustered tightly: 22 of the 120 scored in the 20s or 30s, 94 scored in the 40s or 50s, and only 4 reached the 60s. None scored 20 or below, and none reached 70. Sixty-three percent scored under 50.
The most common failures were the cheap ones. Sites built as JavaScript single-page apps returned an almost empty shell to non-rendering AI crawlers. Homepages carried no LocalBusiness or Service structured data, so the engine had to guess what the company was. About pages, phone numbers, and review signals were thin or missing, leaving the model nothing to quote. In several cases robots.txt or a bot-block returned so little text that the audit could barely read the site at all, which is itself the first reason the company is absent from answers.
What this means for MSP owners
The gap is not a quality gap. The winners' on-page work is often no better than the invisible providers'. The difference is off-site and architectural: metro pages with the city in the URL, structured data on those pages, a client-owned comparison post, placement in the directories AI actually reads, and a real review corpus. That is learnable and repeatable, and because 93 percent of the field has not done it, the metro is effectively open. Being named by AI in managed IT right now is closer to a land grab than a ranking war.
Methodology and limits
We audited 120 US managed IT providers across six metros (Austin, Tampa, Dallas, Charlotte, Phoenix, and Columbus) in July 2026. Visibility was measured with two buyer-intent questions per provider on two engines with live web access, ChatGPT with web search and Perplexity. Audit scores come from our own six-factor rubric, scored 0 to 100.
We are honest about the limits. Each visibility check was a single-run spot check, not an averaged panel, and AI answers vary between sessions, so a provider scored invisible could surface occasionally and one scored visible could drop out. We tested two engines, not the full set (Claude, Gemini, Copilot, and Google AI Overviews were out of scope for this run). The audit rubric is ours, not an industry standard. And we anonymized the audited providers because a single spot-check should not brand a real company as invisible in public; the winners are named because AI already recommends them openly to anyone who asks. Directional confidence is high, per-company precision is not the claim. The 93 percent is a snapshot of a young, wide-open field, not a permanent verdict on any one firm.