AI Overviews are recommending your competitors. Not you.

Across the HVAC and plumbing accounts I audited on the North Shore, Google's AI Overview rendered for most service searches. On nearly all of them, the Overview cited competitors and didn't cite the audited business. Four to six regional contractor domains show up in the citations across nearly every audit. If you're a local operator and your domain isn't one of them, you have an answer-engine visibility problem on top of whatever your paid-search account looks like.

This isn't an SEO ranking issue in the old sense. The Overview pulls from sources Google trusts for *answers*, not necessarily the top organic results. So even if your site ranks page one organically, you can be missing from the answer entirely. The audit captures both.

The other thing the audit data shows: the same handful of competitors keep getting cited. They aren't doing anything magical. They have on-page content that answers real customer questions, a clean local footprint, and structured data that matches what's on the page. That's it. The gap is closeable.

The diagnostic for an HVAC site missing AI Overview citations:

Step 1: Pull a sample of 20-30 service queries in the prospect's market. "AC repair Peabody," "furnace replacement North Shore," "emergency hvac near me," "best hvac contractor Boston," and similar. Run each in incognito. Record which queries trigger AI Overview, which domains get cited, and whether the prospect's domain ever appears.

Step 2: For the queries where AI Overview renders + cites competitors, look at the cited pages. What is the page shape? Direct-answer paragraph at the top? FAQPage schema? Author markup? Recent dateModified? The competitors cited repeatedly will share a substrate pattern; that pattern is the answer to "what does my site need."

Step 3: Audit the prospect's site against the substrate pattern. Most HVAC sites fail at one or more of: hero-block-instead-of-direct-answer (the answer is buried below a carousel), Article schema present but author: Organization instead of author: Person, datePublished missing entirely, structured data declared but mismatched to actual page content, /llms.txt absent, FAQPage schema on pages that are not actually Q&A.

Step 4: Check entity coherence. Site name, address, phone (NAP) consistency across the homepage, GBP, structured data, and external citations. AI Overview cross-checks; mismatches reduce citation eligibility. The 2-minute fix here can move citation-eligibility scores meaningfully.

Step 5: Inventory the existing service pages. Generic value-prop pages do not get cited. Pages that answer a specific buyer question with depth (800+ words, named entities, specific math) do. The fix list usually reads: 4-8 service pages need a content rewrite to direct-answer shape; 1-2 new pages need to be authored to cover topic gaps the cited competitors are filling.

Step 6: Check for /llms.txt and structured-data quality. /llms.txt is cheap to ship; absent is a signal the site has not been retooled for AEO. Structured data quality is checked by running the page through Google's Rich Results Test and Schema.org validator; mismatches between declared schema and actual content reduce citation eligibility.

The audit produces a finding count + a prioritized fix list. The substrate work to start showing up in AI Overview citations typically takes 2-8 weeks after the fixes ship; faster on high-authority sites.