TL;DR
The numbers from the May 2026 round of local-search visibility tracking landed unambiguous: AI Overviews appear on 68% of local intent queries in U.S. and major EU markets. That's up from ~24% twelve months ago. The classic Map Pack — three local listings above the blue links — now sits below the AI Overview in most cases, and on mobile is often pushed off the first screen entirely. For SMBs, this means the visibility battle has moved from "rank in the top 3 of the Map Pack" to "be one of the 2–4 businesses cited inside the AI Overview itself." The four signals that determine AI Overview citation: review density and recency, structured-data quality on your own site, third-party platform presence (G2/Trustpilot/Yelp/Reddit), and entity match consistency across the open web. The 30-day plan: reviews flywheel, schema audit, Trustpilot/G2 claim, NAP cleanup. Same flywheel, higher leverage than ever.
The 68% number, in context
PinMeTo's 2026 ranking factors report tracks AI Overview appearance rates across 12,000 monitored local queries spanning restaurants, services, healthcare, and retail. The May 2026 cut showed AI Overviews appearing on 68% of queries with local intent (queries containing geo-modifiers like "near me," city names, or implicit local context).
The same dataset shows the position shift: in queries where AI Overviews appear, the classic Local Pack drops to position 4–6 on average. On mobile screens (where 73% of local queries originate), the AI Overview alone occupies the first viewport. Users tap the businesses inside the AI Overview at roughly 2.4x the rate they tap businesses in the demoted Map Pack below.
The practical translation: if you're not cited in the AI Overview, you've lost the majority of the click stream — even if you rank #1 in the traditional pack.
What an AI Overview cites (and why)
When Google's AI Overview answers "best brunch spot in Lisbon Príncipe Real," it composes its response from a synthesis of:
- Google Business Profile data — name, address, hours, services, top review themes
- Review text content from Google — recent reviews, weighted heavily on the last 90 days
- Third-party listings and reviews — Yelp, Trustpilot, TripAdvisor, Reddit threads
- Your own website's structured data — LocalBusiness JSON-LD, Review schema, FAQPage
- Entity-match signals — consistency of your name/address/phone across the web
The output is typically 2–4 named businesses with one-sentence rationales drawn from this synthesis. AI Overviews don't cite by raw ranking — they cite by citation density: businesses that show up consistently across multiple sources get the nod.
The 4 signals, ranked by leverage
Signal 1: Review density × recency. A profile with 200 reviews from 3 years ago loses to a profile with 60 reviews where the last 12 came in the last 90 days. Recency is weighted harder than total count in 2026. Practical floor: 5–8 new Google reviews per month, indefinitely.
Signal 2: LocalBusiness JSON-LD on your site.
AI Overviews use structured data as the canonical entity record. A site without LocalBusiness schema is, to the AI, less trustworthy than one with it — even if the content is identical. The schema fields that matter: name, address (full PostalAddress object), telephone, geo (latitude/longitude), openingHoursSpecification, priceRange, image. Add aggregateRating and review if you have real ones, but never fake them — AI Overviews now cross-validate against Google reviews and dock entities with mismatched ratings.
Signal 3: Third-party platform presence. The 3x citation multiplier from G2/Capterra/Trustpilot/Yelp isn't theoretical — it's measurable in citation tracking. AI Overviews pull from these platforms regularly because they're recent, structured, and crawlable. A B2B SMB without a G2 profile is invisible in 50% of relevant AI Overview queries.
Signal 4: Entity-match consistency. Your name, address, phone, and category line up across Google, Yelp, Apple Maps, Bing, your website footer, and any directory listings. Inconsistencies (different phone format, "Inc." in one place but not another, an old address that's still cached somewhere) reduce entity confidence and citation likelihood.
What no longer moves the needle
Three former staples that 2026 AI Overviews have stopped weighting:
- Backlinks from generic directories. The 2010s SEO playbook of paid directory links does almost nothing for local visibility in 2026.
- Keyword-stuffed business names. Google's 2026 GBP spam crackdown is suspending profiles for this (locksmiths, movers, contractors hardest hit). Keep your legal business name, period.
- Generic press releases. PR mentions in publishers that AI Overviews don't cite (WSJ, NYT, Bloomberg — all absent from the top 20 citation sources) produce essentially zero visibility lift.
The 30-day plan
Week 1 — measure and clean.
- Audit your current AI Overview citation rate on 10 representative queries for your business. Use a manual incognito-mode check or a tool like Profound/Searchwell.
- Run a NAP audit across Google, Yelp, Apple Maps, Bing, and your site footer. Fix inconsistencies.
- Inventory your current structured data with Google's Rich Results test.
Week 2 — schema and platforms.
- Add or polish LocalBusiness JSON-LD on your homepage. (Template: see LocalBusiness Schema for AI Overviews — separate post.)
- Claim your Trustpilot, G2/Capterra (B2B), or Yelp (consumer) profile if not done.
- Set up a single review request flow that targets 5–8 fresh Google reviews per month.
Week 3 — content for citation.
- Write or refresh 2–3 service-page sections with the exact natural-language phrases customers would use ("Houston dental implants under $3000," "vegan brunch in Lisbon Príncipe Real"). AI Overviews pull these as quotable answer fragments.
- If you're a B2B SMB, post 1 substantive answer on a relevant Reddit thread (no link, no pitch — just be helpful).
Week 4 — measure again, iterate.
- Re-run your AI Overview citation check on the same 10 queries.
- If you went from 1 → 3 citations, the system is working. Continue.
- If you went from 1 → 1, the issue is usually entity match — keep cleaning NAP and add 3 more Trustpilot/G2/Yelp reviews.
The Map Pack isn't dead. It's just no longer the headline. The headline is whether you appear in the AI Overview above it. The flywheel that gets you there is unglamorous — reviews, schema, platforms, consistency — but it's the only one that compounds.
In 2026, the businesses that win local search are the ones that built the citation flywheel in 2024 and kept it spinning. The cost to start one in 2026 is higher than it was in 2024, but lower than it will be in 2027.
Start this month.
New blog posts. No spam.
Get the next reputation playbook delivered when it drops.
