SOURCE

Claude Code Local SEO: How I Got 50,000 Google Clicks/Mo (Steal This)

Properties1
tagslocal-seo, claude-code, gbp, reviews, citations, ssg, next-js, jono-catliff
AUTHOR
Jono Catliff
PUBLISHED
2026-05-26
ORIGINAL
view original

6 min read

Claude Code Local SEO: How I Got 50,000 Google Clicks/Mo (Steal This)

Author: Jono Catliff (NEW) | Published: 2026-05-26 | Source: transcript (70:20, 102K views)

Summary

Jono Catliff’s 70-minute masterclass is a builder’s local-SEO stack: Claude Code (inside Google’s Antigravity IDE) writes the GBP, the posts, the review gate, and a Next.js site; Make.com ships GBP posts and review replies; Semrush picks the keywords. He claims a prior (sold) business ranked #1 for the industry’s hardest term, ~1,500 clicks/day (~50k/month), and >$500k revenue — unaudited. Pillar is both because half the runtime is AI-Assisted Development (SSG vs CSR, Lighthouse, GitHub→Vercel) in service of Local SEO.

He treats 2026 pack ranking as proximity-first and uncontrollable, then four fields that he says are ~80% of what you can control: primary category, services, service areas, reviews. One primary + nine secondary categories; 10–50 services chosen by Semrush volume for service + primary city (skip zero-volume Claude hallucinations); up to 20 service areas within a two-hour drive (SABs / “you go to the customer” only). “BS test”: if it isn’t real, don’t add it — Google is suspending fake/UPS/PO-box/coworking/shared-office listings. Online-not-his research he cites: max the 10/50/20 caps. Other GBP levers: 100+ genuine photos → 520% more calls (stock photos hurt); hours (9–5 hides you from “open now”; optional 24/7 via an AI receptionist); identity attributes for CTR; citations (Yelp/Facebook/Bing/BBB/Yellow Pages NAP) as a trust layer, optionally via Semrush listings. Multiple GBPs only if each is verified, high-quality, and review-fed — otherwise you dilute.

GBP posts are a conversion tool, not a traffic tool, with four claimed side-effects: more engagement → more impressions; AI Overviews can pull posts; 3 months of silence looks closed; a yellow Maps badge if a post matches the query. Voice files (tone.md, humor.md, vocabulary.md, beliefs.md, businesscontext.md) keep Claude from sounding like “AI slop.” Review & Reputation Management: ≥10 reviews as a visibility floor, then ~8/month velocity; consumers trust 4.7–4.9 more than a suspicious 5.0/25. He builds a Next.js review-gate (1–3 → Slack via Make, 4–5 → Google review URL) and auto-replies only 4–5★.

Organic below the pack is a city × service zipper: emergency plumbing Toronto / Mississauga / Brampton…, 10–400 money pages (not 1,000 — thin-content risk), each with a 4–5 keyword cluster. Localized blogs tag the city and poach adjacent intent (“best wedding venues in Toronto” for a photographer) plus remarketing pixels. Only static site generation (pizza-already-sliced); CSR “will never rank.” Copy top-3 competitor averages (word count, H2s, images), inject humor for dwell, paste a Lighthouse mobile report into Claude (demo: 77→82). Wrap it as a /blogpost skill. Deploy GitHub private repo → Vercel (Next.js preset).

Key Claims

  • Proximity is the #1 2026 pack factor and you cannot optimize it; SABs can rank city-wide, storefronts cannot. (aligns with Local Ranking Factors (Relevance, Proximity, Prominence/Authority))
  • Categories / services / areas / reviews ≈ 80% of controllable GBP SEO. Caps he cites from “research online”: 10 categories, 50 services, 20 areas. Semrush-filter services by keyword + main city volume.
  • 100+ real photos → 520% more calls; stock photos hurt. (specific figure, not his study)
  • ≥10 reviews to get shown, then ~8/month velocity; 4.7–4.9 beats a too-perfect 5.0. Review-gate 1–3★ off Google; auto-reply only 4–5★.
  • GBP posts convert existing views; AI Overviews can cite them; 90 days silent looks closed.
  • City × service matrix, 10–400 pages, not a thousand near-dupes. Cluster 4–5 secondary keywords per URL. Localized blogs must include the city.
  • SSG only for Google crawl/index; CSR never ranks. Lighthouse mobile → Claude to chase Core Web Vitals.
  • Citations = NAP trust (Yelp, Facebook, Bing, BBB, Yellow Pages), not a 2017 directory blast.
  • Headline 50k clicks/mo / $500k from a sold business. (self-reported)

Notable quotes

“The number one indicator in 2026 for whether your profile is going to be shown or not is proximity. And this is completely out of your control.”

Why citable: Independent restatement of Google’s triad, matching Caleb’s “three glasses” without using his language.

“If you have 100 plus photos, you can actually get 520% more phone calls… You don’t want to have like stock images.”

Why citable: Same 100-photo threshold Nico (AI Ranking) treats as a “strong correlation,” with a harder (unverified) multiplier.

“You don’t want to create a,000 pages… Google doesn’t like thin content.”

Why citable: He sells a zipper of location pages and warns the thin-content trap — the tension Joy is already seeing in core updates.

Connections

Entities mentioned: Jono Catliff (NEW), Claude, Claude Code, Google, Google Business Profile, ChatGPT (not used; he is Claude-native) · Semrush, Make.com, Next.js, Vercel, GitHub, Homestars, Yelp, Bing, Antigravity, fal.ai Concepts referenced: Local SEO, Local Ranking Factors (Relevance, Proximity, Prominence/Authority), GBP–Website Alignment (Consistency Signals), Core 30, Review & Reputation Management, Super Citations, AI-Assisted Development, Duplicate Content (for Service-Area Pages)

Contradictions / Tensions

  • vs. Caleb Ulku / Core 30 architecture: both are GBP-first (categories, services, reviews, photos, hours). Caleb’s page set mirrors the GBP entity (~30 pages, homepage→category→service, editorial links). Jono’s page set is a Semrush-driven city×service zipper (10–400) plus city-tagged blogs and keyword clusters — closer to programmatic local than entity-mirror. He never mentions rank maps, NAP character-match, or a 7+1 landing-page checklist. Corroborates the ingredients, disagrees on information architecture.
  • vs. Nico (AI Ranking) stack: same jobs, different pipes. Nico = Pabbly + BrightLocal + Claude API on existing sites. Jono = Claude Code + Make.com + Semrush + greenfield Next.js. Strong corroboration on the review safety policy (auto 4–5★, human 1–3★) — now three independent creators. Citations: Nico pays BrightLocal ~$3.20; Jono does manual/Semrush directories (volume-ish, not Caleb’s verification-gated Super Citations).
  • vs. Joy Hawkins on service-area / city-gate pages: Jono will ship up to 400 zipped pages; Joy (2026-08, after this video) calls SAP/city-gate a liability. His own thin-content warning is the unresolved hedge. Do not treat 400 location pages as endorsed by Joy/Caleb.
  • Hours 24/7 via AI receptionist: matches the loophole Joy says still works but is decaying as everyone copies it.
  • Multiple GBPs: Jono says more quality profiles rank better; Joy campaigns against rank-and-rent/fake listings. Compatible only if every profile is a real verified location — his own suspension list (no UPS/PO box/coworking) is the constraint.
  • SSR/SSG language is sloppy: he says “only SSG,” then “SSR or SSG can be read,” then “never CSR.” For Google, SSG/SSR both emit HTML; CSR is the actual killer. Ahrefs’ ChatGPT-crawler-can’t-render-JS point is a different crawler. Keep the CSR warning; don’t cite him as a rendering-architecture authority.
  • Commercial bias: Skool free→paid, 7-calls/week community, agency with $200 show-up guarantee. 50k clicks / $500k / “two seven-figure businesses” are marketing claims.

Notes

Best as an independent implementation recipe (Claude Code + Make + SSG) that corroborates GBP basics and Nico’s review-gate, not as a competing theory of ranking. Do not fold the zipper matrix into Core 30. Flag NEW: jono-catliff.