How Google and AI Actually Work

Author: Caleb Ulku | Published: unknown | Source: transcript

Summary

Caleb’s model of Local SEO: you are trying to rank the Google Business Profile, not the website. Every article, internal link, and sponsored event exists to push that GBP into the top three farther from the pin. Fourth is “invisible” — nobody clicks “More businesses.” The success metric is top-3 percentage on the Local Rank Map & Top 3% Metric.

He restates Local Ranking Factors (Relevance, Proximity, Prominence/Authority) as proximity, relevance, and trust. Proximity (distance from searcher to the GBP address) is uncontrollable. Relevance splits into topical (Google knows what you do) and geographical (Google knows where you serve) — that split is why the course spends so much time on content. Trust, in his mouth, is “just a fancy word for quality links”: Google verifying you are a legitimate business online. Building relevance and trust is how you “overcome the proximity negative” and expand the radius of green.

“Google is math”: no human reads the web; it is a machine-learning system whose goal is maximizing user happiness via goal completion — the faster the user gets what they came for, the better. For “plumber Houston,” that means showing legitimate plumbers who actually serve Houston and making them easy to contact. He calls the practical implication open-book SEO: type the keyword, and the current top 10 are the answer key for what Google wants. Engagement is how Google measures satisfaction; it tests combinations of on-page variables and keeps what improves goal completion. He contrasts this with agencies still “trying to trick Google with complicated schemes.”

Then the AI-search turn. Consumers increasingly find local businesses via AI; ChatGPT is “by far the winner in AI adoption.” Optimizing for other AI bots is, he says, like optimizing for Bing — possible, not where to focus. But because of Microsoft’s partnership with OpenAI, he claims ChatGPT relies on Bing’s index, not Google’s, so a business that wants ChatGPT recommendations needs a Bing for Business / Bing Places profile, not only a GBP. Most local businesses ignore Bing and are therefore “invisible to ChatGPT recommendations.” The same relevance and trust signals (consistent NAP, clear services, verified citations) matter on both engines; businesses in the top three have topical relevance, geographic relevance, and trust across Google and Bing. Tactic: don’t guess — copy what is already ranking and give both engines more of it.

Key Claims

Local SEO ranks the GBP, not URLs; every on-site action is in service of map-pack radius.
Ranking factors: proximity (uncontrollable), relevance (topical + geographic), trust (= quality links / legitimacy).
Success metric = top-3 % of the rank map; position 4 is practically invisible.
Google is an ML system optimizing user goal completion, not a human editor.
Open-book SEO: the current SERP for your keyword is the spec for the page/GBP you should build.
ChatGPT is the AI surface that matters; other assistants are as low-priority as Bing-the-search-engine.
ChatGPT relies on Bing’s index (Microsoft/OpenAI partnership), so you need a Bing Places profile to be recommendable. (contested — see tensions)
The same NAP / service / citation consistency helps both Google map pack and Bing/ChatGPT recommendations.

Notable quotes

“For local SEO, we’re trying to rank the Google business profile, the GBP, not the website.”

Why citable: The course’s one-sentence local-SEO definition, repeated across the YouTube corpus. Use as the canonical “you rank the pin” line.

“Because of Microsoft’s partnership with OpenAI, ChatGPT relies on Bing’s index, not Google’s.”

Why citable: The most load-bearing — and most dated — AI-search claim in the course. Needs a wiki contradiction note against later ChatGPT browsing / native search / Google-index access.

Connections

Entities mentioned: Caleb Ulku, Google, Google Business Profile, ChatGPT Concepts referenced: Local SEO, Local Ranking Factors (Relevance, Proximity, Prominence/Authority), Local Rank Map & Top 3% Metric, Local Link Building & Authority, Super Citations, GEO / AEO (Getting Recommended by AI), GBP–Website Alignment (Consistency Signals)

Contradictions / Tensions

ChatGPT relies on Bing’s index was a reasonable 2023–early-2024 shorthand (Bing grounding in ChatGPT Plus browsing). It is not a complete description of ChatGPT in 2025–2026 (native search, browsing, and non-Bing sources). Flag on GEO / AEO (Getting Recommended by AI) and any Bing Places entity: this claim is the reason he pushes Bing profiles; keep the action (create Bing Places) even if the mechanism is overstated.
Trust = quality links collapses Google’s documented “prominence” (reviews, citations, links, brand mentions) into PageRank-ish links. Elsewhere he treats chambers, super-citations, and reviews as trust. Don’t let this lesson overwrite that broader prominence bundle.
“Fourth position is invisible” is a map-pack UX claim (the 3-pack), not true of organic local packs or “More businesses.” Fine for map-pack CRO; overstated as a ranking philosophy.
“Open-book SEO” / copy the current top 10 can become circular (everyone clones the same SERP). Tension with his own Core 30 advice to build pages for your GBP taxonomy rather than for competitor URL structures.
Optimizing for other AIs is like optimizing for Bing (don’t bother) vs. you must optimize for Bing so ChatGPT can see you. He means Bing-search-share vs. Bing-as-ChatGPT’s-index; keep that distinction or the paragraph contradicts itself.

Notes

Foundational theory lesson for the course; closest analog to the “three glasses” section of the 2025 tutorial. New entity to flag: bing-places. Concepts: GEO / AEO (Getting Recommended by AI) is the right existing page for the ChatGPT/Bing claim. No product demo. Commercial: tees the agency-process lesson as “how we apply this.” Transcript duplicates several sentences (Whisper overlap) around proximity/trust and the plumber-Houston example.