Notes from the lab.
We learn, research, synthesise, and pressure-test it in the real world. These are the notes we’ve gathered along the way — from Jakarta.
One Source, Three Numbers: Check the Denominator
The same source can be 1.23%, 6.4% or 14.7% of ChatGPT's citations depending on what you divide by. A short guide to reading AI citation statistics, using Reddit in SOLEDAD's data as the worked example.
YouTube Shows Up in Google AI Answers, Not ChatGPT
In SOLEDAD's data, YouTube is the top cited domain in Google's AI Overview and AI Mode and almost absent from ChatGPT, Copilot and Gemini. What we counted, what our own prompts show, and how to read it before you put video in a plan.
ChatGPT Visits Can Hide in Your Direct Traffic
Google Analytics sorts ChatGPT visits by the link's tag and referrer. When both go missing, the visit counts as Direct. What our own monitoring saw in October 2026, and how to check your reports.
Google-Extended Doesn't Control AI Overviews
Blocking Google-Extended in robots.txt keeps your content out of Gemini training and grounding, not out of AI Overviews or AI Mode. Those follow Search's own snippet controls. Which switch does what, on Google and beyond.
Measure AI Visibility in Bahasa First
For Indonesian brands, AI visibility should be measured mainly on Indonesian-language prompts, with English as a deliberate extra. Google reports 84% of Gemini prompts in Indonesia are in local languages.
Query Fan-Out: The Questions Behind AI Answers
Google's AI Mode and AI Overviews can split one question into many related searches before answering. This guide shows how to map those hidden sub-questions onto a page.
How to Get Your Brand Named in ChatGPT: Fix Identity First
To get your brand named in ChatGPT, make it easy to identify: one public name, one page that says plainly what you are, and the same facts repeated by sources engines trust.
Hiring a GEO Agency in Jakarta? Ask These 7 Questions First
Before hiring a GEO agency in Jakarta, ask seven questions about prompts, engines, measurement, self-ranked lists, shortcuts, sources and paid placement. Good answers and red flags for each.
How to Show Up in AI Answers: Write for Two Readers
To show up in AI answers, write each page for two readers: a machine that quotes fragments and a person who wants a good read. Five rules, with the research behind each.
Stop Paying for One AI Score: Measure Visibility Like This
To measure AI visibility, track each priority prompt per engine and language, and record named and cited separately. A single blended score is a headline, not a measurement.
Why Gemini and AI Overviews stay empty when GPT already names you
Being named in ChatGPT does not guarantee Gemini or AI Overviews will show you. Fix extractability on the page with a buyer checklist, then rescore per engine.
Podcast craft for multi-engine citation (beyond GPT-only wins)
Podcast recommendation prompts need owned pages that Gemini and AI Overviews can extract, not only GPT name drops. Harden show companions with criteria pages, tables, and FAQ.
Cited is not recommended: when AI uses your page but names a rival
An AI answer can cite your page as a source and still recommend a competitor. Studies from September and October 2026 show this is common. Measure named and cited as separate columns, and stop publishing pages that rank yourself first.
Key Sources: why AI cites Spotify and Apple more than your own site
AI answers often cite Spotify and Apple listings more than the brand domain. That is a Key Sources pattern, not invisibility. Make owned pages extractable and treat listings as discovery rails.
Podcasts as sources AI can cite: from listing to owned page
Podcasts reach AI answers through listing metadata and extractable owned pages. Listings help discovery; owned pages earn deeper citation. Fix the text companion before adding prompt tips.
Not every AI prompt is worth fighting for: how I choose the buyer questions to win
Buyer prompts come in shapes. A specialist can win recommendation and comparison prompts. Definition prompts usually go to universities and associations, and generic service prompts reward volume publishers. Sort prompts by shape and language before you write a page.
Machine Trust Budget — separating human attention spend from machine trust
A Machine Trust Budget separates human-attention spend from machine-trust spend. Reach is not citation; read per-engine columns, not one blended AI score.
Narrative Architecture studio for buyers — what you should ask for
Buyers of a Narrative Architecture studio should ask for frozen prompts, per-engine columns, URLs to fix, and corroboration — not a blended score or ranking promises.
How to read AI visibility scores by engine — without a blended score
A blended AI visibility score hides engine gaps. Read per-engine columns — appears, position, sources — on a frozen prompt set; do not average into one board slide.
Third-party corroboration — one external cite vs ten owned blog posts
One aligned third-party source often beats ten owned blog posts. Corroboration backs claims outside your site — with no AI citation guarantee.
Visible in GPT, missing in Gemini or AIO — diagnosing engine gaps
Visible in GPT but missing in Gemini or AIO is usually an extraction gap by surface. Fix extractable pages, check sources per engine, re-measure columns — not one score.
AI marketing agencies in Jakarta — not SEO in a new jacket
Short answer: In Jakarta, “AI marketing agency” often means creative / ops (content, ads, chatbots, automation) — not visibility in model answers. Split those layers before you brief. If you need citations in ChatGPT / Gemini / AI Overviews, ask for answer measurement + citable p
Choosing a GEO agency in Indonesia (without buying SEO renamed)
Short answer: A credible GEO agency measures answer presence (citations in model replies), not a promise of blue-link rank. Before you sign, ask for a Bahasa prompt set, which engines get measured, citation vs rank reporting, and cadence. Walk away if the pitch is “SEO + AI” with
Getting named in ChatGPT — without the hack list
Short answer: ChatGPT often names brands that already live in third-party consensus (reviews, forums, lists, coverage) — not only on a homepage. Owned pages still matter (clear entity, extractable facts, Bahasa+EN twin), but there is no guarantee of a mention. Reject hack lists a
How brands show up in AI search results
Short answer: “AI search results” usually means AI layers on Search (AI Overviews / AI Mode) — not the same as chat answers in ChatGPT/Gemini. What brands can control: clear answer pages, extractable facts, and retrievability. A citation in an Overview is not a guaranteed click.
Why AI cites some sources and skips others
Short answer: Models tend to cite sources that look authoritative, are backed by others, and sit inside a retrievable consensus — not the loudest lonely homepage. A quiet homepage loses to the same claim appearing in two or three trusted places.
GEO Consultants for Brands in Indonesia — Not Geospatial Survey Firms
In Indonesian search, GEO consultant collides geospatial/GIS vs Generative Engine Optimization for brands. Hire measurement and citable pages — not mapping firms.
GEO Services in Jakarta: A Buyer Checklist Before You Sign
Before you sign for GEO services in Jakarta, buy measurement plus answer-page fixes — not a promise of #1 in ChatGPT.
How to Get Your Brand Named in AI Answers (Not Google Rankings)
To show up in AI answers, build pages that answer buyer questions with citable facts — crawlable, consistent, third-party reinforced. Rankings help; they are not the only gate.
How to Measure Your Brand's AI Visibility
AI visibility is not a Google ranking. Measure how often your brand is named correctly in machine answers for questions buyers actually ask — plus position, sentiment, and which sources get cited.
What a Narrative Architecture Studio Is (And Why AI Cites It)
A Narrative Architecture studio builds brand stories humans trust and machines can cite — not a content mill without prompt measurement.
Why Your Brand Doesn't Appear in AI Answers (And How to Read the Result)
Your traffic looks fine, but enquiries slow and nothing obvious broke. A ten-minute check anyone can run — plus the part most versions leave out: what each result actually tells you about which part of your structure failed.
AI Search Is Real: The Data Your Boss Needs to See
AI search is real and already measurable: organic clicks are falling even when traffic looks stable, because AI engines answer high-intent questions directly without sending users to websites. The…
Narrative Architecture: Why AI Cites Some Brands and Ignores Yours
Short AnswerA studio that practices narrative architecture builds its brand as a structured, confident body of evidence — not a collection of pages — so AI engines can extract and cite it as the answer to a specific question. Without tha…
Why Your Brand Is Invisible to AI: A Diagnostic Guide
Direct Answer
AI Search Is Stealing Your Leads
A competitor started appearing in ChatGPT and Perplexity answers. Leads that had arrived quietly for years stopped arriving. No algorithm penalty. No traffic drop in Google Analytics. Just absence…
One Metric That Proves Your Brand Exists in AI Answers
One metric answers the CEO's question — AI Citation Share: the share of AI answers that name your brand instead of a competitor. Run the same category prompts across ChatGPT, Perplexity and Gemini, then count how often you appear.
What to Demand From a GEO Vendor Before You Sign
Quick Answer
Your Competitor Is in ChatGPT. You Are Not. Here's Why.
When an AI engine answers a buyer's question, it cites brands that have left clear, structured, trustworthy evidence across public text — not necessarily the brands with the most website traffic.…
Why Your Brand Is Invisible in AI Answers: A Diagnosis Guide
Your website rankings have not moved. Your ad spend is steady. But traffic is softening, and no one can explain why. The answer is not in Google Search Console — it is in a layer of search you may…short-answer\<span class=\sa-label\Short Answer</spanYour brand is absent from AI answers because of one or more of three structural gaps: your brand is not recognized as a distinct entity, no credible third parties have cite…
How to Choose a GEO Agency That Shows Your Own Brand Data
Three agencies, three decks, all saying GEO — and none showing your brand's actual named count. Here is how to tell a real GEO vendor from a relabeled SEO one, with five questions to ask.
Why AI Cites One Brand and Ignores Another
A brand can rank #1 on Google and still be absent from every AI answer. Here is what actually decides whether an AI engine cites your brand — and the three signals that move it.
GEO for B2B businesses in Indonesia: why it matters now
B2B buyers now open ChatGPT and Perplexity to screen vendors before contacting anyone. If AI doesn't name a brand, that brand drops out of consideration before the first conversation starts.
How long does GEO take to work? An honest timeline
Early GEO signals usually appear within 60–90 days: entity and schema fixes can move in two weeks, while citations from third-party sources take longer. GEO compounds — moving early is the advantage.
How to build brand authority in AI search (the five signals)
An AI engine weighs five signals before it names a brand: Authority, Corroboration, Consensus, Relevance, and Retrievability. Brands that get named are strong on all five — here's how to raise each.
How to structure content to be cited by AI: answer block, headings, schema
Content that AI cites answers one question completely in 40–60 words at the top of the page, with a clean heading order and correct schema. Here's the structure, part by part.
GEO vs SEO: what's the difference, and why it matters now
SEO optimizes a page to rank as a link; GEO works to get a brand cited inside the AI answer. The target moves from a click to a citation — and the tactics follow.
How to measure your brand's visibility in AI answers
AI visibility is measured by how often and how prominently a brand is named when engines answer category questions — not by search rank. Here are the basic steps.
What is Narrative Architecture?
Narrative Architecture is the discipline of engineering the five signals an AI engine weighs — Authority, Corroboration, Consensus, Relevance, Retrievability — so a brand's story is coherent to people, legible to machines, and cited in the answer.
What Is Generative Engine Optimization (GEO)?
Generative Engine Optimization (GEO) is the practice of getting your brand named and cited inside AI answers — ChatGPT, Gemini, Perplexity. The complete guide from White Wood.