Motoko Hunt, president of international search marketing at AJPR, published a piece on Search Engine Journal on 3 November 2026 about building a global content factory with AI. Her argument is that the old global content model, built on translation memory and one unified message, is failing in AI search. When every market’s page looks the same, the strongest version wins, usually the US one, and a buyer in France or Japan gets an answer that does not apply to them. Her fix is a governed fact registry for each market, a split between facts that never change and facts that vary by market, AI doing the drafting and translation, and local people supplying the evidence, the questions and the approvals.
The claim, and what actually changed
The headline promise is that you can run content as a factory, across markets, and still keep the human touch. The excitement is in the word factory. Volume across many markets at a cost that used to be out of reach for anyone without an agency in each country.
What changed is that AI made the assembly cheap. Drafting, translating and structuring a page is now close to free.
What did not change is what an AI engine needs before it will quote you. It needs a fact that is true for that market, dated, and specific enough to repeat. Hunt’s own composite example shows the failure: French pages that never mention France, Spanish pages that pull answers meant for Latin America, thin Japanese pages built on old data. Translation never fixed any of that, because the pages had nothing market-specific in them to translate.
Where the human touch actually lives
Read “without losing the human touch” and most teams picture a reviewer at the end of the line, reading AI drafts and fixing the tone. That is the model that produces slop at scale. The reviewer can catch a wrong price. They cannot put in a point of view that was never there.
Hunt’s method, to her credit, puts people at the front. Local teams supply the evidence, write the questions their market actually asks, and approve which claims are allowed. I would go further. That front-end work is the whole human contribution. Everything after it is assembly.
Slop was never in the tool. It comes from expecting the AI to think on its own and mass-produce pages with no direction and no opinion. The output comes out generic, and the engines are now good enough to notice and quote it less. Volume without a frame is the problem, and a frame is something you write before the first draft.
What this looks like for a skincare brand selling in three markets
Take a skincare brand with customers in Singapore, Malaysia and Australia, and a marketing team of three. The old way was one English page, lightly adapted, with the same claims everywhere.
The factory way, done properly, starts with a list. Which products are registered for sale in each market. The price in each currency. Delivery times from each warehouse. Which ingredient claims each regulator allows. One or two pieces of local proof per market, with a date. The team writes that down, approves it, and dates it. Then the AI drafts each market’s page from that list and nothing else.
The human work in that setup is deciding what is true and what the brand’s stance is on it. A team of three can do that for three markets. They could never have written and maintained all those pages by hand.
How I run my own content
I run content at volume, and I use AI for almost all of it. Every piece carries a frame, an opinion and facts I have approved before the drafting starts. Claude does the building, the counting and the checking. I make the decisions.
The results are dated and scoped, because AI answers change week to week. Within three weeks of putting up comparison pages in early 2026, AeroChat was in the same Google AI Overview as Tidio, Gorgias and Intercom. The work before those pages was writing down exactly what AeroChat is, who it is for and how it is different, with no marketing language. My own coffee store, Qwang Coffee, got its first AI Overview citation 48 hours after launch in August 2026, then lost every Google citation within a week while ChatGPT still named it first. Stridec, my agency, was named first in Google’s AI Overview for “best AI SEO agency” in September 2026.
No reviewer at the end of the line produced any of that. The brief did.
What a business should do
- Write the fact registry before any page. One list per market: what is sold, at what price, delivered how fast, with what proof, approved by whom, dated.
- Split the list the way Hunt describes. Facts that never change go in once. Facts that vary by market are sourced from that market.
- Put an opinion in the brief. What you believe about your category, and what you would tell a customer to their face. The AI cannot supply this.
- Date every figure on every page, and keep each claim and its qualifier in the same passage.
- Measure in each market’s language. Ask the engines the questions your buyers ask and record whether you are named and whether the facts are right.
- If the team cannot write the brief, do not run the factory. Fix the brief first, then turn on the volume.
Source: Search Engine Journal: Building A Global Content Factory With AI (Without Losing The Human Touch)