> ## Content Index
> Fetch the complete content index at: https://www.agenticlandmark.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# "Optimize for People," Says the Machine
- URL: https://www.agenticlandmark.com/brief/optimize-for-people-says-the-machine/
- Published: 2026-09-19T19:53:51.000Z
- Updated: 2026-09-19T19:53:51.000Z
- Description: Google's holiday guidance tells retailers to optimize for people, not bots. Its other recommendations are all written for machines: richer product feeds, real-time price and inventory, agentic checkout. The contradiction resolves once you separate the surface layer from the infrastructure layer.
- Author: Jim Cook
- Tags: agentic-commerce, google, two-layer-architecture, ARI, retail

*Google's holiday guidance tells retailers to stop building for bots, then hands them a to-do list only a bot could read.*

## The advice and the playbook disagree

In August, Think with Google published its read on how AI is reshaping holiday shopping, built on Ipsos surveys Google commissioned. The piece breaks the season into four shifts: exploring, evaluating, validating, and deal-hunting. Each ends with a recommended move. The one that will get quoted sits in the middle. Retailers are told to optimize for people, not bots, on the grounds that Google's systems read language the way humans do. Build fast, clean, mobile pages, the argument goes, and the AI will recommend you on its own.

Read the other three moves and the advice inverts. Enrich product data for the Shopping Graph so camera searches and AI summaries can surface you. Keep prices and inventory current in real time so AI assistants don't pass you over. And the companion playbook the piece links to closes on structuring product feeds for AI search and building early agentic checkout infrastructure.

That is not a list for people. People do not read your product feed. They do not query your inventory endpoint. They never see the structured attributes that decide whether an AI summary includes you. Every one of those moves is written for a machine.

## Both things are true, on different layers

The contradiction resolves once you separate the two layers the article collapses into one. The surface layer is where agents and people meet your brand: the page, the summary, the video, the review. The infrastructure layer is what agents query to decide whether you belong on that surface at all: the feed, the attributes, the price, the stock status, the checkout path.

Google's line is sound advice for the surface. Don't stuff keywords. Don't write for a ranking algorithm the way search marketing did a decade ago. But the surface is downstream. A beautifully written product page that sits behind stale inventory data gets skipped, and Google says so in the same article.

There is a second reason the humans-first framing doesn't carry. Google's crawler renders JavaScript. Most AI crawlers don't; they take the initial HTML and move on. Botify and Lumar found that 63% of product detail pages on JavaScript-heavy e-commerce sites return an empty or near-empty HTML shell. A page can be fast and polished for a person and a skeleton for every agent outside Google. "Build for humans and the AI will follow" holds only for the AI that renders like a browser.

## The journey got longer, and moved off your site

The strongest number in the piece is about scope. Shoppers who used AI somewhere in a considered purchase noticed or interacted with 12.1 touchpoints on average, against 4.2 for those who didn't (Ipsos for Google, global average, December 2025).

The article reads this as good news: more exploration, more chances to be found. Look at where those touchpoints live. AI summaries, camera search, YouTube creators, Google Pay. The journey nearly tripled in length, and almost none of the added steps happen on a page the brand owns or measures.

That has two consequences. The first is Traffic Dependency. A brand whose revenue model assumes shoppers arrive at its site to compare is exposed when the comparison happens inside a summary. The second is quieter. A brand that can't see those eight extra touchpoints in its analytics can't tell whether it's winning or losing in them. That is the Measurement Readiness gap, and it widens as the journey lengthens.

## The verification loop is a position, not just a behavior

The piece's most striking claim is that 99% of people who research products in ChatGPT also check Google Search before deciding. Read the sample: Google-commissioned, ChatGPT users who reported commercial use, April 2026\. It describes a real behavior. It is also Google stating where it sits in the journey: the place you go to check the other AI's work.

For brands, this matters less as a Google-versus-OpenAI story than as a reminder about Brand Moat. The validation step exists because shoppers don't fully trust a single recommendation. Being present where they check is necessary. It is not the same as being chosen. A brand with no distinct reason to be preferred survives verification and still loses the sale to the equivalent option that was cheaper, in stock, and one tap from checkout.

## Bypass is the word to keep

The fourth shift is where the article comes closest to naming the real change. Shoppers use AI to hunt coupons, check inventory, try products on virtually, and get through checkout faster. Sixty percent of Google AI users said AI helps them settle the decision, find a better deal, and finish the purchase faster (Ipsos for Google, April 2026).

In the practice's terms, that is a Scoper handing parameters to an assistant (under $150, arrives by the 20th, best price) and moving toward Delegator, where the assistant closes the purchase inside those limits. At that point the agent isn't recommending your product to a person who then visits your site. It is comparing structured offers and acting on the best one. Google's own warning, that stale prices and stock let AI assistants pass you over, is Agent Substitutability described by the platform doing the substituting.

No page experience rescues you from that. The assistant doesn't see the page. It sees your price, your availability, and whether it can complete checkout.

## What to build before peak

The useful reading of Google's piece is the three moves it doesn't lead with. Before peak season, a retailer should be able to answer four questions plainly. Are the attributes a shopper would ask about (materials, sizing, compatibility, use case) present as structured fields in the feed, and not only in marketing copy? How much time passes between a price or stock change in your systems and that change reaching every surface an agent reads? Does your product content exist in the initial HTML, or only after JavaScript runs? Can an agent complete a purchase without handing the shopper back to a browser session?

Then a fifth, which the article doesn't ask: when those eight extra touchpoints decide a sale, will you be able to see it?

These are infrastructure questions. They are the ones that decide whether the surface ever gets a chance to impress anyone.

## The line worth quoting back

Google is right that nobody should write product pages for a ranking algorithm. But "optimize for people, not bots" is the surface half of a two-layer problem, and the rest of Google's own advice is the other half. The shopper at the end of this holiday season is still a person. The thing deciding whether your product reaches that person, more often each quarter, is not.

[4 AI holiday shopping shifts brands should know - Think with GoogleGoogle’s Official Digital Marketing Publication. Discover four consumer behavior shifts reshaping holiday shopping and how brands can respond.![](https://storage.ghost.io/c/e1/c4/e1c4b063-1fa4-4977-8ffa-6add50cd2316/content/images/icon/google-favicon-48-ba4e1210-631c-480e-9a9e-14fb8dcd8858.png)Google BusinessSaba Afreen![](https://storage.ghost.io/c/e1/c4/e1c4b063-1fa4-4977-8ffa-6add50cd2316/content/images/thumbnail/twg-holiday-insights-1-ai-powered-shopping-season-1600x900-final-b5030503-f479-4e3b-abe2-1be07ff24da4.webp)](https://business.google.com/us/think/consumer-insights/ai-holiday-shopping-shifts/?ref=agenticlandmark.com)