AI That Builds the Page for Each Visitor: What Generative Personalization Means for Your Website

Netflix has always personalized what you see, but its engineers recently described something a step further: GenPage, an end-to-end generative model that builds a whole personalized homepage for each user. Not "returning visitors see this banner," but an AI assembling the experience for the individual. Netflix operates at a scale most businesses never will, yet the shift it signals, from picking pre-built variants to AI generating the experience, is one every business with a website should understand. Because the underlying idea is about to become accessible far beyond Netflix.

From picking variants to generating experiences

Personalization isn't new, but it's been fundamentally limited. The traditional approach is rules-based: you define segments ("first-time visitor," "from a paid ad," "returning customer"), hand-build a variant for each, and the system shows the closest match. It works, but it caps out at however many versions you can create and maintain. Generative personalization removes that cap: AI assembles or adapts the content on the fly for the individual, enabling genuinely one-to-one experiences without someone pre-building every combination. GenPage is that idea taken to its logical end, the page itself is generated per user.

You don't need Netflix's stack to benefit

The mistake would be to file this under "big-tech only." The principle, use AI to make each visitor's experience more relevant, scales down through tools most businesses can already reach. Practical entry points: AI-personalized email content, dynamic landing pages that match the ad or source a visitor came from, tailored product or content recommendations, and messaging adapted to visitor behaviour on your key pages. You're not building a generative model; you're applying AI to make your existing site and campaigns speak more directly to each person, which is exactly the kind of high-value marketing work we cover in using AI for marketing.

It also pairs with the broader lesson that AI is only as good as the context you feed it, personalization runs on your customer and behavioural data, so the businesses with clean, well-organized data get better results, echoing why generic AI gives generic answers.

The guardrails that keep it from backfiring

More power means more ways to go wrong, so personalization needs limits. Brand and accuracy: AI-assembled content can drift off-message or state something incorrect, keep guardrails and review, especially on anything customer-facing. Privacy: personalization runs on visitor data, which must comply with PIPEDA and rest on clear consent. The "creepy" line: helpful relevance builds trust; personalization that reveals you're tracking too much breaks it. The goal is to feel considerate, not surveillant, relevant enough to help, transparent enough to trust.

How to start

Begin where relevance pays off and your data is clean. Personalize by traffic source (a visitor from a specific ad lands on messaging matched to that ad), by segment, or by behaviour on your highest-value pages. Use AI to generate and test variations, measure the conversion lift against a baseline, and scale what works. Keep it compliant and on-brand. You don't need to generate every page from scratch to win, you need to make your most important moments meaningfully more relevant, one improvement at a time.

Where this is heading

GenPage is a preview of a web where experiences are increasingly generated for the individual rather than served from a shelf of fixed pages. Most businesses won't (and shouldn't) rebuild their site around a generative model tomorrow. But the direction is clear, and the accessible pieces are here now. Start using AI to make each visitor's experience more relevant, with the right guardrails, and you'll be building the muscle for a more personalized web while your competitors are still serving everyone the same page.

Frequently Asked Questions