Path-Dependent UX: How User Journey History Should Shape What They See

Most web systems are Markovian. They react only to the current state — what page you’re on right now. With no memory of how you got there. A visitor who arrived via a Google search for “personalized Christmas ornaments for grandma” sees the exact same category page as someone who navigated from the homepage through three levels of taxonomy.

This is a fundamental architectural limitation, and it’s costing every ecommerce business money they don’t know they’re losing.

The Markov Baseline

Classical Markov models assume that the next state depends only on the current state, not the full history. In web terms, this means a page reacts only to current context. There’s no memory of the user’s path. No inference about their evolving intent. No adaptation based on the narrative they’ve been building through their browsing behavior.

This works for basic recommendation systems and simple adaptive navigation. It’s computationally cheap and easy to implement. But it’s fundamentally insufficient for meaning-making, authority building, and trust.

Why? Because human meaning-making is path dependent.

Post-Markov Systems

In the Semantic Projection Commerce architecture, we treat user context as a cumulative signal. Where a user came from, how they arrived, and why they might be here all shape how knowledge should be projected to them.

This aligns with several established theoretical frameworks. Higher-order Markov models capture multi-step transition dependencies. Bayesian belief updating provides a formal mechanism for refining intent estimates as new signals arrive. Narrative theory and hermeneutics explain how humans construct meaning through sequences of experiences, not isolated moments.

The practical implication is that pages should not have a single fixed presentation. They should have a canonical semantic core. The knowledge that’s true regardless of context, and multiple contextual projections that adapt based on the user’s journey.

What This Looks Like in Practice

Consider a product category page for personalized ornaments. Under a Markov system, everyone sees the same grid of products with the same sorting and the same descriptions.

Under a path-dependent system, the projection layer considers the user’s accumulated context. A visitor who searched for “first Christmas together ornament” and then browsed the “couples” subcategory sees the category with romance-themed sorting, relationship milestone messaging, and production method details that emphasize personalization quality. A visitor who came from a blog post about gift-giving etiquette sees the same products framed around gifting confidence. Reviews, shipping guarantees, and personalization previews featured prominently.

The underlying knowledge is identical. The canonical topic is the same. The products don’t change. But the projection. The way knowledge is rendered for this specific user in this specific moment. Adapts to serve their actual intent.

The Authority Anchor

Here’s where path dependence intersects with SEO strategy. Search engines need stable referents. If your pages are completely dynamic. Different content for every visitor. There’s nothing for a crawler to index reliably.

SPC resolves this tension by separating the canonical layer from the projection layer. The canonical version of every page is stable, versionable, and indexable. It represents the authoritative knowledge about that topic. Dynamic projections happen on top of the canonical layer for logged-in users or users with accumulated session context, but the base that search engines and AI systems reference remains consistent.

This is the key architectural insight: dynamic user experiences must be anchored to stable semantic artifacts. You can personalize the presentation without destabilizing the authority.

Implementation Realities

Full path-dependent projection is the long-term architecture. On a Shopify store with current constraints, the immediate implementation is more pragmatic. Using UTM parameters, referral sources, and session-level intent signals to adjust above-the-fold messaging and sorting logic on key category pages.

Even this simplified version produces measurable improvements in engagement metrics because it aligns what the user sees with why they’re actually there, rather than showing everyone the same default experience.

In Part 3, we’ll break down the category spine architecture and how canonical topics prevent the semantic cannibalization that kills most ecommerce sites.