Article illustration 1

New Real User Monitoring (RUM) datasets reveal a startling pattern: websites built as Single-Page Applications (SPAs) average just one soft navigation per hard page load. This metric, observed independently in the RUM Archive and Chrome's Soft Navigations Origin Trial data, suggests users engage with SPAs in exceptionally shallow sessions. For Microsoft Edge architect Alex Russell, this data points to "an industry-wide failure of technology selection" where teams incur significant complexity for minimal payoff.

The Broken SPA Value Proposition

SPAs traditionally promised a performance trade-off: accept higher initial loading latency (due to large JavaScript bundles) in exchange for faster subsequent interactions through client-side rendering. Russell mathematically models this as:

L_avg = [latency(navigation) + Σ latency(i)] / N

Where N represents total user interactions per session. With observed N≈1–2, the equation collapses—the upfront JavaScript penalty isn't amortized across sufficient interactions. "If sessions are truly this shallow," Russell notes, "there’s almost no space for SPA technologies in the market" outside specialized applications like GIS or document editors.

The Perfect Storm Driving SPA Adoption

Several factors propelled SPAs to default status despite performance concerns:
1. Browser capability gaps: iOS restrictions blocked progressive enhancement approaches
2. Missing native components: Lack of styleable date pickers, scrollable lists, and gesture support necessitated JavaScript workarounds
3. Toolchain momentum: React-based frameworks became onboarding defaults despite bundle-size risks

This convergence created what Russell terms "frontend’s lost decade," where JavaScript-heavy architectures ignored network/device constraints. The result? Teams now require specialized remediation experts to combat chronic performance issues exacerbated by these stacks.

Three Critical Mysteries

  1. Distribution Mystery: Does N≈2 represent all SPAs equally, or do some application types buck the trend?
  2. Definition Mystery: Are interactions like infinite scroll or non-URL updates underrepresented in current metrics?
  3. Intervention Mystery: Can session-depth insights guide teams toward more suitable architectures before they adopt problematic stacks?

Implications for Web Development

The data suggests most teams adopting SPAs are making an architectural misstep. As Russell observes, "Client-side JavaScript has always been the slowest and most expensive way to accomplish anything in a browser." With most global users on low-end Android devices, the cost of misplaced complexity becomes particularly acute.

This revelation positions web performance practitioners as crucial advisors for architectural decisions. By quantifying real-world interaction patterns, they can steer teams away from FOMO-driven choices toward solutions matching actual user behavior. As Russell concludes: "If there are high callings for the Web Performance community, steering teams away from needing our help... would seem near the top of the stack."

Source: The Curious Case of the Shallow Session: SPAs by Alex Russell