Back to Research Reports
Open Science & Dataset Transparency

Testing Methodology & Datasets

We publish verifiable web performance and SEO benchmarks. This registry documents our active datasets, testing hardware profiles, and statistical treatment methods.

Active Research Cohorts & Telemetry

Dataset A: 100-Agency Website Forensic Audit

100 digital agency homepages
Observation Window:Q1 2026 (January 15 - February 20, 2026)
Primary Objective:Forensic evaluation of agency homepage technical hygiene, mobile video payloads, and Core Web Vitals.
Selection & Sampling Protocol:

Randomized sample drawn from Clutch and Awwwards agency directories across the United States and Western Europe.

Dataset B: 500-Site WordPress Performance Benchmark

500 active WordPress domains
Observation Window:Q1-Q2 2026 (February 1 - March 10, 2026)
Primary Objective:Comparative benchmarking of block themes vs legacy page builders, edge HTML caching vs standard hosting, and script execution weight.
Selection & Sampling Protocol:

Stratified sample across e-commerce (WooCommerce), editorial publications, and business websites with measurable organic traffic.

Telemetry Base: 10,000+ Cumulative Diagnostic Scans

10,000+ cumulative URL inspections
Observation Window:Continuous (November 2025 - Present)
Primary Objective:Validating DOM parsing speed, LCP highlighter accuracy, and client-side inspection stability across diverse layouts.
Selection & Sampling Protocol:

Synthetic regression suites and active tab scans run across VitalsSniper PRO engine dogfooding and automated test runs.

Standard Testing Rig & Network Profiles

To ensure statistical reproducibility, our synthetic lab audits execute within containerized Chromium headless environments configured with standard W3C mobile network profiles:

Mobile Device Emulation

  • • Viewport: 390 x 844 pixels (DPR: 3.0)
  • • CPU Throttling: 4x slowdown (simulating mid-tier mobile hardware)
  • • User-Agent: Chrome Mobile Android (Linux; Android 14)

Network Emulation

  • • Simulated 4G Profile: 4.0 Mbps download, 3.0 Mbps upload
  • • Latency: 40ms RTT round-trip delay
  • • Runs per URL: 3 runs (1 cold cache, 2 warm cache averaged)

Correlation vs. Causation Standard

Our editorial integrity standard strictly differentiates between observed empirical correlations and demonstrated causal relationships. When our studies report that websites utilizing lightweight themes load faster, this reflects observed data across our measured cohort rather than an assertion that a single theme inherently solves all architectural bottlenecks.

Similarly, our research into AI crawlers (GPTBot, ClaudeBot, PerplexityBot) measures access frequency and text extraction success in live server logs; it does not claim to decode proprietary ranking algorithms.