Load Test Result Comparator Across Runs

WHAT IT DOES

Load Test Result Comparator Across Runs takes load test result data from two separate test runs and performs a side-by-side comparison of all reported metrics. The tool accepts JSON input containing run labels and metric objects, then automatically identifies all metrics present in either run and computes the difference between them. For numeric metrics, it calculates the absolute change, percentage change, and applies a heuristic to determine whether the change represents an improvement or regression based on metric type (e.g., lower gas and lower latency are improvements). The output includes a summary cards view showing each run's label and metric count, a detailed comparison table with per-metric changes color-coded green (improvement), red (regression), or neutral, and an overall summary indicating whether the tested changes improved, degraded, or had mixed results.

HOW TO USE

Provide load test result data for two runs in JSON format. Each run must include a "label" field (e.g., "Baseline" or "Optimized") and a "metrics" object containing key-value pairs of metric names and values. Metric values can be numbers, strings, or arrays. Paste the JSON for Run 1 into the first text area and Run 2 into the second. Click "Compare runs" to generate a detailed comparison. The tool displays a summary card for each run, a table showing every metric with values from both runs and the computed change, and a final summary statement indicating the overall trend. This enables quick identification of performance regressions or improvements when comparing changes in contract code, configuration, or infrastructure.

TECHNICAL MECHANISM: METRIC SET UNION AND DIFFERENTIAL ANALYSIS

The tool collects all metric keys from both runs using a Set union, ensuring no metric is missed even if it appears in only one run. For each metric, it compares the values from run 1 and run 2. Numeric values undergo difference calculation (run2 - run1) and percentage change (diff / |run1| × 100). The tool applies a heuristic classification: metrics containing keywords like "gas", "failures", "duration", "latency", "time", or "cost" are treated as "lower is better" metrics, while all others are treated as "higher is better". This allows the tool to automatically color-code changes as improvements (green) or regressions (red). String and object values are compared for equality, with "same" or "changed" status. The summary aggregates total changes, improvements, and regressions to provide an overall trend assessment. All comparisons are deterministic and client-side, with no data leaving the browser.

WHAT IT CANNOT SEE

Cannot execute or run load tests itself; it only analyzes pre-collected data. Cannot automatically determine which metrics are statistically significant; small changes may be noise. Cannot detect external factors that may have influenced test results, such as network congestion, RPC latency, or node performance differences. Cannot guarantee that test conditions were identical across runs—differences in environment, time of day, or test parameters may skew results. Cannot identify performance regressions that are not reflected in the provided metrics. Cannot validate the accuracy or reliability of the load test data; it assumes the user's data is trustworthy. Cannot provide recommendations for optimization without domain-specific context. Cannot compare results that use different measurement methodologies. Cannot detect anomalies that require historical baseline data beyond what is provided.

PLEASE NOTE

EVM chains only. This tool does not interact with any blockchain; it reads no on-chain data and performs no transactions. All processing is local and private.