diff --git a/src/data_analysis/optimal_p_estimation.rs b/src/data_analysis/optimal_p_estimation.rs
index e4edec9..aa28c41 100644
--- a/src/data_analysis/optimal_p_estimation.rs
+++ b/src/data_analysis/optimal_p_estimation.rs
@@ -503,27 +503,37 @@ impl OptimalPAnalysis {
 
         // Compact header - axis name and basic info
         output.push_str(&format!(
-            "{}: Td={:.1}ms (target {}, {:+.0}% dev, windows={}), Noise={}, Consistency={:.0}%\n",
+            "{}: Td={:.1}ms (target {}, {:+.0}% dev, windows={}), Noise={}\n",
             axis_name,
             self.td_stats.mean_ms,
             target_display,
             self.td_deviation_percent,
             self.td_stats.num_samples,
             self.noise_level.name(),
-            self.td_stats.consistency * 100.0
         ));
 
-        // Warning for low consistency (inline)
-        if !self.td_stats.is_consistent() {
-            let cv_percent = self
-                .td_stats
-                .coefficient_of_variation
-                .map_or(0.0, |cv| cv * 100.0);
-            output.push_str(&format!(
-                "  ⚠ Low consistency (CV={:.1}%) — unreliable (>{:.0}%)\n",
-                cv_percent,
-                TD_COEFFICIENT_OF_VARIATION_MAX * 100.0
-            ));
+        // Reliability line — always shown with both metrics
+        {
+            let cv_str = self.td_stats.coefficient_of_variation.map_or_else(
+                || "CV=N/A".to_string(),
+                |cv| {
+                    format!(
+                        "CV={:.1}% (≤{:.0}%)",
+                        cv * 100.0,
+                        TD_COEFFICIENT_OF_VARIATION_MAX * 100.0,
+                    )
+                },
+            );
+            let cons_str = format!(
+                "Consistency={:.0}% (≥{:.0}%)",
+                self.td_stats.consistency * 100.0,
+                TD_CONSISTENCY_MIN_THRESHOLD * 100.0,
+            );
+            if self.td_stats.is_consistent() {
+                output.push_str(&format!("  Reliable: {cons_str}, {cv_str}\n"));
+            } else {
+                output.push_str(&format!("  Unreliable: {cons_str}, {cv_str}\n"));
+            }
         }
 
         // Compact recommendation
diff --git a/src/plot_functions/plot_step_response.rs b/src/plot_functions/plot_step_response.rs
index f73cef2..b71183d 100644
--- a/src/plot_functions/plot_step_response.rs
+++ b/src/plot_functions/plot_step_response.rs
@@ -12,6 +12,7 @@ use crate::constants::{
     FINAL_NORMALIZED_STEADY_STATE_TOLERANCE, LINE_WIDTH_PLOT,
     LOW_AUTHORITY_SETPOINT_THRESHOLD_DEG_S, POST_AVERAGING_SMOOTHING_WINDOW, RESPONSE_LENGTH_S,
     STEADY_STATE_END_S, STEADY_STATE_START_S, TD_COEFFICIENT_OF_VARIATION_MAX,
+    TD_CONSISTENCY_MIN_THRESHOLD,
 };
 use crate::data_analysis::calc_step_response; // For average_responses and moving_average_smooth_f64
 use crate::data_analysis::optimal_p_estimation::{OptimalPAnalysis, PRecommendation};
@@ -629,28 +630,32 @@ pub fn plot_step_response(
                             stroke_width: 0,
                         });
 
-                        // Consistency — always shown; orange warning when poor
+                        // Reliability — always shown with both metrics; orange when poor
                         {
-                            let cv_percent = analysis
-                                .td_stats
-                                .coefficient_of_variation
-                                .map_or(0.0, |cv| cv * 100.0);
-                            let consistency_pct =
-                                (analysis.td_stats.consistency * 100.0).round() as u32;
-                            let (cons_label, cons_color) = if !analysis.td_stats.is_consistent() {
-                                (
+                            let cv_str = analysis.td_stats.coefficient_of_variation.map_or_else(
+                                || "CV=N/A".to_string(),
+                                |cv| {
                                     format!(
-                                        "  Consistency: {}% (CV={:.1}%) — unreliable (>{:.0}%)",
-                                        consistency_pct,
-                                        cv_percent,
-                                        TD_COEFFICIENT_OF_VARIATION_MAX * 100.0
-                                    ),
-                                    COLOR_OPTIMAL_P_WARNING,
+                                        "CV={:.1}% (≤{:.0}%)",
+                                        cv * 100.0,
+                                        TD_COEFFICIENT_OF_VARIATION_MAX * 100.0,
+                                    )
+                                },
+                            );
+                            let cons_str = format!(
+                                "Consistency={:.0}% (≥{:.0}%)",
+                                analysis.td_stats.consistency * 100.0,
+                                TD_CONSISTENCY_MIN_THRESHOLD * 100.0,
+                            );
+                            let (cons_label, cons_color) = if analysis.td_stats.is_consistent() {
+                                (
+                                    format!("  Reliable: {cons_str}, {cv_str}"),
+                                    COLOR_OPTIMAL_P_TEXT,
                                 )
                             } else {
                                 (
-                                    format!("  Consistency: {}%", consistency_pct),
-                                    COLOR_OPTIMAL_P_TEXT,
+                                    format!("  Unreliable: {cons_str}, {cv_str}"),
+                                    COLOR_OPTIMAL_P_WARNING,
                                 )
                             };
                             series.push(PlotSeries {