fix: adoption score consistency uses period days, realistic mock model costs

Consistency metric now divides by the actual time period instead of
max active days across users — gives accurate adoption tiers.
Mock data updated with realistic per-request costs and model selection
that respects cost tiers.

Co-authored-by: Cursor <cursoragent@cursor.com>

Author: ofershap

FEATURES.md

@@ -29,14 +29,15 @@
 ### Members Table
 
 - Sortable by: spend, activity, requests, lines, $/req, context, name
-- Filterable by badge type
+- Filterable by badge type via `Badges ▾` dropdown with collapsible sections and per-badge user counts
+- Active badge filter shown as chip with ✕ dismiss next to table header
 - Columns: rank, name, email, spend, requests, lines, $/req, model, profile badges, ranks
 
-### Badges (per user, max 2)
+### Badges (per user, all shown)
 
 | Category | Badges                                                                                                                                                                |
 | -------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
-| Usage    | Power User, Deep Thinker, Low Usage                                                                                                                                   |
+| Usage    | Power User, Deep Thinker, Balanced, Low Usage                                                                                                                         |
 | Spend    | Cost Efficient, Premium Model, Over Budget                                                                                                                            |
 | Context  | Long Sessions, Short Sessions                                                                                                                                         |
 | Adoption | AI-Native (80%+), High Adoption (55%+), Moderate (30%+), Low Adoption (10%+), Manual Coder (<10%) — based on composite score (accept rate + engagement + consistency) |

data/mock.db

@@ -75,7 +75,7 @@ const STORY_MEMBERS: StoryMember[] = [
     userId: "usr_marcus001",
     storyRole: "expensive_model",
     activityLevel: "high",
-    primaryModel: "claude-4.5-opus-high",
+    primaryModel: "claude-4.6-opus-high",
     clientVersion: "2.5.12",
     group: "Backend > API",
   },
@@ -199,24 +199,27 @@ const MODELS = [
   "gpt-5.3-codex",
   "claude-4.5-sonnet-thinking",
   "claude-4.6-opus-high-thinking-fast",
+  "claude-4.6-opus-max-thinking",
 ];
 
-const MODEL_WEIGHTS = [20, 15, 8, 12, 10, 10, 5, 5, 5, 5, 3, 2, 0];
+const MODEL_WEIGHTS = [20, 15, 4, 12, 10, 10, 5, 5, 5, 5, 3, 2, 0, 4];
 
 const MODEL_COST_PER_REQ: Record<string, number> = {
-  "claude-4.5-sonnet": 0.18,
-  "claude-4.6-opus-high": 1.03,
-  "claude-4.6-opus-max": 1.26,
-  "claude-4.5-haiku": 0.05,
-  "claude-4.6-opus-high-thinking": 2.8,
-  "gpt-5.2": 0.45,
-  "claude-4.5-opus-high-thinking": 0.42,
-  "gpt-5.2-codex": 0.65,
-  "gemini-3-pro-preview": 0.35,
-  "gemini-3-flash-preview": 0.12,
-  "gpt-5.3-codex": 0.85,
-  "claude-4.5-sonnet-thinking": 0.55,
+  "claude-4.5-sonnet": 0.06,
+  "claude-4.6-opus-high": 0.38,
+  "claude-4.6-opus-max": 4.5,
+  "claude-4.5-haiku": 0.015,
+  "claude-4.6-opus-high-thinking": 0.8,
+  "gpt-5.2": 0.18,
+  "claude-4.5-opus-high": 0.35,
+  "claude-4.5-opus-high-thinking": 0.3,
+  "gpt-5.2-codex": 0.22,
+  "gemini-3-pro-preview": 0.12,
+  "gemini-3-flash-preview": 0.03,
+  "gpt-5.3-codex": 0.32,
+  "claude-4.5-sonnet-thinking": 0.18,
   "claude-4.6-opus-high-thinking-fast": 14.55,
+  "claude-4.6-opus-max-thinking": 7.8,
 };
 
 const GROUPS = [
@@ -1414,18 +1417,19 @@ function generateNormalDay(user: StoryMember, date: string, isWeekend: boolean):
   const agentReqs = baseReqs;
   const chatReqs = rand(0, Math.floor(agentReqs * 0.2));
   const composerReqs = rand(0, Math.floor(agentReqs * 0.1));
-  const model = seededRandom() < 0.8 ? user.primaryModel : weightedPick(MODELS, MODEL_WEIGHTS);
+  let model = user.primaryModel;
+  if (seededRandom() > 0.85) {
+    const primaryCost = MODEL_COST_PER_REQ[user.primaryModel] ?? 0.5;
+    const candidates = MODELS.filter((m) => (MODEL_COST_PER_REQ[m] ?? 0.5) <= primaryCost * 3);
+    const candidateWeights = candidates.map((m) => MODEL_WEIGHTS[MODELS.indexOf(m)] ?? 1);
+    model = weightedPick(candidates, candidateWeights);
+  }
   const costPerReq = MODEL_COST_PER_REQ[model] ?? 0.5;
 
-  const events = generateEvents(
-    user,
-    date,
-    Math.min(agentReqs, rand(5, 25)),
-    model,
-    costPerReq,
-    false,
-  );
-  const spendCents = Math.round(events.reduce((s, e) => s + e.totalCents, 0));
+  const eventCount = Math.min(agentReqs, rand(5, 25));
+  const events = generateEvents(user, date, eventCount, model, costPerReq, false);
+  const jitter = 0.85 + seededRandom() * 0.3;
+  const spendCents = Math.round(costPerReq * 100 * agentReqs * jitter);
 
   const linesAdded = agentReqs * rand(3, 15);
   const linesDeleted = Math.floor(linesAdded * (0.2 + seededRandom() * 0.3));

scripts/generate-mock-db.ts

@@ -1218,7 +1218,7 @@ export function getFullDashboard(days: number = 7): FullDashboard {
     cycleEnd,
     cycleDays,
     dailyTeamActivity,
-    rankedUsers: assignBadges(rankedUsers).map((u, i) => ({ ...u, rank: i + 1 })),
+    rankedUsers: assignBadges(rankedUsers, days).map((u, i) => ({ ...u, rank: i + 1 })),
   };
 
   return { days, stats, modelCosts, teamDailySpend, dailySpendBreakdown };
@@ -1247,6 +1247,7 @@ function assignBadges(
     spend_rank: number;
     activity_rank: number;
   }>,
+  periodDays: number = 7,
 ): Array<
   (typeof users)[number] & {
     usage_badge: UsageBadge | null;
@@ -1255,7 +1256,6 @@ function assignBadges(
     adoption_badge: AdoptionBadge | null;
   }
 > {
-  const maxActiveDays = Math.max(...users.map((u) => u.active_days), 1);
   const activeUsersForIntensity = users.filter((u) => u.agent_requests >= 10 && u.active_days > 0);
   const intensities = activeUsersForIntensity
     .map((u) => u.agent_requests / u.active_days)
@@ -1331,7 +1331,7 @@ function assignBadges(
       const acceptRate = u.total_applies > 0 ? u.total_accepts / u.total_applies : 0;
       const intensity = u.agent_requests / u.active_days;
       const intensityNorm = Math.min(intensity / p90Intensity, 1);
-      const consistency = u.active_days / maxActiveDays;
+      const consistency = periodDays > 0 ? u.active_days / periodDays : 0;
       const score = acceptRate * 40 + intensityNorm * 40 + consistency * 20;
       if (score >= 80) adoption_badge = "ai-native";
       else if (score >= 55) adoption_badge = "high-adoption";