diff --git a/.changeset/dashboard-report-drill.md b/.changeset/dashboard-report-drill.md
new file mode 100644
index 0000000000..dd04a2faad
--- /dev/null
+++ b/.changeset/dashboard-report-drill.md
@@ -0,0 +1,5 @@
+---
+"@objectstack/service-analytics": minor
+---
+
+Dataset analytics enrich **dimension** result fields with their display label (so report/dashboard table headers read "Status" instead of the raw field name) and expose drill-through metadata on the dataset query result: the base `object`, a drillable dimension→field map, and a parallel `drillRawRows` array of each row's raw grouped values (captured before label resolution). This lets a host drill a grouped bucket back to its underlying records with an exact-match filter built from the stored value, not the display label. Date dimensions are excluded (a humanized bucket can't be exact-matched).
diff --git a/examples/app-showcase/src/datasets/chart-gallery.dataset.ts b/examples/app-showcase/src/datasets/chart-gallery.dataset.ts
index b6388a3c96..5d41ac7f51 100644
--- a/examples/app-showcase/src/datasets/chart-gallery.dataset.ts
+++ b/examples/app-showcase/src/datasets/chart-gallery.dataset.ts
@@ -41,7 +41,11 @@ export const ShowcaseProjectDataset = defineDataset({
   ],
   measures: [
     { name: 'project_count', label: 'Projects', aggregate: 'count' },
-    { name: 'budget_sum', label: 'Total Budget', aggregate: 'sum', field: 'budget', format: '$0,0' },
-    { name: 'spent_sum', label: 'Total Spent', aggregate: 'sum', field: 'spent', format: '$0,0' },
+    // `budget`/`spent` are currency fields with NO declared currency code, so a
+    // hardcoded "$" misrepresents the amount (an amount with unspecified
+    // currency must not show a $ symbol). Use a plain grouped-number format;
+    // declare a `currency` on the field to get a locale-correct symbol via Intl.
+    { name: 'budget_sum', label: 'Total Budget', aggregate: 'sum', field: 'budget', format: '0,0' },
+    { name: 'spent_sum', label: 'Total Spent', aggregate: 'sum', field: 'spent', format: '0,0' },
   ],
 });
diff --git a/packages/services/service-analytics/src/__tests__/query-dataset.test.ts b/packages/services/service-analytics/src/__tests__/query-dataset.test.ts
index 643f6b3e15..99a8d24c90 100644
--- a/packages/services/service-analytics/src/__tests__/query-dataset.test.ts
+++ b/packages/services/service-analytics/src/__tests__/query-dataset.test.ts
@@ -80,4 +80,55 @@ describe('AnalyticsService.queryDataset', () => {
     });
     expect(svc.cubeRegistry.has('sales')).toBe(true);
   });
+
+  // ── ADR-0021 D2 drill-through metadata ──────────────────────────────────
+  it('exposes drill-through metadata: object, dimensionFields, and a raw-value sidecar', async () => {
+    const captured: { sql: string; params: unknown[] }[] = [];
+    const result = await service(captured).queryDataset(
+      dataset,
+      { dimensions: ['region'], measures: ['revenue'] },
+      { tenantId: 'org_A' } as ExecutionContext,
+    ) as any;
+    // The host drills into the dataset's base object…
+    expect(result.object).toBe('opportunity');
+    // …mapping the drillable dimension name to its underlying field…
+    expect(result.dimensionFields).toEqual({ region: 'account.region' });
+    // …and the RAW grouped value is preserved in a parallel array (rows are
+    // NOT mutated — they keep exactly their measure/dimension columns).
+    expect(result.drillRawRows).toEqual([{ region: 'NA' }]);
+    expect(result.rows[0]).toEqual({ region: 'NA', revenue: 100 });
+  });
+
+  it('enriches dimension columns with their dataset display label', async () => {
+    const labeled = DatasetSchema.parse({
+      name: 'sales2', label: 'Sales', object: 'opportunity', include: ['account'],
+      dimensions: [{ name: 'region', field: 'account.region', type: 'string', label: 'Region' }],
+      measures: [{ name: 'revenue', aggregate: 'sum', field: 'amount', label: 'Revenue', certified: true }],
+    });
+    const result = await service([]).queryDataset(
+      labeled,
+      { dimensions: ['region'], measures: ['revenue'] },
+      { tenantId: 'org_A' } as ExecutionContext,
+    ) as any;
+    const regionField = (result.fields ?? []).find((f: any) => f.name === 'region' || f.name === 'account.region');
+    expect(regionField?.label).toBe('Region');
+  });
+
+  it('does NOT mark a date dimension drillable (a humanized bucket cannot be exact-matched)', async () => {
+    const dated = DatasetSchema.parse({
+      name: 'sales3', label: 'Sales', object: 'opportunity', include: [],
+      dimensions: [{ name: 'closed', field: 'close_date', type: 'date' }],
+      measures: [{ name: 'revenue', aggregate: 'sum', field: 'amount', certified: true }],
+    });
+    const svc = new AnalyticsService({
+      queryCapabilities: () => ({ nativeSql: true, objectqlAggregate: false, inMemory: false }),
+      executeRawSql: async () => [{ closed: 1700000000000, revenue: 100 }],
+      getReadScope: (_o, ctx?: ExecutionContext) => (ctx?.tenantId ? { organization_id: ctx.tenantId } : undefined),
+    });
+    const result = await svc.queryDataset(dated, { dimensions: ['closed'], measures: ['revenue'] }, { tenantId: 'org_A' } as ExecutionContext) as any;
+    // No drillable (non-date) dimension → no drill metadata at all.
+    expect(result.dimensionFields).toBeUndefined();
+    expect(result.object).toBeUndefined();
+    expect(result.drillRawRows).toBeUndefined();
+  });
 });
diff --git a/packages/services/service-analytics/src/analytics-service.ts b/packages/services/service-analytics/src/analytics-service.ts
index 4f0aadd5e1..a5975209b4 100644
--- a/packages/services/service-analytics/src/analytics-service.ts
+++ b/packages/services/service-analytics/src/analytics-service.ts
@@ -21,6 +21,25 @@ import { DatasetExecutor } from './dataset-executor.js';
 import { resolveDimensionLabels, type DimensionLabelDeps } from './dimension-labels.js';
 import { evaluateAnalyticsQueryOverRows } from './preview-evaluator.js';
 
+/**
+ * Analytics result augmented with drill-through metadata (ADR-0021 D2; see
+ * queryDataset). Carried alongside `rows` so the host can drill a clicked bucket
+ * back to the underlying records without the renderer knowing field mappings.
+ */
+type AnalyticsResultWithDrill = AnalyticsResult & {
+  /** The dataset's base object — the host drills into its records. */
+  object?: string;
+  /** Selected drillable dimension NAME → underlying object FIELD name. */
+  dimensionFields?: Record<string, string>;
+  /**
+   * RAW grouped values per row, aligned to `rows` by index — each a map of
+   * drillable dimension NAME → stored value (BEFORE label resolution rewrote
+   * `rows[i][dim]` to the display label). The exact-match drill filter is built
+   * from these, never from the display labels.
+   */
+  drillRawRows?: Array<Record<string, unknown>>;
+};
+
 /**
  * Detect the "backing object/table isn't present in this kernel" class of
  * error so a dataset query can degrade to an empty result instead of failing
@@ -441,14 +460,41 @@ export class AnalyticsService implements IAnalyticsService {
       throw err;
     }
 
+    // Selected dimensions resolved against the dataset definition — shared by
+    // drill metadata, label resolution, and dimension field-label enrichment.
+    const selectedDims = (selection.dimensions ?? [])
+      .map((name) => dataset.dimensions?.find((d) => d.name === name))
+      .filter((d): d is NonNullable<typeof d> => !!d);
+
+    // ADR-0021 D2 — drill-through metadata. A host (dashboard/report) drills a
+    // clicked bucket back to the underlying records, but it only knows the
+    // dimension NAMES, and the label resolution below OVERWRITES the raw grouped
+    // value in each row with its display label. So before that happens, snapshot
+    // the raw grouped values into a PARALLEL array (aligned to `rows` by index —
+    // the result rows are NOT mutated) and expose the dataset's `object` +
+    // dimension→field mapping so the renderer can build an exact-match filter.
+    // Date buckets are excluded — a humanized bucket ("2026-06") can't be
+    // exact-matched against the stored timestamp, so they are not drillable.
+    const drillDims = selectedDims.filter((d) => !!d.field && d.type !== 'date');
+    if (drillDims.length && result.rows.length) {
+      (result as AnalyticsResultWithDrill).object = dataset.object;
+      (result as AnalyticsResultWithDrill).dimensionFields = Object.fromEntries(
+        drillDims.map((d) => [d.name, d.field as string]),
+      );
+      (result as AnalyticsResultWithDrill).drillRawRows = result.rows.map((row) => {
+        const raw: Record<string, unknown> = {};
+        for (const d of drillDims) raw[d.name] = row[d.name];
+        return raw;
+      });
+    }
+
     // ADR-0021 — resolve grouped dimension values to human display labels
     // (select option label, lookup related-record name). Charts render the
     // dimension key verbatim, so this is the single place that turns a stored
     // value / FK id into the text a user expects to read.
-    if (this.labelResolver && selection.dimensions?.length) {
-      const dims = selection.dimensions
-        .map((name) => dataset.dimensions?.find((d) => d.name === name))
-        .filter((d): d is NonNullable<typeof d> => !!d?.field)
+    if (this.labelResolver && selectedDims.length) {
+      const dims = selectedDims
+        .filter((d) => !!d.field)
         .map((d) => ({ name: d.name, field: d.field, type: d.type, dateGranularity: d.dateGranularity }));
       if (dims.length) {
         try {
@@ -480,6 +526,22 @@ export class AnalyticsService implements IAnalyticsService {
         if (f.format == null && m.format) f.format = m.format;
       }
     }
+
+    // Enrich DIMENSION columns with their display `label` too, so a grouped
+    // table header reads "Status" instead of the raw field name "status". The
+    // measure-only enrichment above left dimension headers bare (the renderer
+    // then fell back to the raw dimension name).
+    if (result.fields?.length && selectedDims.length) {
+      const dimByName = new Map(selectedDims.map((d) => [d.name, d]));
+      const dimByField = new Map(selectedDims.filter((d) => !!d.field).map((d) => [d.field as string, d]));
+      for (const f of result.fields) {
+        if (f.label != null) continue;
+        // Result fields may be keyed by the dataset dimension NAME or the
+        // underlying cube FIELD depending on strategy — match either.
+        const d = dimByName.get(f.name) ?? dimByField.get(f.name);
+        if (d && typeof d.label === 'string') f.label = d.label;
+      }
+    }
     return result;
   }