diff --git a/packages/root-cms/cli/cli.ts b/packages/root-cms/cli/cli.ts
index 725ddbffd..40f484429 100644
--- a/packages/root-cms/cli/cli.ts
+++ b/packages/root-cms/cli/cli.ts
@@ -2,6 +2,7 @@ import {Command} from 'commander';
 import {bgGreen, black} from 'kleur/colors';
 import {generateTypes} from './generate-types.js';
 import {initFirebase} from './init-firebase.js';
+import {pruneTranslations} from './prune-translations.js';
 
 class CliRunner {
   private name: string;
@@ -37,8 +38,15 @@ class CliRunner {
         'generates root-cms.d.ts from *.schema.ts files in the project'
       )
       .action(generateTypes);
+    program
+      .command('prune_translations <doc>')
+      .alias('prune-translations')
+      .description(
+        'removes unused translations tagged with the provided doc id'
+      )
+      .action(pruneTranslations);
     await program.parseAsync(argv);
   }
 }
 
-export {CliRunner, generateTypes, initFirebase};
+export {CliRunner, generateTypes, initFirebase, pruneTranslations};
diff --git a/packages/root-cms/cli/prune-translations.ts b/packages/root-cms/cli/prune-translations.ts
new file mode 100644
index 000000000..405bd0236
--- /dev/null
+++ b/packages/root-cms/cli/prune-translations.ts
@@ -0,0 +1,274 @@
+import path from 'node:path';
+import {fileURLToPath} from 'node:url';
+
+import {loadRootConfig, viteSsrLoadModule} from '@blinkk/root/node';
+import {FieldValue} from 'firebase-admin/firestore';
+
+import {RootCMSClient} from '../core/client.js';
+import type {Field, Schema} from '../core/schema.js';
+import type {
+  RichTextBlock,
+  RichTextData,
+  RichTextListItem,
+  RichTextTableRow,
+} from '../shared/richtext.js';
+
+const __dirname = path.dirname(fileURLToPath(import.meta.url));
+
+type ProjectModule = typeof import('../core/project.js');
+
+type NormalizeStringFn = (str: string) => string;
+
+export async function pruneTranslations(docId: string) {
+  if (!docId || !docId.includes('/')) {
+    throw new Error('doc id must be in the format "<collection>/<slug>".');
+  }
+
+  const [collectionId, ...slugParts] = docId.split('/');
+  const slug = slugParts.join('/');
+  if (!collectionId || !slug) {
+    throw new Error('doc id must include both collection and slug.');
+  }
+
+  const rootDir = process.cwd();
+  const rootConfig = await loadRootConfig(rootDir, {command: 'root-cms'});
+  const cmsClient = new RootCMSClient(rootConfig);
+  const normalizeString = cmsClient.normalizeString.bind(cmsClient);
+
+  const modulePath = path.resolve(__dirname, './project.js');
+  const project = (await viteSsrLoadModule(
+    rootConfig,
+    modulePath
+  )) as ProjectModule;
+  const collectionSchema = await project.getCollectionSchema(collectionId);
+  if (!collectionSchema) {
+    console.warn(`collection schema not found for ${collectionId}`);
+    return;
+  }
+
+  const docStrings = new Set<string>();
+  for (const mode of ['draft', 'published'] as const) {
+    const doc = await cmsClient.getDoc(collectionId, slug, {mode});
+    if (!doc) {
+      continue;
+    }
+    extractFields(
+      docStrings,
+      collectionSchema.fields,
+      doc.fields || {},
+      collectionSchema.types || {},
+      normalizeString
+    );
+  }
+
+  const taggedTranslations = await cmsClient.loadTranslations({tags: [docId]});
+  const unusedTranslations: Array<{hash: string; source: string}> = [];
+  Object.entries(taggedTranslations).forEach(([hash, translation]) => {
+    const source = normalizeString(translation.source);
+    if (!docStrings.has(source)) {
+      unusedTranslations.push({hash, source});
+    }
+  });
+
+  const taggedCount = Object.keys(taggedTranslations).length;
+  console.log(
+    `found ${taggedCount} translation(s) tagged with "${docId}".`
+  );
+  if (unusedTranslations.length === 0) {
+    console.log('no unused translations found.');
+    return;
+  }
+
+  console.log(`removing tag from ${unusedTranslations.length} translation(s)...`);
+  const translationsPath = `Projects/${cmsClient.projectId}/Translations`;
+  const batch = cmsClient.db.batch();
+  unusedTranslations.forEach(({hash}) => {
+    const ref = cmsClient.db.doc(`${translationsPath}/${hash}`);
+    batch.update(ref, {tags: FieldValue.arrayRemove(docId)});
+  });
+  await batch.commit();
+
+  console.log('removed tags from the following sources:');
+  unusedTranslations.forEach(({source}) => {
+    console.log(`- ${source}`);
+  });
+}
+
+function extractFields(
+  strings: Set<string>,
+  fields: Field[],
+  data: Record<string, any>,
+  types: Record<string, Schema> = {},
+  normalizeString: NormalizeStringFn
+) {
+  fields.forEach((field) => {
+    if (!field.id) {
+      return;
+    }
+    const fieldValue = data[field.id];
+    extractField(strings, field, fieldValue, types, normalizeString);
+  });
+}
+
+function extractField(
+  strings: Set<string>,
+  field: Field,
+  fieldValue: any,
+  types: Record<string, Schema> = {},
+  normalizeString: NormalizeStringFn
+) {
+  if (!fieldValue) {
+    return;
+  }
+
+  function addString(text: string) {
+    const str = normalizeString(text);
+    if (str) {
+      strings.add(str);
+    }
+  }
+
+  if (field.type === 'object') {
+    extractFields(strings, field.fields || [], fieldValue, types, normalizeString);
+  } else if (field.type === 'array') {
+    const arrayKeys = fieldValue._array || [];
+    for (const arrayKey of arrayKeys) {
+      extractField(
+        strings,
+        field.of,
+        fieldValue[arrayKey],
+        types,
+        normalizeString
+      );
+    }
+  } else if (field.type === 'string' || field.type === 'select') {
+    if (field.translate) {
+      addString(fieldValue);
+    }
+  } else if (field.type === 'image') {
+    if (field.translate && fieldValue && fieldValue.alt && field.alt !== false) {
+      addString(fieldValue.alt);
+    }
+  } else if (field.type === 'multiselect') {
+    if (field.translate && Array.isArray(fieldValue)) {
+      for (const value of fieldValue) {
+        addString(value);
+      }
+    }
+  } else if (field.type === 'oneof') {
+    const fieldTypes = field.types || [];
+    let fieldValueType: Schema | undefined;
+    if (typeof fieldTypes[0] === 'string') {
+      const typeId = fieldValue._type as string;
+      if ((fieldTypes as string[]).includes(typeId)) {
+        fieldValueType = types[typeId];
+      }
+    } else {
+      fieldValueType = (fieldTypes as any[]).find(
+        (item: any) => item.name === fieldValue._type
+      );
+    }
+    if (fieldValueType) {
+      extractFields(
+        strings,
+        fieldValueType.fields || [],
+        fieldValue,
+        types,
+        normalizeString
+      );
+    }
+  } else if (field.type === 'richtext') {
+    if (field.translate) {
+      extractRichTextStrings(strings, fieldValue, normalizeString);
+    }
+  }
+}
+
+function extractRichTextStrings(
+  strings: Set<string>,
+  data: RichTextData,
+  normalizeString: NormalizeStringFn
+) {
+  const blocks = data?.blocks || [];
+  blocks.forEach((block) => {
+    extractBlockStrings(strings, block, normalizeString);
+  });
+}
+
+function extractBlockStrings(
+  strings: Set<string>,
+  block: RichTextBlock,
+  normalizeString: NormalizeStringFn
+) {
+  if (!block?.type) {
+    return;
+  }
+
+  function addString(text?: string) {
+    if (!text) {
+      return;
+    }
+    const str = normalizeString(text);
+    if (str) {
+      strings.add(str);
+    }
+  }
+
+  function addComponentStrings(components?: Record<string, any>) {
+    if (!components) {
+      return;
+    }
+    Object.values(components).forEach((component) => {
+      collectComponentStrings(component);
+    });
+  }
+
+  function collectComponentStrings(value: any) {
+    if (typeof value === 'string') {
+      addString(value);
+      return;
+    }
+    if (Array.isArray(value)) {
+      value.forEach((item) => collectComponentStrings(item));
+      return;
+    }
+    if (isPlainObject(value)) {
+      Object.values(value).forEach((item) => collectComponentStrings(item));
+    }
+  }
+
+  function extractList(items?: RichTextListItem[]) {
+    if (!items) {
+      return;
+    }
+    items.forEach((item) => {
+      addString(item.content);
+      addComponentStrings(item.components);
+      extractList(item.items);
+    });
+  }
+
+  if (block.type === 'heading' || block.type === 'paragraph') {
+    addString(block.data?.text);
+    addComponentStrings(block.data?.components);
+  } else if (block.type === 'orderedList' || block.type === 'unorderedList') {
+    extractList(block.data?.items);
+  } else if (block.type === 'html') {
+    addString(block.data?.html);
+  } else if (block.type === 'table') {
+    const rows = block.data?.rows || [];
+    rows.forEach((row: RichTextTableRow) => {
+      const cells = row.cells || [];
+      cells.forEach((cell) => {
+        const cellBlocks = cell.blocks || [];
+        cellBlocks.forEach((cellBlock) => {
+          extractBlockStrings(strings, cellBlock, normalizeString);
+        });
+      });
+    });
+  }
+}
+
+function isPlainObject(value: unknown): value is Record<string, unknown> {
+  return typeof value === 'object' && value !== null && !Array.isArray(value);
+}