diff --git a/src/app/alternatives/[slug]/page.tsx b/src/app/alternatives/[slug]/page.tsx
index 73df1a31..f58b3563 100644
--- a/src/app/alternatives/[slug]/page.tsx
+++ b/src/app/alternatives/[slug]/page.tsx
@@ -159,7 +159,10 @@ export default async function AlternativePage({
         ...(citableAsOf(bench) ? { dateModified: bench.lastRunAt } : {}),
         author: { "@id": `${SITE.url}/#org` },
         publisher: { "@id": `${SITE.url}/#org` },
-        about: { "@id": `${url}#dataset` },
+        // `about` describes the topic. Use a Thing rather than @id-refing
+        // the same-page Dataset (Google validates @id-only refs as
+        // standalone incomplete Datasets even within the same @graph).
+        about: { "@type": "Thing", name: bench.metric },
       },
       buildBreadcrumbJsonLd([
         { name: "Home", item: SITE.url },
diff --git a/src/app/benchmarks/[slug]/[chain]/page.tsx b/src/app/benchmarks/[slug]/[chain]/page.tsx
index 61b993c5..7564acef 100644
--- a/src/app/benchmarks/[slug]/[chain]/page.tsx
+++ b/src/app/benchmarks/[slug]/[chain]/page.tsx
@@ -375,8 +375,20 @@ export default async function BenchmarkChainPage({
         author: [{ "@id": PERSON_ID }, { "@id": `${SITE.url}/#org` }],
         reviewedBy: { "@id": PERSON_ID },
         publisher: { "@id": `${SITE.url}/#org` },
-        about: { "@id": `${benchmarkUrl}#dataset` },
-        isPartOf: { "@id": `${benchmarkUrl}#article` },
+        // `about` describes the topic. Point at the metric as a Thing
+        // rather than @id-referencing the parent bench Dataset (cross-doc
+        // @id refs get validated as standalone incomplete Datasets by
+        // Google, same class of issue as the isPartOf/isBasedOn fixes).
+        about: { "@type": "Thing", name: benchmark.metric },
+        // isPartOf: parent bench article. Inline the required fields so
+        // the reference validates on its own (Google does not stitch
+        // cross-page @id refs).
+        isPartOf: {
+          "@type": "TechArticle",
+          "@id": `${benchmarkUrl}#article`,
+          name: benchmark.title,
+          url: benchmarkUrl,
+        },
       },
       buildBreadcrumbJsonLd([
         { name: "Home", item: SITE.url },
diff --git a/src/app/benchmarks/[slug]/page.tsx b/src/app/benchmarks/[slug]/page.tsx
index 966eda10..a7170b75 100644
--- a/src/app/benchmarks/[slug]/page.tsx
+++ b/src/app/benchmarks/[slug]/page.tsx
@@ -419,7 +419,12 @@ export default async function BenchmarkPage({
         author: [{ "@id": PERSON_ID }, { "@id": `${SITE.url}/#org` }],
         reviewedBy: { "@id": PERSON_ID },
         publisher: { "@id": `${SITE.url}/#org` },
-        about: { "@id": `${benchmarkUrl}#dataset` },
+        // `about` describes the topic. Use a Thing (matches the
+        // StatisticalReport `about` pattern above) rather than @id-refing
+        // the same-page Dataset. Google's Dataset validator was flagging
+        // @id-only refs as standalone incomplete Datasets even within
+        // the same @graph (see the isBasedOn fix in dataset-jsonld.ts).
+        about: { "@type": "Thing", name: benchmark.metric },
       },
       buildBreadcrumbJsonLd([
         { name: "Home", item: SITE.url },