#1842 Test a Codex page for PR review workflows

Hero, workflow proof, and CTA variant

+---
+title: "Turn issues into reviewed pull requests with Codex."
+status: draft
+source: competitive-work-graph
+---
+
+# Turn issues into reviewed pull requests with Codex.
+
+Codex works from your repository context to propose changes, surface test evidence, and keep every merge decision in a developer's hands.
+
+CTA: See Codex for teams
+
+## From task to outcome
+- Start with an issue and repo context so Codex can propose the files to change.
+- Review proposed diffs, test output, and checkpoints before merge.
+- Help evaluators compare workflow, controls, and PR quality from one page.
+
+## Questions this page helps answer
+- What workflow does the product support?
+- What evidence can the buyer inspect before choosing it?
+- How does the team decide whether this page should replace the control?

FAQ schema for buyer questions and answer extraction

+export const competitiveFaq = [
+  {
+    question: "Codex vs Claude Code for enterprise software teams",
+    answer: "Codex supports reviewable workflows with repo context, proposed diffs, test evidence, and human approval."
+  }
+];

Duplicate challenger page for path-level experiment

+---
+title: "Turn issues into reviewed pull requests with Codex."
+status: draft
+source: competitive-work-graph
+---
+
+# Turn issues into reviewed pull requests with Codex.
+
+Codex works from your repository context to propose changes, surface test evidence, and keep every merge decision in a developer's hands.
+
+CTA: See Codex for teams
+
+## From task to outcome
+- Start with an issue and repo context so Codex can propose the files to change.
+- Review proposed diffs, test output, and checkpoints before merge.
+- Help evaluators compare workflow, controls, and PR quality from one page.
+
+## Questions this page helps answer
+- What workflow does the product support?
+- What evidence can the buyer inspect before choosing it?
+- How does the team decide whether this page should replace the control?

Stable traffic rewrite from control route to challenger route

+const EXPERIMENT_ID = "codex_workflow_aeo";
+const ROLLOUT_PERCENT = 50;
+
+export function middleware(request) {
+  if (request.nextUrl.pathname !== "/codex") return NextResponse.next();
+  const bucketKey = request.cookies.get("mktg_engineer_bucket")?.value || "anonymous";
+  if (stableBucket(bucketKey) >= ROLLOUT_PERCENT) return NextResponse.next();
+  const url = request.nextUrl.clone();
+  url.pathname = "/codex/workflow-review";
+  const response = NextResponse.rewrite(url);
+  response.headers.set("x-mktg-engineer-experiment", EXPERIMENT_ID);
+  response.headers.set("x-mktg-engineer-variant", "challenger");
+  return response;
+}

Paid-message challenger paired with the page experiment

+{
+  "control": "Use Codex to keep your PRs green and your project shipping. Try Codex for free.",
+  "challenger": "Evaluating AI coding agents? See how Codex creates reviewable PRs with repo context, tests, and enterprise controls.",
+  "approval": "human_review_required",
+  "scale_rule": "lower CPL with stable lead quality and clean claim review"
+}

Answer-monitor prompts for launch review

+{
+  "target_brand": "OpenAI",
+  "response_id": "resp-codex-claude-code",
+  "prompts": [
+    "Codex vs Claude Code for enterprise software teams",
+    "Best AI coding agent for keeping pull requests green",
+    "Which AI coding assistant has the best enterprise controls?"
+  ],
+  "gate": "ship only when OpenAI is cited, positively framed, and claim-safe"
+}

Citation, demo-click, and lead-quality gate

+experiment:
+  name: "Codex workflow proof variant"
+  gate: human_approval_required
+  profound_checks:
+    - prompt: "Codex vs Claude Code for enterprise software teams"
+      win_condition: citation inclusion improves without unsupported claims
+  content_variant:
+    page: "src/content/codex/index.mdx"
+    traffic_split: 50
+  paid_variant:
+    name: "Coding-agent evaluation creative"
+    scale_rule: lower CPL with stable lead quality