404/5xx SOP: The Repeatable Process We Use in 2025

Introduction

Downtime and dead links quietly drain revenue. Our 2025 SOP turns 404/5xx chaos into a repeatable, measurable workflow you can run every week—or during incidents. Use SEO Horizan tools for fast QA (redirects, headers, snippet checks) and keep error budgets under control.

Scope & Definitions

• 404 (Not Found): Client requested a path that doesn’t exist. Often caused by URL changes, typos, or removed content.
• 410 (Gone): Resource intentionally and permanently removed. Prefer over 404 for deliberate retirements.
• 5xx (Server Errors): Origin or edge failed to serve (500/502/503/504). Affects crawl, UX, and trust signals.
• Error budget: Agreed monthly threshold for 5xx rate and % of broken internal links.

SOP Overview (Trigger → Actions → Owner → SLA)

Trigger: Rising 404s or 5xx in logs/monitor
Actions: Triage → Prioritize → Fix/Redirect → QA → Monitor → Postmortem
Owner: SEO Eng (links), Platform Eng (5xx), Content Ops (mappings)
SLA: P0 live issues ≤ 2h, P1 ≤ 24h, P2 weekly batch

Step 1 — Detect & Aggregate

• Sources: server logs, edge logs, uptime monitors, GSC coverage, analytics 404 pageviews.
• Normalize: Lowercase paths, strip params, collapse trailing slashes. Group by template or referrer.
• Snapshot: Export a CSV with URL • Hits • First seen • Last seen • Referrer • Status.

Step 2 — Triage (classify fast)

• Internal link regressions: Broken links in nav/footer/related blocks.
• Moved/renamed content: Slug changes, category moves, site merges.
• Programmatic noise: Facet/search params, bots probing old paths.
• Platform outages: Spikes in 5xx across templates or regions.

Step 3 — Prioritize (impact × fix speed)

• P0: 5xx spikes; top-landing URLs; checkout/docs/login flows.
• P1: High-referrer 404s from internal links or fresh campaigns.
• P2: Long-tail/bot noise; archive cleanup; low-referrer paths.

Step 4 — Fix Paths (404/410/301 rules)

• Internal link bugs: Update the source link to the final canonical. Kill chains and mixed hosts.
• Moved content: 301 one-to-one to the closest topical match. Avoid blanket folder-to-home redirects.
• Retired content: Serve 410 (Gone) when there’s no equivalent replacement.
• Param noise: Add routing rules to normalize or drop known junk params.

Verify final status with URL Redirect Checker and headers via HTTP Headers Lookup.

Step 5 — Stabilize Server/Edge (5xx playbook)

• Traffic shape: Enable circuit breakers and autoscaling; serve stale-while-revalidate if cacheable.
• Origin health: Roll back the last deploy if 5xx started post-release; check DB connections and timeouts.
• Dependency map: Identify failing upstreams (search, payments, CMS APIs) and degrade gracefully.
• Status selection: Prefer 503 + Retry-After for maintenance; avoid generic 500 loops.

Step 6 — Rebuild Internal Links

• Run a crawl of nav/footer/related blocks; update CMS menus, MDX, and internal indexes.
• Ensure hub → spokes → proof patterns link to final 200 pages only.
• Re-publish key hubs to refresh discovery; confirm entries in your Sitemap.

Step 7 — QA Checklist (copy this)

• Target URL resolves 200 with no intermediate 3xx (Redirect Checker).
• Headers sane: Content-Type, cache, canonical (Headers Lookup).
• Meta/OG parity on top targets (Meta Tags • OpenGraph).
• Snippet (40–55 words) present under H1 (Text Extractor).
• Performance guardrails: TTFB < 600 ms, payload < 2 MB (TTFB • Page Size).

Step 8 — Monitor & Communicate

• Dashboards: 404 hits/day, 5xx rate, top referrers, affected templates.
• Alerting: P0 if 5xx > budget or 404s spike > X% in 15 minutes.
• Changelog: document rule adds/removals, redirect maps, and deploys.

Step 9 — Postmortem (within 72 hours)

• Root cause: code, config, content, infra, or third-party.
• Blast radius: sessions lost, revenue impact, organic entrances affected.
• Prevention: tests, monitors, ownership, rollback criteria.

Redirect Mapping Template

From URL, To URL, Match Type (exact/prefix/regex), Reason (moved/typo/merge), Date, Owner, Notes

Incident Worksheet (CSV)

URL, Status, Hits, First Seen, Last Seen, Referrer (internal/external), Template, Priority (P0/P1/P2), Fix (301/410/Update Link/Origin Fix), Owner, Due, QA Passed (Y/N)

Governance & Error Budgets

• 5xx budget: e.g., <0.2% of requests/month. Breach triggers freeze + review.
• Broken link budget: <0.5% of internal link clicks may result in 404/410.
• Ownership: Each template has a technical owner and content owner.

FAQs