Website Technical Audit Checklist: What to Check Before Launch or Handover
Use a repeatable checklist to catch the failures that a pretty homepage can hide, before handover or a big deployment.
5 min readRead guide →
A web app can feel excellent to a human and still give a search engine almost nothing useful on the first request. The page loads a shell, fetches data, paints the interface and assumes every crawler will run all the same code. Google can render JavaScript, but the crawl, render and indexing steps are not identical.
Quick answer: For any public page you want indexed, serve useful HTML with a meaningful title, main heading, description, canonical, text and real links on the initial response. Make the page return the right HTTP status. Keep private screens authenticated and outside the public sitemap.
1. Look at the HTML response without JavaScript
Open View Source rather than just Inspect Element. The former shows the initial server response. The latter often shows content added after JavaScript execution. Search Google’s JavaScript SEO documentation for the same distinction: a `200` page may be queued for rendering, but relying entirely on that later step creates extra dependencies.
Search-important content should not be hidden behind an API that needs a browser session, a cookie consent click, a failed WebSocket or a Cloudflare challenge. Use server rendering, static generation or a pre-rendered HTML fallback. Progressive enhancement lets the interactive app remain rich without making discovery fragile.
2. Make the status code tell the truth
| Page state | Useful HTTP signal | Why |
|---|---|---|
| Valid public guide or item page | 200 | Content exists |
| Moved permanently | 301 or 308 | Update canonical destination |
| Unknown item or deleted article | 404 or 410 | Do not pretend an error is a real article |
| Private page requiring sign-in | 401 or protected application flow | Do not expose private content |
| Rate limited | 429 with appropriate handling | Can interfere with crawl access |
Do not return a branded "page not found" design with status `200`. That can trigger a soft 404 and confuse monitoring. If the user is looking for a guide that no longer exists, link to related guides from a genuine error page.
3. Keep one stable URL per resource
Every article should have its own clean URL and a matching canonical link. If you provide sorting, category toggles and internal search, avoid generating thousands of indexable duplicates from every filter combination. Google's faceted-navigation guidance describes how that can exhaust crawl resources.
Example URL policy
/blog/guide-title/: indexable standalone guide./blog/: indexable directory./blog/?q=guide: same directory canonical, not a new article./account/tickets/: private area, protected by authentication.
4. Test actual robots headers, not only meta tags
Two different systems can set indexing instructions. A page may contain <meta name="robots" content="index, follow"> while the server sends X-Robots-Tag: noindex. The more restrictive instruction wins. A successful sitemap upload does not override a `noindex`, a `403`, a challenge or missing content.
Inspect raw response headers and page source for the exact canonical URL. Also check whether security rules challenge legitimate, verified search crawlers. Never turn off security for everyone just to fix indexing.
5. Audit one URL from top to bottom
- Request it without cookies and verify the response code.
- Check the original HTML for the H1, helpful body text and internal links.
- Confirm robots meta and HTTP headers agree.
- Confirm canonical URL matches the sitemap URL.
- Check mobile usability and essential resources.
- Inspect the live URL in Google Search Console and compare Google's selected canonical.
- Make a single material fix, then validate after recrawling.
For a game database, do not confuse changing the page title with changing the content. A unique title above the same generic paragraph on hundreds of pages adds little value. Give high-priority pages genuinely unique data, explanations and links.
6. Measure pages, not just sitewide impressions
Break Search Console results down by page group and query. Separate navigational queries, game-guide searches and product-intent queries. A new tool can get an early burst of impressions that never becomes stable. A large unindexed total can also include legitimate canonical variants rather than site errors.
Bottom line: build the useful page in HTML first, then let JavaScript add interaction. It is simpler to diagnose and more resilient when browsers, crawlers or third-party scripts behave differently.
A small server-response test that catches big problems
Pick an important app URL and open its initial HTML without executing JavaScript. Does the response contain its actual page heading, a helpful description and enough unique information to explain the product or dataset? If it only says "Loading...", search engines may struggle to assess the page even if the interactive application works perfectly in a browser.
Repeat the test for a second route. The title, canonical, headings and structured data should describe that route rather than copy the application homepage. A dynamic filter combination that has no independent purpose usually does not need a new indexable URL.
| Route type | Search-friendly default |
|---|---|
| Main application landing page | Index, useful HTML explanation, self-canonical |
| Substantial standalone guide or item page | Index if it contains unique useful data |
| Search input and sort states | Consolidate rather than index each combination |
| Authenticated client dashboard | Require login; exclude from search |
| Invalid dynamic identifier | Real 404 or appropriate redirect |
Test with more than a lighthouse score
Inspect server headers, then view the actual rendered DOM in a browser. Test the response that a verified crawler receives if Cloudflare or a web application firewall is present. Finally, check internal anchor links: navigation implemented only through JavaScript event handlers may leave routes hard to discover. Our Search Console troubleshooting guide covers indexation symptoms.
