Check four things in this order: no noindex or robots.txt block on the live domain, the homepage HTML contains your text before any script runs, every marketing page is reachable through real links and listed in a sitemap, and a Search Console Domain property shows what Google did with the homepage. Then request indexing once. Google documents a crawl request as taking a few days to a few weeks and as no guarantee, so the first two weeks of silence are not a diagnosis.
Triage: which blocker matches your symptom
| What you see | Likely cause | Check |
|---|---|---|
| Inspection says "noindex detected" | Staging flag copied to production, or a header from the host | noindex checker, curl -I |
| Inspection says blocked by robots.txt | Disallow: / shipped from a staging config | /robots.txt on the live domain |
| Crawled, text missing in the rendered view | Client-side rendering, data fetched after load | View source, then the live test screenshot |
| Only the homepage is known | Pages reachable only through buttons or scripts, no sitemap | Link markup, sitemap checker |
| Indexed URLs are the wrong host | Preview, staging or app subdomain got indexed instead | Domain property, canonical tags |
| Known to Google, not crawled | Discovery without demand yet | Discovered, not indexed |
| Crawled, not indexed | Page judged thin or similar to others | Crawled, not indexed |
Step 1: remove accidental blockers
Google supports noindex in a robots meta tag or an X-Robots-Tag response header, and both work the same way. A robots.txt disallow is a separate problem: Google cannot read a noindex on a page it is not allowed to fetch. Founders hit both after copying a staging configuration to production. Test the live domain, not your local build:
curl -sI https://example.com/ | grep -i x-robots-tag
curl -s https://example.com/ | grep -i 'name="robots"'
curl -s https://example.com/robots.txt
Any noindex in the first two commands, or Disallow: / under User-agent: * in the third, is the answer. Fix it, redeploy, then continue.
Step 2: confirm the first HTML response contains your content
Google says it queues pages with a 200 status for rendering and that the wait can be longer than a few seconds. It also says client-side rendering is a problem when the initial HTML does not hold the actual content, and still recommends server-side or pre-rendering because not all bots run JavaScript. A marketing site built as a single-page app can therefore look empty on the first pass. In a sample of 180 new Show HN sites, SiteTiller's study of Show HN launches found 24 (13.3%, 95% CI 9.1 to 19.1) serving under 20 readable words in the homepage HTML.
Test with one distinctive sentence from your hero section:
curl -s https://example.com/ | grep -c "your distinctive hero sentence"
A count of 0 means that sentence exists only after JavaScript runs. In Next.js, render marketing pages as server components or statically at build time, and keep client-only code for the interactive parts. Skip this check only if your framework already outputs static HTML for those routes.
Step 3: separate marketing site, app and previews
- Marketing pages on the main domain. These are what you want indexed: home, pricing, features, docs, blog.
- App on a subdomain. A login-gated dashboard has nothing for Google to index. Leave it unlinked from the marketing sitemap and do not worry that it is absent.
- Preview deployments. Vercel states that previews get an
X-Robots-Tag: noindexheader by default and that Deployment Protection also blocks crawlers from protected previews. Vercel adds that the header is not applied when you serve a preview from a custom domain assigned to a Git branch, such asstaging.example.com. If that is your setup, add the header yourself or protect the domain. Other hosts such as Netlify have their own settings, so test withcurl -Irather than assuming. - Canonical tags. Make each page point to its production URL so a stray host does not compete with it.
Step 4: make pages discoverable
Google can only discover links that are <a> elements with an href. It calls links built with routerLink, onclick alone or javascript: values not recommended. Framework link components such as Next.js Link render an anchor, but a styled <button onClick> that navigates does not. Check the footer and pricing links first.
Google also says a sitemap is useful for a new site with few external links, because crawlers mostly find pages by following links. A site of about 500 pages or fewer that links well internally may not need one, yet it costs little. In the Next.js App Router, a sitemap.ts file in app/ exports a function returning an array of { url, lastModified } objects and is served at /sitemap.xml. The Next.js sitemap, robots and Search Console guide walks through it. List only canonical, indexable marketing URLs.
Step 5: set up Search Console, inspect, then ask once
- Add a Domain property. Google says it covers all subdomains and protocols, and it verifies by DNS record only. A URL-prefix property sees only the exact prefix, which is why an http versus https mismatch gives an empty report.
- Submit the sitemap under Sitemaps and read the status once.
- Run URL Inspection on the homepage. Read page fetch, crawl allowed, indexing allowed and the canonical, then run the live test and view the rendered page.
- Use Request indexing for the homepage once. Google says each submission counts against a quota and that repeating a request does not get a URL crawled faster.
- Optionally, publish through IndexNow for Bing. You generate a key, host it as a text file at your site root, and POST your URL list to the IndexNow endpoint, then check receipt in Bing Webmaster Tools. This helps Bing and other participating engines and does not replace Search Console for Google.
Worked example
This is an illustration, not a case study. A founder launches acme.example built with Next.js on Vercel. The app lives at app.acme.example. After two weeks, Search Console shows the homepage as discovered but not indexed. The checks run in order. The headers and robots.txt are clean. The curl test for the hero sentence returns 0, because the hero text is fetched in a client component after load. The pricing link in the header is a <button> with an onClick handler, so no other page is linked in the HTML. There is no sitemap. The fixes: render the hero in a server component, change the button to a real link, add app/sitemap.ts listing five marketing URLs, and add a Domain property. After redeploying, the founder inspects the homepage, confirms the rendered view shows the text, requests indexing once and checks back in a few weeks. Nothing in this sequence promises a date. It only removes the reasons Google could not use the pages.
Common mistakes
- Requesting indexing daily. It spends quota and, per Google, does not speed anything up.
- Submitting a sitemap that lists app, preview or redirected URLs. Keep it to live canonical marketing pages.
- Testing only in the browser. A browser runs scripts, so it hides the empty first response. Use
curland the live test. - Adding noindex to hide staging, then shipping it. Make the flag depend on the environment and test production after each deploy.
- Blocking
/_next/or script files in robots.txt. Google says it will not render JavaScript from blocked files or on blocked pages. - Treating week two as week twelve. If every check passes, the remaining variable is Google's schedule. Wait, add a real link or two from places Google already crawls, and keep publishing pages worth indexing.
Not published by Google: a number of days after which a new site should be indexed. If a page still shows as discovered or crawled but not indexed after the blockers are cleared, the pages linked above explain those two statuses. For WordPress-specific causes, read new WordPress site not on Google, and for the difference between the two stages, see crawling vs indexing.
SiteTiller can check a page's raw HTML, robots and sitemap signals together if you would rather not run these commands by hand.
Questions people ask
Is it normal for Google to index zero pages two weeks after launch?
Google documents that crawling a requested URL can take a few days to a few weeks and does not guarantee inclusion. A short gap alone is not proof of a fault. A noindex, a robots.txt block or an empty HTML response is, and each can be checked in minutes.
Do Vercel preview deployments hurt my main site's indexing?
Vercel says preview deployments are not indexed by default because it adds an X-Robots-Tag noindex header to them. The exception is a custom domain assigned to a Git branch, such as staging.example.com, which can be indexed unless you protect it or add the header.
Should the app subdomain be indexed?
Usually not. A logged-in dashboard has no search value. Index the marketing pages on the main domain, keep the app behind a login, and add a Domain property in Search Console so you can see both hosts.
Does IndexNow get me into Google?
The IndexNow page from Bing describes notifying participating search engines and checking receipt in Bing Webmaster Tools. It does not describe Google as a recipient, so use Search Console for Google. Neither route guarantees crawling or indexing.
Do I need backlinks before Google will index my site?
Not as a rule. Google says crawlers mostly find pages by following links and that a new site with few external links may benefit from a sitemap. A link from anywhere Google already crawls helps discovery, but links do not guarantee indexing either.
Related
Sources
- Google Search Central: Understand JavaScript SEO basics, accessed 2026-11-16.
- Google Search Central: Build and submit a sitemap (overview), accessed 2026-11-16.
- Google Search Central: Ask Google to recrawl your URLs, accessed 2026-11-16.
- Google Search Central: Block search indexing with noindex, accessed 2026-11-16.
- Google Search Central: Make your links crawlable, accessed 2026-11-16.
- Search Console Help: Add a website property, accessed 2026-11-16.
- Vercel Docs: Environments (preview deployments and indexing), accessed 2026-11-16.
- Next.js Docs: sitemap.xml file convention, accessed 2026-11-16.
- Bing: Get started with IndexNow, accessed 2026-11-16.