You published. You inspected. You clicked. Google said no.
The flow every SEO knows by heart: publish an article, open Google Search Console, paste the URL into the inspection box, wait for “URL is not on Google”, click Request Indexing — and then:
Oops! Something went wrong We had a problem submitting your indexing request. Please try again later.
No error code. No reason. No retry timer. And “try again later” is a lie often enough — for some accounts, the dialog comes back tomorrow, next week, every single time.
We hit this exact wall while building our own indexing automation, burned a few days debugging it, and ended up mapping out the rules Google never wrote down. Here’s everything we learned, so you don’t have to repeat the experiment.
Rule 1: There is no visible quota — anywhere
The first thing everyone tries to find is the quota meter. It doesn’t exist.
Google does not expose the Request Indexing limit in the Search Console interface, in any API, or in any documentation. You cannot ask how many requests you have left. The URL Inspection API has a published quota, but it only inspects — it cannot request indexing. The Indexing API can, but only for JobPosting and BroadcastEvent pages; regular blog content is explicitly out of scope.
So the only quota signal you will ever get is the refusal itself. Community testing converges on roughly 10 requests per property per day, with the real number floating lower for new or low-trust properties.
Rule 2: The error message hides three different problems
The same generic dialog covers at least three distinct situations:
1. The daily cap. You submitted ~10 URLs on that property today. The dialog won’t say “quota” — older versions of GSC did, the current one doesn’t.
2. Burst throttling. Submitting 6–10 URLs back-to-back can trigger a soft block even below the daily cap, and reports of how long it lasts range from a few hours to several days. Pace matters as much as volume.
3. Account-level trust. This is the one nobody expects, and it deserves its own rule.
(There’s a fourth dialog that looks similar but isn’t: “Indexing request rejected” means Google’s live test found the page unindexable — noindex, robots.txt, a redirect, a 404, or a canonical pointing elsewhere. That’s a page problem, not a quota problem. Fix the page; resubmitting changes nothing.)
Rule 3: New Google accounts often can’t request indexing at all
Here’s the finding that cost us the most time.
We were running submissions from a recently registered Google account — verified owner of the property, everything green in Search Console. Every Request Indexing attempt returned “Something went wrong.” First submission of the day, first submission ever: refused. We suspected our automation, then the property, then the site.
Then we ran the control experiment: same browser, same property, same URL — but signed in with an aged, regularly-used Google account that we’d added as an owner. It worked on the first click.
The conclusion is hard to avoid: Google applies trust scoring at the account level, and fresh accounts can be locked out of Request Indexing entirely — silently, behind the same generic error it uses for quotas. A new Gmail managing a new property on a new website is precisely the profile that abuse systems are built to distrust. Nothing in the UI will ever tell you this is what’s happening.
The diagnostic is beautifully simple: submit one URL by hand. If manual submission fails the same way, stop debugging your tooling — it’s the account.
What actually fixes it
Based on the rules above, the playbook:
- Use an established Google account. Add it as an Owner in Search Console (Settings → Users and permissions), or better, have it verify the domain independently via DNS so it doesn’t depend on anyone else’s verification.
- Keep one account per browser profile. GSC routes tabs through the browser’s default Google session; with two accounts signed in, your requests can silently run under the wrong one. A dedicated Chrome profile removes the ambiguity.
- Respect the pace. Stay under ~10 requests per property per day, leave seconds — not milliseconds — between submissions, and stop at the first refusal rather than hammering through it. Hammering extends the block.
- Let the queue drain over days. A 36-article backlog at ~10/day is a 4-day job. That’s not a failure state; that’s the system working.
- Treat Request Indexing as a supplement. Your sitemap is the bulk discovery channel — keep it submitted and fresh in GSC. Request Indexing is for the URLs you care about today. (For Bing and Yandex, IndexNow does the same job with an actual public API — use it.)
Automating it without getting flagged
Everything above also explains how indexing automation should be built — and why most of it breaks:
- It has to run in your browser, on your GSC session, because that’s the only sanctioned path for regular content. Server-side tricks with the Indexing API are off-label and get cleaned up in waves.
- It has to recognize Google’s refusal dialogs and classify them: quota-style refusals should pause that property and retry tomorrow without burning retry budget; page-level rejections should be logged with the reason and skipped.
- It has to pace like a human — randomized delays, small batches, stop on refusal.
- And it should never guess URLs. Submitting a guessed permalink that 404s doesn’t just waste a request — it feeds Google a rejection signal on your property. The URL you submit must be the real, deployed permalink (from your CMS, or matched against your verified sitemap).
That’s exactly the design of BlogShoot’s Auto-Index extension: it pulls the queue of your published-but-unsubmitted articles, drives the Request Indexing flow in your own signed-in browser with human pacing, defers quota refusals automatically, and reports per-URL results back. Set it on a 24-hour schedule on any machine that stays on, and the backlog drains itself at whatever rate Google allows that day.
TL;DR
- “Oops! Something went wrong” = Google refusing your submission, not a bug in your site.
- There is no quota meter. ~10/property/day, less if Google doesn’t trust you.
- New Google accounts can be refused 100% of the time. Test manually; switch to an aged owner account.
- One Google account per browser profile, slow pacing, stop on first refusal.
- Sitemap for bulk discovery, Request Indexing for priorities, automation only if it behaves like a patient human.