We build Mythex, an AI app builder. In September we added 1,734 landing pages under /ai/: one for each pairing of a business type and an app type ("booking system for dental clinics", "dashboard for hotels"), all prerendered to static HTML with their own title, description, FAQ and breadcrumb schema.
A week later, Search Console said 289 pages indexed. Most of those were the 200 template pages and the docs. The 1,734 new pages barely showed up at all.
This post covers what was actually wrong, what fixed it, and what didn't change. The numbers are straight from Search Console.
The number that gave it away
Search Console has a Sitemaps report. Ours said:
- Sitemap:
https://mythex.ai/sitemap.xml - Last read: Sep 18
- Discovered URLs: 231
The live file had 1,961 URLs. The /ai pages had gone live after Google's last read, and Google does not re-read a sitemap on any schedule you control. Nothing was broken. Google simply hadn't looked since before the pages existed.
Most of the "why isn't it indexed" advice we found assumes a quality problem: thin content, duplicate pages, crawl budget. We could have spent a week rewriting pages that Google had never been told about.
Check the discovered-URL count on the sitemap before touching content. If it's lower than your live sitemap, nothing else matters yet.
What we did
- Resubmitted the sitemap in Search Console (Sep 20). Google re-read it the same day and reported 1,961 discovered URLs.
- Added IndexNow for Bing and the engines that share it. Our deploy pings it with the difference between the sitemap being served and the one just built. We snapshot the old sitemap before the CDN deploy, because afterwards the "live" sitemap is already the new one and every diff comes out empty. One catch: our sitemap has no
lastmod, so only new URLs get announced. Rewrites of existing pages have to be submitted by hand.
What happened
| When we checked | Indexed | Not indexed |
|---|---|---|
| Sep 20, before resubmitting | 289 | 103 |
| Sep 22 | 305 | 105 |
| Sep 24 (data through Sep 21) | 1,700 | 454 |
About 1,400 pages entered the index within a day of Google re-reading the sitemap. "Discovered – currently not indexed" went from 32 to 375, which means Google knows those pages and has queued them. That's the next bottleneck, and it's a different problem.
A few days later the new pages started getting impressions for queries nobody on our side had targeted by hand: "hotel dashboard builder" (position 4), "nutrition dashboard builder" (4), "boutique dashboard builder" (5), "law firm kpi dashboard builder" (10). By Sep 24, 157 different URLs had search impressions over the previous 28 days.
Two bugs we found along the way
Every page declared two FAQPage blocks. Our index.html shell carried a site-wide FAQPage in its JSON-LD @graph, pointing at /faq, and prerendering then injected each page's own FAQ on top. So /privacy was telling Google it was the FAQ page. We removed the shell copy and added a build check that fails if it comes back.
Three quarters of the template pages were orphans. "Related templates" returned the first four templates in the category, so all 48 portfolio pages linked to the same four portfolios, and the other 44 had no inbound internal link except from the category index. The prerendered HTML also printed no related links at all, so for any crawler that doesn't run JavaScript, every template page was a dead end. We now link them in a ring ordered by subcategory (four in, four out) and print it in the static HTML. A test asserts there are no orphans.
What did not change
Clicks. Non-brand clicks went from about 20 a month to 26. Indexing makes a page eligible to rank; it doesn't make it rank. Most of the new pages sit on page 2 or lower, with a handful of impressions each.
The honest read is that a young domain with one real backlink can get its pages indexed but can't push them to page 1. More pages won't fix that. The next lever is pages worth linking to, and people who link to them.
Checklist, if you are shipping a lot of pages at once
- After the deploy, open Sitemaps and compare "Discovered URLs" with your live file. Resubmit if they differ.
- Prerender real HTML. Google renders JavaScript eventually, but "eventually" is on its schedule, and other crawlers often don't render at all.
- Make sure each page is reachable through links in the static HTML, not only through the sitemap.
- One FAQPage per page. Check the raw HTML, not your component tree.
- If your sitemap has no
lastmod, IndexNow and sitemap diffs only announce new URLs. Rewrites need a manual push. - Track "Discovered – currently not indexed" week over week. If it climbs while Indexed stays flat, the problem is authority, not content.
Mythex builds and publishes web apps from a description, and these pages are how it describes itself to search engines. If you want to see what the generated pages look like, here is one.
Top comments (0)