Lullwood：十個已收錄頁面和一條外部連結帶來的啟示

Lullwood, a free browser horror game, had ten pages indexed by Google, yet Search Console gave us no query data to see our name in. The 14 impressions it did report cannot be tied to any query. Crawling, speed and schema were all fine. Google's URL Inspection tool lists exactly one outside page linking to us, a post we published ourselves.

That is the whole diagnosis. If your pages are indexed but get almost no impressions, the first thing to count is who links to you, not how well Google can read you. Below is the evidence, the API behind each number, and what we do not know. We make Lullwood, so this is our own site's data.

What the SEO checklist covered

We did what a search checklist says, and measured each item on 2026-10-28:

A sitemap submitted to Search Console. It went from 11 URLs to 18 in one release, and it listed 18 when we measured on 2026-10-28.
IndexNow pings that run on their own after a deploy that changes a page. The run after that release returned HTTP 200 for all 18 URLs.
Valid WebSite, Organization, VideoGame and FAQPage JSON-LD on the home page. All of it parses.
A 0.121 second time to first byte on the home page.
HTTP 200 for nine search and AI crawlers (GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-SearchBot, PerplexityBot, Googlebot, Bingbot, Applebot) on /, /robots.txt, /llms.txt and /sitemap.xml.
One H1 per page and self-referential absolute canonicals.

What that bought us

In the 28 days from 2026-09-28 to 2026-10-26, the Search Console API reported 1 click, 14 impressions and an average position of 2.7. The home page had 13 of those impressions and the one click.

Do not read 2.7 as page one. It is an average over 14 impressions on queries Google will not name. The same API returned no query rows at all, so there is nothing to filter for our name and nothing to confirm those 14 were for it.

The URL Inspection API said 10 of the 11 sitemap URLs at that time were "Submitted and indexed". The exception was a form page, /suggest.

Why indexed pages can get almost no impressions

Across everything we inspected, the referringUrls field listed four distinct pages linking to ours. Three are our own: the home page, /credits and /devlog. The fourth is a dev.to cross-post of one of our devlog posts, published 2026-10-27. So the external set is one URL, and we made it. That list is what Google reports for the pages we inspected, not a full backlink audit.

The page ranking for our name makes it worse. When we searched for Lullwood on 2026-10-22 and 2026-10-26, with a web-search tool and not a logged-in Google results page, our own GitHub repository was the result that ranked, first on the 26th. We fetched that repo page on 2026-10-28: its website link to www.lullwoodgame.com carries rel="nofollow". The repo does not appear in our referring URLs at all.

The one place we could make a followed link ourselves was the body of a dev.to article. The links in our published one had no nofollow. A GitHub README, an itch.io game page and the Three.js Discourse all mark outbound links nofollow when we fetched an example of each.

Crawlability and authority are different problems

Crawlability is whether a search engine can fetch and understand a page. Authority is whether anyone else vouches for it. A checklist fixes the first, because every item on it is something you control from your own repo. The second needs people you do not control to link to you, so it is not on the list and cannot be shipped in a release.

The rule we took from our numbers: when Search Console says indexed and the impressions are close to nothing, stop editing the site and start counting referring URLs.

What we are doing next

As of 2026-10-28 the number of distinct external pages linking to Lullwood is still one. This is a diagnosis, not a result.

More cross-posts are scheduled, with the canonical set to the page on our site. We also plan to ask curated lists of browser and AI-made games to include Lullwood. Those lists mark links nofollow too, so we expect them to help readers and answer engines more than rankings. We will report the referring-URL count again, whatever it is. If it rises and impressions for our name stay near nothing, authority is lagging and we keep going. If it does not rise, the one link we have was luck and this diagnosis needs revising.

The original, with the source of every number, is on the Lullwood devlog. If you want to see the site all this is about, you can play Lullwood free in your browser, or read how to survive Lullwood.

We are one human founder plus a team of Claude agents building Lullwood in the open.