Running a network of indie sites alone: keywords, launch, indexing and GEO — methods and trade-offs

Hugozen · 2026-11-02

1. Scope

This post is about methods and trade-offs, not results. I'm Hugozen. I do product, engineering and operations as one person, and I also run a dozen-plus personal indie sites for overseas readers on my own. One person can't carry many sites, so I try to push the "is this worth building?" question as early as possible: a little more time on keyword choice saves a lot of rework later.

Four parts: keyword selection, launch, indexing, GEO. For each I'll say what I do and why I don't do more. Where I haven't verified something, I'll say so.

First, one of my own failures, because many of the trade-offs below come from it: I have a content site that put up a large number of AI-generated pages in a short time and then lost a lot of Google indexing. It still gets by on Bing traffic and visibility in GEO / AI search, and it's slowly recovering. The lesson: don't flood a site with AI content pages. Go for a few carefully edited, high-quality pages first, then scale up gradually. Nearly every newcomer learns this the hard way. Read everything below with that in mind.

2. Keywords: pick smaller, and pick something you can win

The most common way an indie site dies isn't poor execution; it's choosing a keyword that's too big. A new domain with no backlinks and no history can't take on an established site over a head term.

My approach has two steps.

Step one: find low-competition long-tail needs. Long-tail queries have low volume but specific intent, and one page can fully answer them. For one person, "one query, one page, answer that exact problem well" is the best fit.

Step two: screen with KGR. KGR (Keyword Golden Ratio) is the number of results returned by an allintitle: search for the phrase, divided by the phrase's monthly search volume. The usual rule of thumb: when monthly volume is modest (commonly cited as up to roughly 250), a ratio under 0.25 suggests weak competition, 0.25 to 1 is middling, and above 1 you should skip it.

KGR has real limits:

It's a rule of thumb, not a law. Volume tools give estimates, and different tools can disagree widely.
It only measures title competition, not search intent. If the first page is full of big sites' tool pages or encyclopedias, a low ratio doesn't help you.
Low competition can simply mean nobody needs it. So KGR is a sieve, not a decision.

After the screen I do one more thing: search it myself and look at who ranks on page one. Forum threads, stale pages and mediocre small sites mean there's a gap; authoritative sites and official pages mean walk away. Then ask: can I actually solve this need with one page or one small tool? If not, I skip it.

One more trade-off: I don't try to pick many keywords at once. One site that goes deep on one clear need is easier for a search engine to understand than ten shallow pages across ten queries.

3. Launch: keep it as simple as possible

Most of my sites are static content hosted on Cloudflare Pages. The reasons are plain:

No server to maintain; it only serves files.
A global CDN, so overseas readers get fast loads.
DNS, HTTPS and hosting in one place, so problems are easier to trace.
Easy rollback: every deploy is a version, and a bad one can be reverted.

Before launch I only do what's necessary:

Each page has a unique title and description, and a canonical URL pointing to itself.
Generate sitemap.xml and reference it in robots.txt.
Pick one hostname (with or without www) and 301-redirect the other.
If there's analytics, load the script only after consent for visitors in the EEA, UK and Switzerland (Consent Mode). That isn't SEO, but doing it at launch saves rework.

Sites with payments, accounts or uploads need a much higher bar. This post covers static content and tool sites only.

4. Indexing: don't just wait, and don't pester

A new site isn't known to search engines the moment it goes live. I work in three layers.

Google Search Console. Verify the whole domain by DNS (a Domain property), then submit the sitemap. For important pages you can use URL Inspection to request indexing. That request has a quota and no guarantee, so don't click through hundreds of pages one by one.

Bing and IndexNow. Submit the sitemap in Bing Webmaster Tools too. IndexNow is an open protocol that notifies participating search engines when a page is added or updated. To be clear: Bing and Yandex participate; Google does not. So it doesn't replace Search Console; it's one more channel.

Internal links and outside signals. The most reliable indexing isn't submission but discovery: every page has at least one link from another page on the site, and one or two real references exist elsewhere. Submission only speeds things up.

When indexing is slow, I check these before deciding to keep waiting: a stray noindex, a canonical pointing the wrong way, thin content, a wrong URL in the sitemap. Most of the time the problem is on my side.

How long indexing takes, I won't promise: it varies a lot from site to site, and between pages on the same site. My experience is that slow indexing is usually not a missing submission but a search engine that doesn't yet think you're worth crawling more. The right response is to make the existing pages better, not to pile on new ones. The site that lost its Google indexing is the counter-example: the faster the pages went up, the more the search engine saw volume and the less it saw quality. That it survives on Bing and AI-search visibility suggests those two channels are worth treating as supplements to Google, but they don't replace healthy Google indexing.

5. GEO: how far to go on AI-search visibility

GEO (Generative Engine Optimization) means making it easy for AI search and assistants to read your pages and be willing to cite them. The field is new and there are no agreed rules. What follows are things I find reasonable and cheap, not guaranteed tricks.

Answer first. Open with two or three sentences that answer the question in the title, then expand. This helps readers and machines.
Clear structure. Descriptive headings, lists, short paragraphs; one page, one subject.
Checkable facts. Say who wrote it, when, and on what basis. Sources that are clear are easier to cite.
Allow crawling. Don't accidentally block every AI crawler in robots.txt; if you choose to allow them, say so.
Structured data. Use JSON-LD for the article, the author and FAQs. It's a low-cost addition for both classic search and AI.
llms.txt. A proposed convention: a short, model-friendly summary at the site root. I haven't seen reliable evidence that it changes citations, so I treat it as a "while I'm there" item, not a priority.

My stance: these are what good content looks like anyway, so I do them. But I won't change the content itself just to be cited by an AI, and I won't claim any of this works for certain.

6. A few general principles

These practices are common in the overseas indie-site community (for example in the sharing from the Gefei circle). Below are the general principles I took away, restated in my own words. They aren't quotations, and I haven't verified every one of them myself:

Favour new terms and low-competition openings. The earlier you get into a new term, the better your chances; chasing news-driven queries that go cold in a few days is poor value for one person. Prefer needs that people will still search for next year.
Judge supply and front-page quality separately. Few pages with the phrase in the title means thin supply; but if the front page is all big sites' inner pages, the bar may be lower or higher than it looks. Keep the two dimensions apart.
Inner pages on the front page can mean a homepage opportunity. If the top results are mostly inner pages from big sites and nobody has built a dedicated page, a purpose-built site homepage can break in; if several dedicated sites already rank, don't try to squeeze in.
One query, one page; long tails get their own pages. Different phrasings of the same idea (word order, synonyms) are often separate leaderboards. Answering the phrasing nobody covers well beats covering everything vaguely.
Ask three questions before building. What is the user trying to get done? Where do current results fall short? Why would my page satisfy it? If the third has no answer, skip it.
Be patient about indexing. Don't expect results from one submission, and don't pester search engines with hundreds of manual requests.
Bing and GEO are supplements. Pages that answer the questions people ask AI, with clear structure and clear sources. This line doesn't depend on Google, but it doesn't replace Google either.
Quality first, then volume. That's the one I paid for with a failure, and it's the premise behind all the others.

7. Summary of trade-offs

Pick smaller keywords. KGR is a sieve, not a verdict.
Launch on the simplest hosting and spend the effort on content.
Index with sitemap + Search Console + internal links; IndexNow is a supplement. Patience beats submission count.
Quality first, volume later: don't ship a flood of AI-generated pages.
Do the low-cost parts of GEO that also help readers, and don't oversell.
At every step ask: can one person maintain this long-term? If not, don't do it.

Next, how these sites make money, again methods and trade-offs only: Revenue paths for indie sites. There's no public contact address on this site; to follow new posts, subscribe to the RSS feed.