Two Pages Made It Into the Index This Week

By Georg Malahov

On September 11 I submitted my sitemap — the list of addresses a search engine reads to learn what a site is made of — to Google and Bing. A week later there are two pages in the index out of two hundred and fifty-one, and not a single visit from search. The one stranger who did show up this week came from a client's site, not from search.

All week I was basically testing one question: can I do anything to make search pick the site up faster, or is waiting the only option. I tried three ways of speeding up indexing, and in parallel I put links to my site in the places where that is up to me.

My own console

Search Console is the panel where a site owner sees what Google knows about their pages. On Saturday the 13th it still said "processing data" instead of numbers, but it did surface my first mistake: only the default-language pages had made it into the sitemap, seventy-seven addresses out of three hundred and twenty-seven. Fixed.

On the 16th an email came saying Google had started counting impressions. The report showed two indexed pages out of two hundred and fifty-one known ones, plus five reasons the rest are not in. The number doesn't match the sitemap because Google counts staging and subdomains towards my domain as well.

Two hundred and forty-nine pages not in the index, with a list of five reasons

On the 17th I went through those reasons. The nastiest one was on an old site I built for the Petri Heil Schweinfurt angling club: its pages declared my former staging address as their main version, so Google filed them as duplicates. I fixed the addresses there and put a link to myself at the bottom of that site while I was at it. Every page in the sitemap now carries its publication date instead of today's date as its last-modified stamp, and I submitted some addresses for indexing by hand. All of that needed doing, but none of it made indexing any faster: the crawler re-reads what I fixed on its own schedule.

Somebody else's pages

Under my LinkedIn and GitHub profiles, search was still showing an old description — "Expert Full Stack Developer in Next.js" and a line about more than a decade of experience — even though I rewrote both pages a while ago. You can't request indexing for someone else's site from the console, but Google has a separate, fairly obscure panel: Refresh Outdated Content. The request takes the address as it appears in search plus one word that is in the description and gone from the page. I filed two requests, with "decade" and "freelance". Both were denied the same day: "Outdated content not in index".

Both requests denied the same day

I read the denial as good news at first, then opened search: both old descriptions were still there. On September 20, three days on, it still says "Fullstack developer with more than a decade of experience", while the GitHub profile description has said something else for a while now.

Google's own help explains the denial like this: the text you described isn't in Google's copy of the page, so the outdated content is already gone from the index and nothing needs doing. Search disagrees, and I can see two explanations, both from that same help page. First: the copy the tool checks and the description shown in search are different things. About a successful request it says the description is removed from search and then refreshed the next time the crawler visits — so the description is stored separately and moves at its own pace. Second: the address. I filed for de.linkedin.com, because that's how the profile shows up in German search, but the index may hold the page under a different address, in which case the tool was looking at the wrong record.

What the denial definitely does not mean is that the description in search got updated. One more thing: checking "through the crawler's eyes" by putting a Googlebot signature in your request proves very little. Large sites recognise the real crawler by the address it comes from, not by the signature, and what they hand it back may well not be what they hand me.

The link in a comment

On the 16th I left a few comments under other people's LinkedIn posts for the first time. In one of them, under an article by an author with forty thousand followers, I put a link to my post about how I work with agents. I put it there for the clicks: so that whoever reads the comment comes over and reads the post. Search wasn't on my mind at all at that point; the next day I decided to look at what had made it into the index.

The link turned out to carry a nofollow mark: Google generally doesn't count those in a page's favour. And the page I linked to was still sitting in the console as "discovered, currently not indexed". The comment link gave me neither weight nor speed.

What did work

Something else worked instead: not the weight of the links, but the pages they sit on.

The first one is the comment itself. Google indexed it within a day, along with my new profile headline, "AI Agents & Automation". My own profile sat lower in the same results, still with the old description.

The comment carries the new headline, my own profile the old one. The post author's name is blurred

And for "Ralph Loop" plus my last name, Google put together an AI Overview — a paragraph about who I am — and cited that same page with the comment as its source. The word "famously" in it is the model's own addition.

Google assembled an answer about me out of a comment under somebody else's post

The second one is a client's site. I built the Ukrainian Verein's site in July, and the contract says a link to me as the developer goes at the bottom. The link itself only went up on the 11th, the day I submitted both that site and mine to Google: before that it would have pointed at a page behind a password. On the 14th Google said it was indexing the Verein's site, and the same day my stats showed a visit from there. Someone found the Verein's site in search, saw the link at the bottom and came over to look: opened the pricing and the products. That's the first visitor I don't know, and they arrived before my own site showed up in search at all.

The week, summed up

I sped up exactly nothing this week: not my own indexing, not somebody else's descriptions, not weight through comments. What worked were the links I put up myself this week, in the places where it's my call: at the bottom of a client's site — the stranger followed it three days later — and in a comment under someone else's post with a large reach.

And here are the mistakes I found in my own setup this week. Next time I'll check for them before submitting a sitemap:

1. Sitemap in the default language only. Seventy-seven addresses instead of three hundred and twenty-seven — the translations never made it in.
2. The same last-modified date on every page. Every page in the sitemap looked freshly updated, old posts included.
3. noreferrer on the link at the bottom of a client's site. The attribute stops the browser passing on which page a visitor came from, so some visits may have reached my stats without a source.
4. Forgotten main-version addresses on the old client site. They pointed at my former staging, and duplicates turned up in Google's report.

What comes next

External traffic is something I never had, so the next steps are about links and queries. On the evening of the 19th I updated the Call Copilot listing in the Chrome extension store and put links to this site into it; it's with Google for review right now. That was the last place carrying a link to me that is mine to decide. After that: work out which multi-word queries people use to look for what my products do, and write for those.

I'll know search has kicked in when Search Console starts showing the first visits from search to my site and to the product pages.