A few weeks ago I thought I had a technical SEO problem.
Pages weren't showing up.
Search Console looked empty.
So I spent a lot of time worrying about indexing.
Then the pages finally got indexed.
Nothing happened.
That was the moment I realized indexing and understanding are completely different problems.
I'm building a small browser-side utility.
The product removes metadata from photos and documents.
From a technical perspective the site was fine.
Pages were crawlable.
Sitemap was submitted.
Search Console showed indexing progress.
But search traffic still barely moved.
At first I thought Google needed more time.
Then I started looking at the queries that actually appeared.
Something interesting showed up.
Users searched for things like:
"remove location from photo"
"remove gps from image"
"remove author from pdf"
Almost nobody searched for:
"metadata processing"
"metadata extraction"
"browser-side metadata cleanup"
Those were the phrases I had been using when describing the product.
The product and the user were talking about the same thing.
Just in completely different languages.
That changed how I think about SEO.
I used to assume:
Indexing
→ Ranking
→ Traffic
Now it feels more like:
Indexing
→ Understanding
→ Ranking
→ Traffic
And the understanding step is surprisingly slow.
Especially for small sites.
The weird part is that AI makes this easier to miss.
Building is cheaper than ever.
Adding another page takes minutes.
Adding another feature takes hours.
Adding another tool feels almost free.
So it's easy to create more things.
But search engines still need to understand what those things are.
Lately I've been spending less time building pages and more time trying to understand how users describe the problem in their own words.
That has probably taught me more than any SEO guide I've read so far.
Top comments (2)
This maps exactly to something I hit building my own app. There are actually two gaps stacked on top of each other.
The first is the vocabulary one you describe (people search "remove location from photo", not "metadata processing"). Slow and humbling to learn, fully agree.
The second is the "where" gap. Being indexed by Google is not the same as being found by the thing people now ask. ChatGPT leans on Bing, Claude on Brave, Gemini on Google, the open models on Common Crawl. I had Google indexing me fine and still got near zero pickup from the assistants people actually use to discover tools.
What moved the needle for me was matching the channel, not adding pages. An llms.txt plus an IndexNow push got Bing crawling within 24h, and I found out third party software directories (Slashdot, SourceForge auto-scraped my store listing) were feeding the AI citations more than my own site was.
Your "understanding step is slow" framing is the right one. I would just add that the understanding has to happen in the place your users are actually searching, and that place is fragmenting fast.
I like the distinction between being indexed and being understood.
What I've started noticing is that there's now a third layer as well: being discovered.
A page can be indexed.
A search engine can understand it.
And yet users may never encounter it because they're searching through a completely different channel.
The challenge keeps moving further away from publishing and closer to distribution.