4,191 products. Ten in Google.
Friends of mine in Lithuania run a site — airbag.lt. Parts, repair, store. It's been up for 22 years — the domain was registered back in 2003.
I'd been wanting to look under the hood for a while. Not as an auditor — just out of curiosity. We do roughly similar things, I'd been talking about neural nets, and I wanted to test what my agents would show on a real, familiar business.
I launched 4 agents in parallel. Exactly the way I described in the previous post. Each one on its own slice: — one on the technical side (speed, stack, headers) — a second on SEO (title, meta, schema, hreflang) — a third on indexing (sitemap, robots, Wayback, search cache) — a fourth specifically on the store (shop on a subdomain)
About two hours of work. I mostly read what the agents were finding.
What surfaced about the store
In the shop.airbag.lt sitemap — 4,191 URLs. Of those: 1,535 products, 1,551 OEM, 1,080 tags, 17 categories.
The sitemap was last updated yesterday. Meaning the catalog is alive, products are being added, money is being put into the store.
In Google's index — about ten pages. For the entire site. Including the main one.
Product schema on 1,500+ products — zero. Mobile performance for the store — 22 out of 100. LCP 6 seconds. Nobody waits six seconds to look at a car part.
For Google, this store effectively doesn't exist.
Why it can't see it
The main site's robots.txt has the directive: Crawl-delay: 10. Translation: Google can hit one page every ten seconds. With 4,191 products, a full crawl takes 11+ hours. Googlebot doesn't finish that. It came in, looked at a dozen pages, left.
It's like having a warehouse with 4 thousand SKUs and a door that lets one customer through every ten seconds.
The main site
The store isn't visible from outside. But the main site doesn't help either.
— jQuery version 1.9.1. That's 2013. — 8 render-blocking resources on load. Foundation, select2, nivo slider — all from the same era. — 0 security headers. — Title of the homepage — airbag.lt. Just the domain. — Meta description — airbag.lt. — H1 — airbag.lt. — H2 — none. Schema — none. OG tags — none. Canonical — none. Hreflang — none.
The site advertises three languages — Lithuanian, English, Russian. /en and /ru redirect to the homepage. There's no multilingual support. There's the illusion of one.
Analytics
The site has Google Analytics installed. Universal Analytics. Google itself shut that down in July 2024.
Which means my friends have spent a year and a half not knowing what's happening on their site. No traffic, no behavior, no conversions. A black box.
Any decisions — by feel.
It's not their fault. Nobody told them.
Wayback
I checked the archive over 10 years. The main site has barely changed — once a year, cosmetic edits. Development stopped around 2020. Meanwhile, the store keeps living and growing.
Meaning money is being put into the catalog. Into anyone actually finding that catalog — no.
What I did about it
I showed them the report. The rest is up to them. I'm not selling anything, these are friends. If they want it fixed, I'll tell them how. If not, that's fine too.
But for me, the point was different.
The takeaway isn't about them
I went in to test the agents on a familiar business. I expected to see minor mistakes. I saw a picture in which the business exists, basically, only for insiders — for people who already know where to drive and who to call.
And it's not a rare case. I've run other friends' sites the same way — it's almost always the same. Sometimes a bit better.
This is the normal reality of small and medium business on the internet. The site was built once, long ago. Launched. Forgotten. They add products to the storefront, but don't touch what's on the outside.
For people who know how to see this, that's exactly the field of work. Not hunting for rare niches. Just look around — at friends, at locals, at people next door.
Try it yourself
If you don't believe the numbers — check. The site is open: airbag.lt and shop.airbag.lt. No logins needed. Everything I found was gathered from outside — public data, what any browser sees.
Run it through your own tools. Through agents. Through Lighthouse. Through whatever you have.
Look for yourself: — how many URLs are in the store sitemap — how many pages are in the index for site:airbag.lt — what's in robots.txt — what title, description, H1 the homepage has — the library versions in the source
If they match mine — the tools work. If you find more — tell me. I'm curious.
But mostly — try the same thing on your friends' sites. On the business next door. On the store where you go for bread. The picture is almost always roughly the same. It's just that usually nobody sees it.
In the next post, I'll show how this gets fixed.