The finding, in one line
No meta description, a generic title and eleven h1 tags mean the lines AI reads first are empty or confused. These are quick, mechanical fixes, and they are holding back everything else.
The proof
These are the questions a brand or marketing lead would actually type when looking for an agency like Javelin. We then checked the directory lists AI assistants draw on to answer them, and recorded who was named.
Names that recur across the directory lists AI cites for Dublin agencies, from top pick to passing mention. Javelin did not appear in that set.
On the method, plainly. Reachability, schema, on-page text, titles and headings are measured directly from your homepage and robots.txt on 14 September 2026. The shortlist read is taken from the directory lists AI assistants cite for Dublin agencies, checked the same day. Because AI answers vary between runs, the firm standard is five runs per buyer question on each assistant before a rate is locked. That five-run pass is the immediate next step.
Three problems
None of these is about the quality of your work. They are about what the machines that now shape a buyer’s shortlist can see, understand, and repeat.
There is no meta description, and the title is a generic “Home - Javelin”. The two lines AI reads first are empty or unhelpful.
Eleven h1 tags on a single page blur what the page is actually about for a parser.
“Javelin” collides with the sport, the missile and the car, and the directory lists AI cites do not name you.
Why this is fixable
Nothing here is deep or expensive. A real title, a meta description and a single h1 repair the first thing every AI reads, and disambiguating the name does the rest. These are days of work, not months.
One honest note before the upside. Direct traffic from AI is still small today. The prize here is not the click. It is being on the shortlist at all, because when an assistant recommends you it has already told the buyer why you fit, so the introduction arrives pre-qualified.