How a US apparel manufacturer recovered its Google traffic — and opened an AI-search lead channel

The short version

A US clothing manufacturer was watching its free Google traffic disappear. It wasn’t a content problem — the site’s own settings had quietly told Google to ignore most of its pages, and the pages that stayed contradicted each other. In one fixed-scope project we found the cause and handed the client’s team a page-by-page fix plan. Within about two months the slide reversed: organic search recovered — daily clicks roughly doubled off the low — the site now ranks at the top of Google for competitive wholesale terms, above the global brand it sources, and a brand-new AI-assistant channel opened and is growing, converting visitors to enquiries better than any other source.

Every number below comes from the client’s Google Search Console and Google Analytics 4.

What “the floor” actually looked like

The owner came to us because traffic was falling and nobody could explain it. Here’s what we found:

• The site had slipped to the back of Google. Its average position across search fell from about 20 to about 29 — page 3 territory, where almost nobody clicks.
• Free search traffic was falling fast. Organic visits in Analytics dropped sharply over the spring.
• Most pages were missing from Google — because the site was blocking itself. Google’s own coverage report showed hundreds of pages it had given up on or been told to skip: 113 it crawled but chose not to index, 217 blocked in the site’s robots.txt file, 55 carrying a “do not index” tag — and 131 pages that were indexed despite being blocked, a dead giveaway of settings fighting each other.
• The pages contradicted themselves. The same minimum order quantity appeared as 25, 50, and 150–200 in different places; positioning was vague (“ultimate sourcing partner”); and dozens of near-identical location pages competed with each other.

In plain terms: Google was getting mixed signals about which pages to trust, so it stopped showing them.

What we did

This was a strategy-and-plan engagement — we diagnosed the problem and wrote the instructions; the client’s own team made the changes. We delivered:

1. A page-by-page verdict — keep, fix, or remove — so the cleanup had a clear map instead of guesswork.
2. A keyword map of what buyers actually search, grouped by real commercial intent.
3. A clean-up list of which URLs to delete, redirect, or keep, to clear the clutter Google was choking on.
4. Rewrite briefs for the pages that make money — homepage, the Custom and Wholesale sections, About, Contact — with clearer headlines and real calls to action.
5. Behind-the-scenes markup (structured data) that spells out, in code Google reads, what the company is and does.
6. An “entity model” — a plain map of who the company is, what it makes, and where it operates — written specifically so AI assistants like ChatGPT can understand and recommend it.

What happened

Weekly organic clicks (Google Search Console). Engagement began mid-April; clicks bottomed out in late May and recovered through July.

The site climbed back out of the hole.

• Average Google position: ~29 → ~19. Off page 3, back onto page 2 and the edge of page 1 (GSC).
• Daily organic clicks recovered from a low of ~25/day to ~50–60/day — roughly back to pre-collapse levels (GSC).
• Pages indexed by Google grew from 255 to 333 (+31%) as the clean-up cleared the blockages and the right pages were let back in (GSC).

The slide didn’t just stop — it reversed: clicks climbed back off the late-May low and held through July.

And a brand-new channel opened: AI search.

Once the markup and entity model went live, the site began appearing inside answers from AI assistants (ChatGPT, Perplexity and similar). In Analytics this shows up as a distinct AI Assistant source. For the first five months of the period it produced zero enquiries — the channel simply didn’t exist for this site. After the entity and structured-data work, it switched on and started growing: 21 form submissions from a source that had none before.

Weekly form submissions from the AI-assistant channel (Google Analytics 4): zero for five months, then it switched on and climbed. Weekly split is indicative; the 21-lead total is exact.

And these visitors convert better than anyone else:

• AI-assistant visitors submitted a form 7.5% of the time — the highest rate of any channel, ahead of regular Google search (3.5%) and people typing the site in directly (1.7%).

These are real form submissions tracked in Analytics. The AI channel is still small in volume, but it punches far above its weight on actual enquiries — and most competitors in this space aren’t in those answers at all.

One commercial page became a genuine commercial asset.

A single wholesale category page pulled ~1,850 clicks from ~57,000 views in six months (GSC), and today it ranks at the top of Google for its main wholesale search term — above the global brand it sources and above the big B2B directories — proof that a cleaned-up site plus the right brief wins money terms, not just the company’s own name.

Live Google result for the client’s core wholesale term, reconstructed with the client’s name withheld. A Google AI Overview also appears on this query.

Why it worked

A traffic collapse like this isn’t fixed by writing more content. It’s fixed by clearing up how search engines — and now AI assistants — understand the site: stop blocking your own pages, stop contradicting yourself, and state plainly who you are. One fixed-scope project reversed the slide and, as a bonus, opened a channel the client didn’t have before.

This is what our AI search optimization and SEO work is built to do. Request a growth audit to see where your site stands.