Shown 415,000 times. Clicked under 1%. Invisible to AI. Here is what we did about it.

The starting point

This was not a struggling business. It had hundreds of products, a large catalog of reviews, and a genuine reputation in its category. What it did not have was a digital presence built so that search and AI engines could understand and recommend it. Service pages were thin or missing. There was no validated schema. The core product collection was shown twenty thousand times in ninety days yet sat at position seventeen, off page one, where almost nobody scrolls.

And underneath the Google problem was the bigger one. Search is splitting in two: the old blue-links world, and the new one where a customer asks an AI engine "where do I buy this" and gets one recommended answer. In that world the winner is not whoever has the most pages, it is whoever's content is written so an engine can lift a clean factual sentence and quote it. This company was absent from that entirely.

What got built

The data chose the priorities. We let the company's own Search Console numbers select the first pages to rebuild, the ones already shown thousands of times but ranking too low to capture the traffic.

• A priority page rebuild, driven by real dataThe top traffic pages rewritten from scratch from the company's real product facts, each opening with a direct factual sentence built to be the line an AI engine quotes. The highest-impression collection led the list, because pulling it onto page one was the single highest-value move on the site.
• Validated schema across the siteOrganization, Service, FAQ, and Product schema, built and validated, ready to deploy at the site's relaunch so a theme change would not wipe it.
• A true FAQ page, not contact-page blocksThirty-eight questions written in direct-answer format for featured-snippet and AI capture, with its own FAQPage schema, replacing a handful of blocks that had been buried on the contact page.
• Three service pages with real specificsDedicated pages for each core service, written from the company's actual pricing and turnaround, the kind of concrete detail both customers and AI engines reward.
• A directory and citation campaignConsistent name, address, and phone submitted across roughly twenty directories, led by Foursquare, which feeds a large share of ChatGPT's local business data, plus Bing, Apple, Crunchbase, and the major trade and business listings.
• Sitewide meta and a clean handoffUnique titles and descriptions for every page, plus a plain-language handoff sheet so the company's developers could apply the theme-level pieces correctly at relaunch.

The judgment calls that mattered most

The most valuable part of this engagement was not a deliverable. It was what we chose not to do.

The client was interested in modeling new pages on a competitor's high-ranking content. We declined to lift it, and explained why: copying fails twice over. Legally it is the competitor's material, but more importantly, AI search actively buries duplicate content because it is looking for the original source to credit. Copying would not just be risky, it would make the site invisible to the exact engines we were trying to win. So we did the harder thing, original copy from the company's own facts, and those pages will outperform anything borrowed.

The client also wanted a competitive piece that named and called out rivals. We built the piece, but framed it around methods rather than companies, because a page that attacks a named competitor is the one thing that can get content filtered by AI engines, flagged by directories, and into legal trouble. A page that confidently explains why your method wins reads as the market leader talking, more persuasive and far safer.

And when the client's same-day turnaround policy turned out to apply to one product line but not another, we left the claim off the page where it was not true, rather than paste it everywhere for a tidier result. Advertising a same-day service the shop floor might not run is how you get angry customers and chargebacks. Protecting a client from a promise they cannot keep is worth more than a clean-looking page.

The takeaway

This company already had the hardest thing to earn: visibility. Google was showing it constantly. The work was converting that visibility into clicks, and building the structured, original, consistent foundation that lets AI engines recognize and recommend a business by name. That foundation is in place and ready to compound as the site relaunches and the citations finish publishing. The honest framing throughout, original over copied, methods over attacks, true claims over convenient ones, is not a constraint on the work. It is what makes the work hold up.

If you run a product or retail business, the same dynamics apply to you. We go deeper into how shops and brands get surfaced in AI shopping answers on our AI visibility for retail page.