An AI Search Case Study in Housing and Construction: Clicks on the #1 Result Fell 60%, and Only AI Traffic Raised the CV Rate 3.2×

Company	Ecoprocoat Co., Ltd.
Business	Ecoprocoat Co., Ltd. develops and installs floor coatings that use ultraviolet (UV) curing. Its main business, and one of the strongest track records in the category, is Aiken no Yuka — a floor coating for dogs, developed with veterinarians. Ordinary flooring is slippery. Aiken no Yuka adds grip, which takes pressure off an indoor dog’s joints and helps prevent injuries such as dislocations, so the home is safer for the dog.
Objective	Make paid ads and phone inquiries more efficient in floor coating Rankings were holding, but conversions had stalled — fix that
Services	AI Search Consulting (Queue Ltd)
Results	Work started in April 2026. Ecoprocoat is now cited in ChatGPT and AI Overviews for core queries such as “floor coating.” Over the last 90 days, AI traffic brought 260 sessions and 49 conversions. The key-event rate per session was 3.2× organic search for Ecoprocoat, and more than 9× for Aiken no Yuka. In August, branded searches rose 39% from the previous month.

What the work with Queue produced

Cost per acquisition keeps rising. Rankings have not slipped. And yet inquiries are the thing that fell. A lot of companies are in exactly this spot.

When those show up together, the cause is rarely how the ads or the SEO are being run — not now that AI search is widespread. The entrance to search itself has changed shape.

Queue Ltd started the AI search work with Ecoprocoat in April 2026. The company is now cited in ChatGPT and Google AI Overviews for queries such as “recommended floor coating companies,” and inquiries have gone up.

Here is what that looks like.

These figures are from the 90 days after Queue Ltd began the work.

Metric	From AI	Organic search (average)	Difference
Key-event rate per session (Ecoprocoat)	20.2%	6.4%	3.2×
Key-event rate per session (Aiken no Yuka)	9.4%	1.0%	More than 9×

In 90 days: 260 sessions and 49 key events. Once the company was showing up in AI search, the value of a single session moved by a factor of 3 to 9. The data confirms it.

The problem: you can hold the ranking and still lose the clicks

Start with this: it is not a problem particular to Ecoprocoat.

In a study Ahrefs published in July 2026, when an AI Overview is on the page, the click-through rate of the #1 result falls from an expected 24.1% to a measured 9.0%. That is a drop of 62.7%. The same study put the drop at 38% in December 2025. In six months, sites have started losing the clicks faster, even when the ranking holds.

The other direction shows up too. A Semrush study found that people who visit after reading an AI answer convert at about 4.4 times the rate of organic search. They have already done most of the comparison inside the answer. The higher rate is what you would expect.

Fewer people come in. The ones who do are better qualified. That is the shift underway right now.

Three symptoms already in the numbers (Ecoprocoat Co., Ltd.)

1. “Buying” inquiries got less efficient

We compared more than two years of paid-search data, from the start of the account to the most recent period.

Monthly acquisitions fell to about half, and cost per acquisition kept climbing. Spend was not cut, and the account was not being run carelessly. The same effort produced half as many inquiries. Phone inquiries moved the same way, and those are down sharply from the first half of 2025.

The team had already used the levers inside the ad account, and the decline did not stop. That was enough to judge that the cause was probably outside the account.

2. There was enough traffic. It was not turning into purchases.

This is the easy one to miss. Traffic itself was not short.

The dog columns on Aiken no Yuka alone have drawn 33,000 sessions since February 2026. They produced 39 key events, a conversion rate of 0.12%. Case-study and product pages on the same site convert at 8.2%. The gap is about 70 times.

The reason is clear. People who open a dog column want to understand how dogs behave. Floor coating is not even on the list of options they are considering. The site had been gathering a large audience with a different goal.

3. The data was already inside the company

One side finding: matching Search Console with the paid-search query report turned up several queries where the company was still paying for ads on terms it already ranked for. Most of them were searches for the company’s own name — branded queries.

This is not an AI search point. It is one example of what you find when you look across the whole picture: the data for the next move was already in the company. You cannot see that overlap from the ad account alone, or from the SEO data alone.

Why AI search optimization

Laid out together, the direction was fairly clear.

Stacking on still more new traffic fell down the list. Adding volume to a section that already has 33,000 sessions and converts at 0.12% does not fix the economics. What was needed was a path to people whose goal already matched the product.

AI search is that path.

People do not hand AI a keyword. They describe a goal. “My dog keeps slipping on the floor — which flooring do you recommend?” The model then names companies as a way to solve that. To be in the answer is to meet the user when the goal already lines up.

Readers of the dog columns did not act, and visitors from AI did. The explanation is the same: whether the goal and the means fit. That is why AI search was chosen as the next place to invest.

Why Queue

M, marketing lead

I’ll be honest. I started from “So what does LLMO actually do?”

I heard pitches from several firms. Queue was not the usual web-marketing conversation — “let’s do off-page SEO,” “let’s rethink the SEO from scratch,” “let’s publish as many articles as we can.” Their AI engineers were looking at how the model works, how it searches for information, and how it chooses what to cite, and building the plan from that. That stayed with me.

We do not have unlimited budget or people. It mattered that they would not try to do everything, and would work out with us where to start so the resources we do have would actually produce results.

Before the contract, the sales contact and the people who would do the work also explained the technical parts in language we could use. They knew the subject, and it was still easy to talk. I felt I could ask when I did not understand, and we could sort it out as we went.

What decided it was that Queue was thinking past “get mentioned in AI search,” through to inquiries and sales.

What we did first

An AI search analysis, to see where the company stood

The first step was to see how the company is treated in AI search today.

A normal SEO tool cannot measure that. Using Queue Ltd’s free AI search analysis, we looked at which queries cite the company, which ones do not, and who is cited instead.

You may be in a similar place:

• SEO is in place, but the numbers have stopped moving.
• You want a new marketing bet, and you do not know what to run.
• You are starting to doubt the return on ad spend.
• Traffic is fine, and you want a higher conversion rate.

If that is you, check whether your company actually shows up in AI search.

If even one of these fits, request a current-state analysis.

A free AI search diagnosis (Queue Ltd)

See, for free, how AI search treats your company.

Queue Ltd runs a free diagnosis of where you stand in AI search.

We send the AI 100 non-branded prompts written around the business, then show how often your company name and service name are cited, and how often they are mentioned. (The Ecoprocoat example.)

▶ Request a free AI search analysis

Queue Ltd: project lead (Einar Soderberg)

Ecoprocoat had already taken SEO and advertising seriously, and already met a lot of people in search. In AI search — ChatGPT and Google AI Overviews — the company was still barely present.

Individuals and companies now research and compare with AI, then decide from the answer. If the content and the record Ecoprocoat has built can be understood properly in AI search, and recommended to the people who need them, that will produce results. I was confident of that.

The Ecoprocoat team picks up the work quickly, and the communication is genuinely easy, so we keep improving it together.

The project is still underway. As the person on the account, I am glad the results are already showing up as traffic and conversions from AI search. From here, I want to widen the answers in which AI says “for floor coating, Ecoprocoat” and “for dog floor coating, Ecoprocoat.”