How a UK interior designer got cited by AI without filming a single TikTok

Reading time: 9 minutes

Last verified: 11 August 2026

She never filmed a video, never opened the app everyone told her she needed, and her trust-signal score climbed anyway.

The client here is a UK interior designer in high-end residential. Her buyers are 45 to 65, they spend serious money on their own homes, and they take weeks to decide. When we started, her score on our 19-signal trust framework was 8 out of 19. Those scores are our own records. She did the operational, sales and delivery work, Ambitions AI rebuilt the marketing layer underneath, and one business's result is not a promise to yours.

She has asked not to be named, so there are no revenue figures, project values or fees here. What is left is the part you can copy: what was broken, what we fixed, and in what order.

→ The fix was unfashionable and unglamorous. It was also finite, which is the point.

This article covers her starting position, the four changes we made, the buyer journey behind them, and the comment-to-call path she ran afterwards. By the end you will know whether the sequence has legs in your category, and which part to run first.

What was the designer's starting position?

Quick Answer: An elegant website with no named author, no service-level schema, a booking funnel nobody was tracking, and no process for asking happy clients to review her. Her score on our 19-signal trust framework was 8 out of 19.

She had done plenty right. The portfolio was strong and she was clear about the client she wanted. What was missing sat underneath the aesthetic, where buyers and AI answer engines look.

• No named author on any service or project page. The site read as though a brand had built itself.
• No service-level schema, just basic LocalBusiness tags.
• A booking funnel with no tracking. Visitors landed, clicked, dropped, and she could not see where.
• No review process. Her clients loved her. Nobody was asking them to say so in public.

None of that is dramatic, and all of it is cheap to close. The 19 signals are the Ambitions scoring framework, built from editorial patterns we observe in sources that AI answers cite. Nobody outside the labs knows what a language model checks, so we score what is observable.

Why did skipping TikTok not hurt her?

Quick Answer: Because her buyers were not choosing an interior designer on TikTok. They were searching, comparing and asking AI assistants for shortlists, then arriving at her site to check whether a real person stood behind the work.

I am not going to hand you a demographic statistic to settle this, because the honest evidence is her own account of where her enquiries came from: search, the guide download, and referrals that ended with someone checking her website first. The people commissioning full-house projects were reading her pages.

That is the test for any channel. Not "is this platform big", but "is this where my buyer decides".

There is a search-side argument too. Ahrefs analysed 1.9 million citations across 1 million Google AI Overviews in July 2025 and found 76.1% of cited pages already rank in Google's top 10. That covers Google AI Overviews specifically, not ChatGPT or Perplexity. It still tells you something useful: the page the answer layer quotes is usually one already doing the classic search job well.

What did we actually fix?

Quick Answer: We named her on every page, added Service and Article schema across her service and project pages, shortened the booking funnel and added tracking, then switched on review automation. Four jobs, in that order.

First, authorship. She became the named author on every service page. Photo, biography, her qualifications, and a short paragraph on why she cared about that category of work. Project case studies carried a named senior designer, so who did what was obvious to a reader and a machine.

Google's E-E-A-T framework, for the record, is the guidance its human quality raters use when grading search quality. It is not a ranking factor, and AI engines cannot be said to apply it. The reason to name a person is simpler: a buyer handing over the keys to their house wants to know who is doing the work.

Then schema. Service schema on each design category, Article schema on the project case studies, FAQPage schema on the top three landing pages. Google's Rich Results Test went from "no items detected" to four valid items on every page we touched.

Then the funnel. We took a step out of the booking flow and added tracking, so she could finally see where people left. The argument for a shorter path is not a statistic, it is mechanics: every extra step is another chance to leave, and a path you cannot describe is a path a stranger cannot follow.

Then reviews. Every completed project now triggers a review request one week after handover. Whitespark's 2026 Local Search Ranking Factors puts review signals at roughly 19 to 20% of local pack weighting, and in that survey review quantity ranks above recency. Google's own local documentation says more reviews and positive ratings can help local ranking, and says nothing about velocity. Our observed practice, labelled as exactly that, is that a live feed reads as a stronger current signal to a buyer than three-year-old reviews. A judgement, not a ranking claim.

Her trust-signal score moved from 8 out of 19 to 12, and then to 16. Those are entries in our scorecard, not an independently audited outcome, and they are one business's result rather than a typical one.

What does the buyer journey actually look like?

Quick Answer: Google search, an AI Overview naming a shortlist, a click through to her website, a check of recent reviews, then a consultation booking. Five stages, and her site had to hold up at three of them.

If you take one thing from this article, take this sequence. Most owners never map it, and every fix above hangs off it.

1. Search. Someone types a question about a room, a style, a budget approach or a local designer. They are not searching a brand name yet.
2. The AI answer. An AI Overview or an assistant returns a summary, often naming a few businesses or quoting a page. If your pages are not the sort of thing that layer quotes, you are invisible at the moment the shortlist is drawn up.
3. The website. The buyer clicks through, checking one thing: is there a real, qualified human here. A named author with a biography answers that immediately. An anonymous brand voice does not.
4. Reviews. They open another tab and look at what recent clients said. Not 2023. Recent.
5. The consultation. Only now do they book, and the shorter the path to that screen, the fewer places there are to lose them.

Each fix maps onto a stage. Schema and editorial quality feed stage two, named authorship stage three, review automation stage four, the funnel work stage five. None of that is a content-volume problem, which is why the channel everyone wanted her on was never the answer.

How did the comment-to-call path work for her?

Quick Answer: Five steps. A buyer comments a keyword, ManyChat sends the guide by DM, they click to a landing page, enter an email, and the follow-up email carries a calendar link.

We installed the same comment-to-call engine we run on our own channels. The mechanic is not clever, it removes friction.

1. Comment. A buyer comments a keyword on one of her posts.
2. DM. ManyChat detects the keyword and sends the download link.
3. Click. The buyer clicks through to a landing page.
4. Email. They enter an email address to receive the guide.
5. Calendar. The follow-up email includes a link to book a consultation.

A buyer who downloads a guide is under no obligation to book that day, and on a considered purchase most of the value sits in the follow-up rather than the first click. That is why owners switch this engine off too early, and why it repays the ones who leave it running.

Comment SIGNAL on any Ambitions AI post on Instagram or LinkedIn and ManyChat sends the 19 Trust Signals briefing to your DMs.

Why does the order matter more than the effort?

Quick Answer: Because each fix only pays once the one before it is in place. A shortened funnel cannot rescue a page a buyer does not trust, and schema cannot make an anonymous brand look human.

Name the human first, because that is the question a buyer asks fastest. Mark up the pages next, because structured data only helps once there is something worth marking up. Shorten the path to the booking screen after that, so the people you have convinced can act. Then keep reviews arriving, because a page that was credible last year has to stay credible this year.

The pace of that sequence depends on your category, not on how hard you work. Where competitors already name their people, carry structured data and run review automation, the same four jobs take you to parity rather than ahead, and the next gain costs more. I would rather say that now than sell you an expectation that belongs to someone else's market.

Liyana's insight

This is the client who made me put my full weight behind trust-signal work for service businesses. She chased nothing. No trend, no new platform, no aggressive ad spend. She made her website credible to the people and the machines doing the qualifying, asked her happy clients to say so in public, took a step out of her funnel, then left it alone long enough to work. The owners who win here are not the ones with the most time for content. They are the ones willing to do four boring jobs properly, then stop fiddling.

Will the same approach work in my category?

Quick Answer: It works in any service category where buyers search before they buy. The specifics shift, but the order, named authors, then schema, then funnel, then reviews, does not.

Run the 5-step audit on yourself this week and score honestly. A low score is a roadmap, not an embarrassment: this designer started at 8 out of 19, and the US luxury bed retailer we work with started at 4 out of 19. Score yourself, take the lowest signal in the list, and fix that one before you touch anything else.

The exception is a category where buyers never search. If your work is entirely referral-led, start with the buyer-question habit and reviews.

FAQ

Why will you not name this client?

She asked us not to. She is comfortable with the method being published, not with her business identified alongside it, and that is her call. Ambitions AI publishes named client results only when the client agrees.

Her score moved, so how do I know that came from this work?

You do not, and neither do I, with certainty. The scores are our records, they are not independently audited, and she ran the consultations, quoted the work and delivered the projects. What we can point at is the score, 8 to 12 to 16 out of 19, and the authorship, schema, funnel and review changes behind it. Treat the rest as context, not proof.

Could she have done this herself?

She could have done the build herself, given the time. She did not have it, and most owners do not. If you do, the sequence in this article is the sequence, and the Rich Results Test is free. We run the same order for US businesses, including the US plumbing company we work with.

Sources

• Ahrefs, July 2025, 1.9 million citations across 1 million AI Overviews
• Whitespark 2026 Local Search Ranking Factors
• Google on reviews and local ranking
• Google Search Quality Rater Guidelines
• Google Rich Results Test