From 7th to 1st in ChatGPT in 3 Months: How LAGE Lightings Grew Sessions 233% and Won Inquiries From Top Architects

What was the starting point?

LAGE Lightings sells luxury lighting and home décor to buyers in Nigeria through an online store. Its co-founder, Balikis Oke, describes the situation plainly: “I was surprised we could beat our competitors since we had tried for years to do it.” In July 2026, when I asked ChatGPT about luxury chandeliers in Lagos, the brand appeared seventh. For a business selling high-value items to architects, developers and hotels, seventh on a list means the first calls go elsewhere.

This kind of purchase is considered and often involves a professional intermediary. A homeowner may ask an assistant, but an architect or a hotel developer is also asking: who supplies this, who can deliver at this scale, who has done it before. Assistants are now part of that research. DataReportal’s Digital 2026 analysis found that 77 percent of online adults in Nigeria say they are “excited” by AI, and that well over a billion people use standalone AI tools each month, so asking an assistant is no longer unusual. See AI search statistics for 2026.

What was the problem to solve?

• Low visibility in assistant answers. Seventh place in ChatGPT for a core product query.
• Competition that had been entrenched for years. The co-founder says the team had tried for years to beat competitors.
• A high-value, considered pipeline. Fewer but larger deals, where being on the shortlist early matters most.

What did we build?

Through my consultancy Vatam, I built city-level collection pages and buyer-intent articles. The collection pages were designed to give assistants a clear, specific page for each place a buyer might name. The articles answered the questions buyers ask before they choose a supplier. The principle is the one I set out in how to get cited by ChatGPT: a specific page for each specific question, with facts an assistant can verify.

What city-level collection pages are

For a retailer, a collection page groups products. A city-level version frames that collection around where a buyer is: luxury chandeliers in Lagos, for instance, with the availability, delivery and showroom details that matter to someone in that place. It serves two readers at once: a buyer who wants products, and an assistant that needs a specific page to cite for a location-based question.

What buyer-intent articles are

These answer the questions that come before a purchase: how to choose a chandelier for a high ceiling, how lighting affects a hotel lobby, what a supplier needs to know before a quote. Each is written for one question, with plain facts, and links to the relevant collection. This is the same one-page-per-question structure that took Lens & Co by Tiara to page one and ChatGPT’s first place in a different category.

What changed, and when?

Alongside the assistants, the store’s own data moved. Total sessions rose 233% year on year (Shopify, September 2026 against September 2025), sessions from ChatGPT reached 34 in the 61 days to 12 October (below), and the site earned 29,400 search impressions between 20 March and 10 September 2026 (Search Console).

What does ChatGPT say about LAGE Lightings now?

The July and October checks used one prompt, luxury chandeliers in Lagos. On 12 October the question was wider and plainer: the words a homeowner or a hotel buyer might type before they know what they want. ChatGPT listed LAGE Lightings first, under its own heading “Best places to buy luxury lights in Lagos”, and gave the buyer what a shortlist needs:

• What it is best for: luxury chandeliers, contemporary pendant lights, and premium residential or commercial lighting.
• Where it is: a showroom at 15C Chief Waheed Eletu Street, Osapa London, Lekki, Lagos.
• What it sells and for how much: chandeliers, pendant, ceiling, wall and decorative lighting, at roughly ₦8,500 to ₦1.5 million depending on the design.
• How it arrives: nationwide delivery, with Lagos deliveries typically taking 24 to 48 hours.
• Who it is for: a living room, a duplex, a hotel or a luxury apartment.

Two details are worth noticing. ChatGPT wrote “formerly Lights Architect”, so it knows the brand changed its name and links the old name to the new one. And the answer is built from facts a buyer can act on: an address, a price range and a delivery time. Those are the plain, checkable statements the city-level pages were written to carry. I have not audited each figure against the live store for this study, so treat the prices and delivery times as ChatGPT’s wording of the store’s information, not as the store’s own published terms. The showroom address does match the address on the page Google ranks first, shown below.

Is ChatGPT sending visitors to LAGE Lightings?

An assistant can name a business without sending anyone to it. The store’s Shopify analytics suggest that, for LAGE Lightings, the names are turning into visits. Across the 61 days from 13 August to 12 October 2026, Shopify recorded 34 sessions from ChatGPT, up 240% on the 61 days before (13 June to 12 August), which implies about 10 sessions in the earlier period. The two periods are the same length, so the comparison is like for like.

Read this carefully, because the numbers are small. Thirty-four sessions is a trickle, and a base of about ten makes any percentage look large. What matters is the pattern and the path. Sessions from ChatGPT appeared in a business that sells high-value items to considered buyers, they rose when ChatGPT began naming the brand first, and a single visit from the right architect or hotel buyer can be worth more than hundreds of casual ones. This is a different measure from the 233% year-on-year growth in total sessions, which counts every source. It is a signal to keep tracking, not a revenue figure.

What does Google show for “where to buy droplights in lagos”?

Different engines answer the same need differently, and this Google result is a useful counterweight to the ChatGPT answer. Google’s AI Overview for the search lists physical markets and showrooms, starting with sellers in Alaba International Market. What sits behind its “Show more” button is not visible in the screenshot, so the capture cannot say whether LAGE Lightings is named further down. Directly below the overview, the first organic result is the store’s own droplights page.

Three observations follow:

• The page follows the one-page-per-product-per-place pattern. A page about droplights in Lagos, with the showroom address in its snippet, is the kind of specific page this case study describes. It is also the first organic result for the search.
• The old brand still shows. Google lists the result under “Lights Architect” on the lightsarchitect.com domain, while ChatGPT names “LAGE Lightings (formerly Lights Architect)” and links to lagelightings.com. It is worth checking that the old domain redirects to the new one and that the new name is used on every page, so every engine resolves a single business.
• Winning one engine does not win them all. ChatGPT lists LAGE first for a broad question while Google’s overview, in the part that is visible, leads with market traders. That is why this case study tracks several assistants and Google, not one.

How should you read “7th to 1st”?

Position in an assistant’s list is not the same as a Google ranking, and it helps to be precise about what the number means:

• It is a position in one answer to one prompt. ChatGPT answers vary between runs and users, so I record the date and repeat checks.
• “First of five” is a list. In October the answer listed five picks and LAGE Lightings was first.
• It covers a defined query. Luxury chandeliers in Lagos, not every lighting search.
• The 12 October answer is a different prompt. “Where can i buy luxury lights in lagos” is broader than the luxury chandeliers question behind “7th to 1st”. It supports the direction of travel, but it is not a before-and-after for the same wording. The capture also does not show whether the account was signed in, so personalisation cannot be ruled out.
• It is consistent with other assistants. Gemini and Claude also name the brand first for luxury lighting in Nigeria, which makes it less likely to be a one-off.

What did it do for the business?

LAGE Lightings has done nine figures in naira revenue as a business. That is the total for the business, not revenue I can attribute to one channel, and I would not claim otherwise. What its co-founder reports is a change in who was getting in touch: “Victor came in, and within weeks, we were getting inquiries from top architects, 4-star hotels real developers, eventually leading to some of the biggest deals we’ve ever closed.”

That is the pipeline effect worth paying attention to. Being first in an assistant’s answer does not only add visits. It changes the quality of the shortlist a buyer builds before contacting anyone. For a supplier whose customers are architects, hotels and developers, getting named early in a decision process can matter more than raw traffic. I explain the mechanism in AI citation versus AI recommendation.

Why might this have worked?

This is my analysis, not a measured result. I build every page with three qualities, and they are the most likely reasons it worked:

• Specific. Each page answers one question or serves one place, so an assistant has a precise page to cite.
• Plain facts. Products, locations and details are written as short, checkable statements.
• Consistent. The same facts appear across the site, so an assistant finds the same answer wherever it looks.

Google’s documentation on helpful, reliable, people-first content asks for the same qualities of accuracy, completeness and originality, which is why I do not treat AI visibility as separate from good SEO. Its guidance on AI features says the SEO fundamentals remain relevant and no special optimization is needed. The three disciplines are compared in SEO, AEO and GEO explained.

What could another premium retailer take from it?

1. Map your buyers’ questions, including the ones professionals ask, such as architects or procurement teams.
2. Build one page per place and per question rather than one generic category page.
3. State facts plainly: lead times, delivery areas, showrooms, ranges, minimum orders where relevant.
4. Be consistent on every page and in your profiles.
5. Test monthly across assistants, with a baseline recorded before you start. The method is in how to check what ChatGPT says about your business.
6. Ask enquirers how they found you and track the quality of leads, not only their number.

How far should you trust these numbers?

• AI answers change between runs, users and models. Each result here is a dated snapshot, and I look for direction over months.
• This is one brand in one market. Another business with different competitors may see a different result.
• The nine-figure revenue is a business total, and the 233% is year-on-year sessions on Shopify. Neither isolates the effect of any one activity.
• The ChatGPT sessions figure is small, 34 in 61 days against about 10 in the previous 61, and its ChatGPT label comes from the store’s Shopify analytics, not from a filter visible in the screenshot. Treat it as an early signal.
• The Google capture was made while signed in with the location set to Lagos, and the AI Overview in it is truncated.
• The inquiry quality is reported by the client, which makes it a testimonial, not an independent measurement.
• No ranking or AI placement can be guaranteed.

What the case does show is that a focused set of specific pages can move a business up an assistant’s shortlist within a quarter. To test it for your own business, start with this 30-minute check. For a second example, read the Lens & Co by Tiara case study, and see the original case study on Vatam. My service is described on the GEO strategist page.

How does AI visibility fit a high-value, considered purchase?

For a retailer of everyday items, the buying path is short: ask, click, buy. For premium lighting supplied to architects, hotels and developers, the path is longer and has more people in it. A simple model of where assistants can matter:

The honest conclusion is that AI visibility acts mainly at the top of this funnel, deciding who gets considered. What happens after that depends on the business. That fits the co-founder’s description of a change in the quality of inquiries, followed by deals that came through the normal sales process.

What would I measure going forward?

1. Position and coverage in assistants for a fixed panel of buyer questions, including professional ones such as “lighting suppliers for hotel projects in Lagos”.
2. Enquiry quality: the type of buyer, the project size and the source, recorded when an enquiry arrives.
3. Source of closed deals: asking buyers how they first heard of the business.
4. AI referral visits and branded search to catch influence that does not click.
5. Revenue by source, with the caveat that attribution for considered purchases is imperfect.

How does this compare with Lens & Co?

Different categories, buyers and timescales, the same underlying method: specific pages for specific questions, plain facts, consistency and monthly measurement. Read the full Lens & Co case study to compare.

What risks and caveats should a premium brand consider?

• Dependency. Assistants change. A position held today can move, so keep investing and keep measuring.
• Accuracy. An assistant can describe your range or lead times wrongly. Check what it says and fix the sources.
• Attribution. Considered purchases involve many touches. Do not claim a single cause.
• Brand fit. Premium positioning requires premium pages, not thin location pages. Google’s spam policy on scaled content abuse is a reminder that pages generated at scale without value are a risk.
• Competitors respond. A first place invites imitation, so publish genuinely useful pages that are hard to copy.

How should you brief a provider?

1. Share your target buyers and the questions each asks, including professional intermediaries.
2. Agree the panel of questions and a baseline before work starts.
3. Agree what will be reported monthly and how enquiries will be attributed.
4. Set expectations: results vary, rankings and AI answers cannot be guaranteed and answers change between runs.
5. Ask to see a comparable result with its sources and limits.

What does a standard 90-day approach look like?

This is my standard method for a considered-purchase brand, described as a method and not as a record of what happened month by month in this engagement.

Month 1: baseline and foundations

• Run the assistant panel and record where the brand and its competitors appear.
• Audit crawl access, indexing, titles, descriptions and structured data.
• List the buyer questions, including those asked by professionals.
• Map each question to a page, and identify the gaps.

Month 2: build and structure

• Publish city-level collection pages and the first buyer-intent articles, with plain facts and named authorship.
• Link the pages together and to the product collections.
• Make the business identity consistent across the site and its profiles.

Month 3: measure, fix and extend

• Re-run the panel and compare with the baseline.
• Fix the gaps the test exposes, such as missing facts or unclear pages.
• Extend to the next set of questions, and begin tracking enquiry quality and source.

What should a founder ask after reading this?

1. Where do we appear today? Run the 30-minute test and record it.
2. Who are the intermediaries in our market, and what do they ask an assistant before they contact suppliers?
3. Do we have a specific page for each of those questions?
4. Are our facts identical everywhere?
5. How will we know a result is real, and what will we measure besides traffic?

How do you protect a first place?

A position in an assistant’s list is not owned. To defend it:

• Keep content fresh. Update pages when ranges, lead times or prices change, and show the updated date.
• Keep measuring monthly, and investigate any drop quickly.
• Keep earning real mentions and reviews, because reputation signals support recommendations.
• Keep expanding into the next questions buyers ask, so you are not dependent on a single query.
• Keep the foundations healthy: crawler access, speed, indexing and structured data.

What objections do founders raise about a result like this?

• “Isn’t one assistant’s list just luck?” It could be for a single check, which is why I record dates, repeat questions and compare across assistants. Here Gemini and Claude also named the brand first, which makes a fluke less likely.
• “Did the ranking cause the big deals?” I cannot prove that. The co-founder reports a change in who was contacting the business and describes the deals that followed. That is a credible account, and it is a testimonial, not a controlled test.
• “Will it last?” Not automatically. Assistants change, and so do competitors. That is why measurement and fresh content continue after the first result.
• “Is 233% growth a lot?” It is year-on-year sessions on Shopify, from a base that I do not publish. Percentages from a small base can mislead, so read it alongside the other signals.
• “Can we do this ourselves?” Yes, in principle. The steps are documented in how to get cited by ChatGPT. What most teams lack is time and a measurement discipline, which is what a specialist adds.

Sources

• LAGE Lightings
• Vatam case study: LAGE Lightings
• Google Search Central: creating helpful, reliable, people-first content
• Google Search Central: AI features and your website
• DataReportal: Digital 2026, more than 1 billion people use AI