Data as of 18 September 2026 · Site live for about 2 months
Who this case is for
If you run a service business in a tourist town, such as transfers, private taxis, tours or boat hire, and you are about to launch a new website, this case shows the first 60 days after launch: what to do first, what to measure, and what kind of numbers to expect.
We wrote it as a guide you can follow, with the work in date order, how to count contact taps, how to choose the search terms that move most easily, and a checklist at the end.
We do not name the business, to protect the client’s privacy.
1. The starting point: a new domain with no history and no visitors
The owner runs a fixed-price private taxi and airport transfer service in Pattaya, Chonburi. The main routes are to and from the airports in Bangkok and in the east, into Bangkok, to other resort towns, and to piers for island ferries. There are 6 vehicle classes, from taxis and sedans up to SUVs and vans.
Most customers are foreign tourists who book before leaving their own country, families and business travellers. They want to know the price first, want a driver waiting at the airport, and want to talk on the channel they already use.
When work started on 16 July 2026, the domain was newly registered, with no history, no links from anywhere and no visitors at all. The owner sent the fixed price list for each route and vehicle class, and we handled the site structure, content and measurement.
This website has no payment button
The site has no booking or payment system, and the booking page has no form. Customers book through WhatsApp, phone, LINE, a Chinese chat app and email. The upside is a fast launch. The downside is that without extra tracking, the owner has no idea how many people the website leads to get in touch. This is the heart of the case, covered in section 2.
Low search volume, but searchers are ready to book
Before the big page expansion, we pulled keywords from Ahrefs. Most route searches had a difficulty (KD) of 0–2 and low volume, but almost everyone typing a route name is about to travel.
A point tourism business owners should know: the Thai term for this business gets only about 10 searches a month, because the real customers are not Thai. A Thai-only website would be almost impossible to find. So this site started in English, then expanded into the languages of the main tourist groups.
2. What we did in the first 60 days
The actual order of work, from day 1 to day 60
| Date (2026) | Work done | Pages generated |
|---|---|---|
| 16 Jul | Built the English site, then added German, Russian, Thai, Simplified Chinese and Vietnamese, with the owner’s fixed prices | 157 |
| 17 Jul | Refined the design, 22 cover images, 6 vehicle images, menus, 9 guide articles, then launched the live site | 166 |
| 18–19 Jul | First site quality check: speed, rendering and structure | 166 |
| 13 Aug | Grouped content into clusters and built 5 pillar pages | 172 → 184 |
| 16 Aug | Installed GA4 | 184 |
| 25 Aug | Researched keywords in Ahrefs, added return-route pages, added Traditional Chinese, compressed images | 203 → 236 |
| 29 Aug | Checked Search Console and GA4, connected the two, submitted a new sitemap, started counting contact taps | 236 |
| 2 Sep | Fixed issues from the SEO audit, translated 10 guide articles into 5 more languages | 296 |
| 18 Sep | Set contact taps as a Key event and adjusted pages for page-2 search terms | 296 |
The page counts in the table are pages the system generated. The sitemap submitted to Google had 274 pages as of 19 September 2026, because some pages are intentionally kept out of Google’s index.
Site structure: every route has its own page
Tourists do not just search “taxi”. They search for the route they want. So we built the pages around what people actually search.
- 8+ route pages, such as to the airport, into Bangkok, to other resort towns and to island ferry piers
- 3 return-route pages: people at the destination search the reverse, such as from the island back to town. Without these pages, nothing would match their search
- 3 airport pages, one for each airport customers fly into
- Service pages: private driver, family vehicles, business vehicles and airport transfers
- 3 accommodation-area pages for people who search by the name of the area where they booked their hotel
- Pricing, reviews and booking pages that show contact options clearly
- 5 pillar pages as the hub of each category, with internal links generated from central data on every service page
- Guide articles: 13 in English and 13–14 in each other language, such as a private taxi vs ride-hailing app comparison
7 languages, chosen from who actually visits
English is the main language, followed by German, Russian, Thai, Simplified Chinese and Vietnamese. Traditional Chinese was added on 25 August, after keyword research showed that tourists from Taiwan and Hong Kong spell terms differently from mainland China. The Traditional Chinese pages are generated automatically from the Simplified Chinese pages, so there is no second set of content to maintain.
Every page has hreflang telling Google which language each page is in, covering all 7 languages plus x-default, so German users see the German page and Taiwanese users see the Traditional Chinese page.
Technical basics that had to be right from day 1
- A light static site: about 2.2 KB of JavaScript and no layout shift while loading (CLS about 0), from the check on 18 July
- Images 43% lighter: the whole site went from 12.8 MB to 7.3 MB on 25 August
- Complete schema: business details, taxi service, price list, FAQs and breadcrumb navigation
- Open to AI: robots.txt allows AI bots, and an llms.txt file summarises the business so AI can read it easily
Re-checking as the site grew
A site with nearly 300 pages in 7 languages needs audit rounds after launch, because small problems are always multiplied by the number of pages. The audit round from late August to early September found and fixed the following:
| Issue found | Count | What we did |
|---|---|---|
| Search Console flagged duplicate review ratings on one page | 30 pages | Kept a single review data set per page |
| Headings skipping from H1 to H3 | 140 pages | Reordered heading levels |
| Broken internal links | 2 links | Fixed the destinations |
| Meta descriptions too long | 63 pages | Shortened them |
In the re-check on 2 September, every item was at 0.
Count what makes money, not visitor numbers
This is the most important lesson of this case. By default GA4 counts visitors and page views, but not taps on a phone link, opening WhatsApp or adding a LINE contact. For a business that closes sales in chat, these taps are the signal closest to revenue that a website can measure.
So on 29 August we created an event called contact_lead that fires every time someone taps phone, WhatsApp or LINE. On 18 September we set this event as a Key event in GA4, so every report shows which channel brings people who get in touch.
3. Results: real numbers at about 2 months old
Google sees the site: 273 pages indexed
23 pages are not yet indexed. Of these, 17 are intentionally excluded, 3 are redirects and 1 no longer exists, leaving only 2 pages that Google has crawled but not yet indexed. For a 2-month-old site, that is a clean state.
More people see the site every week
At first the site appeared on Google about 30–50 times a day. By mid-September it was at about 350 times a day (the early figures are read from the Search Console chart, so they are approximate). This chart shows the latest 10 days for which we have daily numbers.
(5)7
(3)8
(4)9
(1)10
(1)11
(3)12
(4)13
(2)14
(5)15
(5)
Splitting the data into two periods makes the picture clearer.
| Period | Clicks | Times appeared on Google | Average position |
|---|---|---|---|
| First 34 days (16 Jul – 18 Aug) | 29 | 3.2K | 20 or higher (approx.) |
| Latest 28 days (19 Aug – 15 Sep) | 80 | 6.88K | 13.9 |
The first-period figures are calculated as the overall total minus the latest 28 days. The two periods do not overlap, but they are not the same length, and every new site gets more visibility as it ages anyway. So part of the growth comes from time passing, not from our work alone. Cumulative clicks stood at 44 at the check on 29 August and at 109 on 15 September.
Other languages started getting clicks within a few weeks
In the latest 28 days, clicks came from Thailand 25, Germany 9, India 5, Taiwan 4, Australia 4, the United States 3 and Russia 2. A German route page got 3 clicks, and the Traditional Chinese route page into Bangkok got 3 clicks. The numbers are still small, but they confirm that pages in tourists’ languages are actually being found, not just sitting there.
The page with the most clicks was the article comparing private taxis with ride-hailing apps, with 10 clicks from 2,150 appearances. Articles that answer pre-booking questions often rank before sales pages, because long questions are easier to compete for.
52 contact taps in the 20 days since tracking began
From 21 August to 17 September 2026, the site had 316 sessions from 285 users, with a 48% engagement rate. The contact_lead event fired 52 times from 36 users, or about 12.6% of all users, with only about 20 days actually tracked, because the event started working on 29 August.
To be clear: these 52 are taps on phone, WhatsApp or LINE, not bookings. Some people tap and never send a message, and some tap two channels. There is no data yet on real bookings or revenue from the site. The most accurate numbers have to come from the owner’s own chats and phone.
Visitors from AI got in touch about 4x more often than visitors from Google
GA4 separates the channels people arrive from clearly, and the most interesting number is for visitors from AI assistants such as ChatGPT or Perplexity.
A likely reason is that people who ask AI already have specific questions, such as how to get from the airport to their hotel and roughly how much it costs. When the AI recommends the site, these people are more ready to get in touch. This is why the site lets AI bots read it, has an llms.txt file, and writes key facts such as routes, vehicle classes and contact channels as easy-to-read text, not hidden in images.
How to push terms up from page 2: start with the closest ones
A CTR of 1.2% at an average position of about 14 says the site is visible, but still too low for people to click. At this stage, building more new pages does not help much. What helps is bringing terms that are almost on page 1 up. On 18 September we followed these steps:
- Open Search Console and look at search terms at positions 8–20 that people actually see, not terms that appeared only 1–2 times
- Check which page each term sends people to, then open that page and see whether it directly answers what people searched
- Fix the parts that match the search term most closely, such as the page title, H1, subheadings, or add a section that answers that question directly
- Wait 1–3 weeks, then check the ranking of the same term. Do not re-edit every day, because Google needs time to re-read the page
| Search term found (paraphrased, town name omitted) | Position, latest 28 days | What we changed |
|---|---|---|
| Reviews of transfer services in the town | 8.4 | Adjusted the title and H1 of the reviews page |
| Ride-hailing app fare from Bangkok to this town, current year | 9.4 | Added a section explaining ride-hailing app fares |
| Route to a popular island | Not recorded | Adjusted the page title and added a question about the ferry |
| Trips to the islands off the town | Not recorded | Adjusted an H2 to match the search term |
| Airports in eastern Thailand (Thai) | Not recorded | Added an airport section to the Thai page |
Another term worth knowing is “town name + cab”, which is already at position 6.6, while the broad term “town name + taxi” is still at position 42 and the Thai term is at position 65.9. Broad terms like these do not move from on-page fixes alone.
Expected results from here
All the numbers above are real results. This chart shows what we expect from the work just done, and it is not a result yet, because the page changes were only made on 18 September and it is too early to measure.
The next job, not yet done, is building trust from outside the site, such as a Google Business Profile and listings in tourism business directories. This is necessary for broad terms like “town name + taxi” in English, Thai and Russian. This site has not been set up as a project in Ahrefs yet, so there are no Domain Rating or external link numbers to compare.
First-60-days checklist for transfer and tourism service websites
Open your website and answer each point.
Content
- Every route you actually sell has its own page, including return routes
- Your site’s languages match your customers’ real nationalities, not just Thai and English
- Vehicle classes, seat counts and what the service includes are written as text, not only shown in images
- You have articles that answer pre-booking questions, such as comparisons with ride-hailing apps or how to get to the island ferry pier
- Phone, WhatsApp and LINE buttons are clearly visible on every page on mobile
Technical and measurement
- hreflang covers every language, and your sitemap has been submitted in Search Console
- You have checked the Search Console indexing report to see whether non-indexed pages are intentional or a problem
- GA4 counts phone, WhatsApp and LINE taps, and they are set as a Key event
- You look at the AI assistant channel separately from Google in GA4
- Every 2–3 weeks, you pick terms at positions 8–20 and adjust the matching pages
If you cannot tick more than half, people may be seeing your website, but you do not yet know who is ready to book.
Questions transfer business owners often ask us
How long does a new site take before Google sees it? In this case the site launched on 17 July. By 29 August Google had indexed 149 pages, and 273 pages by 18 September. What we took care of from the start was a light site, a clear internal link structure and a correct sitemap. But being indexed is not the same as ranking. Clicks come later.
Is a multilingual site worth it if most customers speak English? For a specialised tourism business, yes. Search terms in other languages have far less competition than English. In this case, the German and Traditional Chinese pages started getting clicks within a few weeks. But choose languages based on data about who visits and what they search, rather than translating into every language.
If people tap to call or chat from the site, does that mean I have customers? Not yet. A tap is the interest closest to a booking that a website can measure, but you need to check your chats and phone to see how many people actually got in touch and how many booked. The simplest way is to ask every customer where they found you.
What should I do so AI like ChatGPT recommends my website? Allow AI bots in robots.txt, have an llms.txt file, and write key facts as clear text, such as routes, vehicle types and contact channels. In this case, visitors from AI made up only 80 sessions, but they got in touch at about 4x the rate of visitors from Google.
We build websites and do SEO for service businesses in tourist towns
If you run transfers, a private taxi, tours or any service customers book before they travel, and you want tourists in many languages to find you and to know how many people actually get in touch through your website, start with a free website audit or see our website development services and SEO services.