Skip to main content

กำลังโหลด...

Southern Whale

Case Study: 5-language taxi and airport transfer website grows Google clicks about 12x in 3 months

Private Taxi / Airport Transfer / Travel ServicesPrachuap Khiri Khan (Hua Hin)seo

A private taxi and flat-rate airport transfer operator in Hua Hin, Prachuap Khiri Khan (business name withheld for privacy)

Kreangsak S.
Kreangsak S. · Southern Whale team
Illustration of an airport transfer website on a computer screen, tablet and phone, with a whiteboard showing an upward trend chart, no people in the image

1,300Google clicks in the latest 3 monthsPrevious 3 months: 105 · about 12x · Search Console, read on 18 Sep 2026

218WhatsApp taps from the websiteFrom 174 people, about 18% of users · GA4, 21 Aug–17 Sep 2026

12%Sessions from AI assistants156 of 1,269 sessions, e.g. ChatGPT and Perplexity · GA4, same 28 days

10Average position on GooglePrevious 3 months: 20 · Search Console, same period as clicks

In short: Comparing the latest 3 months with the previous 3 months in Search Console, this site got 1,300 clicks from Google, up from 105. Average position moved from 20 to 10, and in the latest 28 days 174 people tapped WhatsApp on the site. Part of this growth comes simply from the site getting older. We explain this in section 3.
Data in this article as of 18 September 2026 · Site about 8 months old, counting from when it was first built in January 2026

Who this case is for

This case is for you if you run an airport transfer, day-hire taxi, VIP van or day-trip business in a tourist town, and you have a multilingual website for foreign visitors, but you are not sure whether it actually brings in customers or just works as an online business card.

We explain what problems a 5-language website had that you cannot see by looking at the pages, what we fixed, and how the real numbers in Search Console and GA4 changed. We are also honest about which part likely came from our work and which part came from time passing.

We do not name the business, to protect the client’s privacy.


1. The starting point: a 5-language website with good foundations, but few people finding it

The client is a flat-rate private taxi operator in Hua Hin, Prachuap Khiri Khan. They drive between the town and Bangkok, including both Bangkok airports, offer rides around town, run same-day return trips, and have VIP vans for larger groups.

Most customers are foreign tourists from English-speaking countries, Germany, China and Vietnam, followed by Thai holidaymakers, families, business travellers and hotel guests. So the website runs in 5 languages: English, Thai, Chinese, Vietnamese and German.

In mid-July 2026 the site had 186 pages. We audited the whole site and found that many basics were already done right, such as canonical tags, hreflang, schema and the sitemap. But the numbers in Search Console were still very small.

105
Google clicks in 3 months (roughly mid-Mar to mid-Jun)
9,640
Times the site appeared on Google in the same period
1.1%
CTR, the share of people who saw it and clicked
20
Average position, which is page 2 of Google

There was also a measurement gap. Until late August, GA4 was not connected to Search Console, and nobody was counting how many people tapped WhatsApp or the phone number on the site. For a transfer business this is the most important number, because tourists usually book by chat rather than by filling in a form.

How we worked

We did not rebuild the site. We fixed things in rounds and checked the results before starting the next round.

Date Work done
18–19 Jul 2026 Audited the whole site, fixed 7 main issues and pushed them live
13 Aug 2026 Organised hub and child pages, added in-article links to 29 articles across all 4 languages that have a blog
25 Aug 2026 Researched keywords in Ahrefs, added 10 new pages, compressed images
29 Aug 2026 Reviewed Search Console and GA4, connected the two, requested indexing for new pages, set up tracking for WhatsApp taps and phone taps
2 Sep 2026 Second audit, added Article schema to every article, removed duplicate pages with no inbound links
18 Sep 2026 Reviewed the trend, marked contact taps as Key events, and rewrote the titles and descriptions of Thai pages

2. What we found and what we fixed

Wrong language codes: Google may not know which language a page is in

A multilingual site has to tell Google, using standard codes, which language each page is in. This site used “cn” for Chinese pages and “vn” for Vietnamese pages, 18 pages in total. Those are country codes, not language codes. The correct codes are “zh” and “vi”.

You cannot see this problem on the page at all. A Chinese visitor can still read the Chinese page as normal. But search engines may match the language wrongly and show the wrong-language page to searchers. We fixed all 18 pages on 19 July.

Wrong heading order on every page, and almost no images with a set size

The site menu used h3 and h4 heading tags in 24 places, and these appeared before the main h1 heading on every page. So the page outline that search engines read was out of order. We restructured every page to start with h1, followed by h2 and h3.

The second issue was that 505 of 507 images had no width and height set. This can make the page jump while it loads. Also, the main hero image at the top of the homepage was not set to load first.

Fix Before After Date
Pages with a wrong Chinese or Vietnamese language code 18 pages 0 19 Jul 2026
Images with no set size 505 8 (only logos from external sites remain) 19 Jul 2026
Images with no image description (alt) 4 0 19 Jul 2026
Links reading “Click Here” 3 0 19 Jul 2026
Skip-to-main-content button for keyboard users None On all 186 pages 19 Jul 2026
Homepage hero image 122 KB, not preloaded 59 KB, preloaded in 5 languages 19 Jul and 25 Aug 2026
Weight of 35 compressed images 6.8 MB 4.6 MB (down 32%) 25 Aug 2026
Articles with Article schema and a publish date 0 of 34 34 of 34 2 Sep 2026

We did not measure Lighthouse or PageSpeed scores before and after the fixes, so we have no speed numbers to quote in this case. The numbers in the table are what we could count by checking the live site files.

Articles barely linked to each other, so Google could not tell what the site specialises in

When we checked in mid-August, the review article had no links to any other page on the site, and 3 other articles had only 1 link each. On top of that, the Thai, Chinese and Vietnamese article index pages had a bug that produced broken links.

We grouped the content into 3 clusters based on what customers actually book.

  1. Airport transfers: every route to and from the airport
  2. Taxi services: rides around town, day hire, VIP vans
  3. Nearby trips: same-day return trips around the town

Each cluster has one hub page, with 21 child pages in total. Child pages link back to their hub and to sibling pages in the same cluster, using link text in that page’s own language. Thai pages link to Thai pages, German pages link to German pages, with no mixing. We built this as a single central data set, so when a new page is added, the links appear in every language automatically.

We then added about 98 more links inside the body of 29 articles.

In-article links added, by language
29 articles · 13 August 2026 · about 98 links in total
English29
Chinese25
Vietnamese24
Thai20
Source: check of the site files after going live · 343 article links checked, no broken links · 1,040 internal links from the hub and child page clusters, no broken links

This work needed no new content at all. It simply connected the pages that already existed into one story, so Google can see more clearly that the site covers every angle of transfers in this town.

10 new pages, chosen from keywords with near-zero difficulty

On 25 August we pulled keywords from Ahrefs and kept only those with a difficulty (KD) of 0–2. We then built 2 new topics, each in 5 languages.

  • A VIP van page: a direct sales page targeting searches for vans from Bangkok to this town
  • A guide to the town’s airport: answers the questions tourists search before they travel, then leads them to the booking page

We also adjusted the existing Thai taxi service page to match the phrase “เช่ารถพร้อมคนขับ” (car rental with driver), which gets about 600 searches a month. The site grew from 186 to 196 pages, and dropped to 195 pages after we removed a duplicate on 2 September. We checked all 1,964 links on the site and found no broken links.

Tracking set up to match how customers really book

On 29 August we set GA4 to count 2 things: taps on WhatsApp links, which appear in 1,277 places across the site, and taps on the phone number, which appears in 701 places. On 18 September we marked both as Key events. We also connected GA4 to Search Console to see what visitors from Google do next on the site.

Rewritten page titles, so people see a reason to click before they visit

On 18 September we rewrote the titles and descriptions of the Thai homepage and the Thai taxi service page. We added the starting price, the 4.9 review rating and the phrase “no hidden extras” to what people see in the search results. This was done after the data period used in this article, so it has no effect on the numbers below yet.


3. Results: real numbers from Search Console and GA4

Google clicks, latest 3 months vs the previous 3 months

These numbers were read from Search Console on 18 September 2026. The latest period is roughly mid-June to mid-September. The previous period is roughly mid-March to mid-June 2026.

Google clicks per 3-month period
Search Console, read on 18 September 2026 · grey = previous · blue = latest
105
1,300
Mid-Mar to mid-JunMid-Jun to mid-Sep
Source: the client's Google Search Console · Search Console shows latest-period clicks as "1.3K", so the true figure may differ slightly from 1,300 · The best days near the end reached about 40 clicks a day, read from the chart
Metric Previous 3 months Latest 3 months Change
Clicks 105 1,300 About 12x
Times the site appeared on Google 9,640 47,800 About 5x
CTR 1.1% 2.7% About 2.5x
Average position 20 10 10 positions better

In the latest period, 1,052 search terms made the site appear on Google.

The part that comes from site age. This site was first built in January 2026, so the previous period was when the site was still very new. New sites usually rank better on their own after some time. Also, the latest period starts in mid-June, but our first round of fixes went live on 19 July. So roughly the first month of the latest period had none of our work in it. We have no control group to separate the effects, so the 12x figure is the combined result of our work, site age and other factors we cannot measure separately. It is not the result of our work alone.

The fastest-growing terms were route searches, not the brand name

If clicks grew only because more people knew the brand, the growing terms would be the business name. The data says otherwise.

Clicks for the main search terms, previous 3 months → latest 3 months
Search Console, read on 18 September 2026 · town name and brand name hidden
The business's brand name (English)14 → 41
Taxi from Suvarnabhumi Airport to this town (English)0 → 17
Bangkok airport to this town (English)1 → 13
How much is a taxi from Bangkok to this town (German)1 → 13
Taxi + town name (English, variant 1)3 → 9
Taxi + town name (English, variant 2)0 → 9
Taxi + town name (English, variant 3)1 → 8
Taxi price from this town to Bangkok (English)1 → 7
Taxi + town name (Thai)0 → 6
Taxi from Bangkok to this town (German)0 → 6
Bar length = clicks in the latest period · grey = brand name · blue = non-brand terms · Source: the client's Google Search Console, top 10 terms by clicks

The brand name grew about 3x, from 14 to 41 clicks. The 9 non-brand terms in this list grew from 7 to 88 clicks combined. These people did not know the business before. They searched for the route they wanted and found the site.

A clear example is the English term “Bangkok airport to this town”. In the previous period the site appeared 95 times for it, and 450 times in the latest period. The Thai term “taxi + town name” went from 186 to 451 appearances.

The German terms are interesting too. 2 terms in the list are German questions about taxi fares and travel from Bangkok. The first went from 1 to 13 clicks. During the same period we fixed the language codes and kept links within the same language across the site, but we did not edit the German pages directly. So we can only say this happened at the same time. It does not prove it was caused by that work.

Ahrefs saw only 2 keywords, but Google saw 1,052

On 25 August, Ahrefs saw this site ranking for only 2 keywords: “taxi + town name” at position 6 and “Suvarnabhumi Airport to this town” at position 8. Ahrefs estimates both at about 10 searches a month. Meanwhile, Google’s Search Console showed 1,052 search terms and 1,300 clicks in the latest period.

External tools estimate from their own crawl data, which cannot keep up with new sites. A business owner who looks only at Ahrefs would think nobody finds the site at all. The data you should base decisions on is your own site’s Search Console.

Visitors from Google actually tap WhatsApp, not just browse

GA4 for the 28 days from 21 August to 17 September 2026 showed 1,269 sessions from 971 users.

People who tapped to get in touch from the site, in 28 days
GA4, 21 August – 17 September 2026 · number of people
174 people
5 people
Tapped WhatsApp (218 taps)Tapped to call (6 taps)
Source: the client's GA4 · Tracking started on 29 August, inside this 28-day window, so it does not cover the full period · A tap is a contact attempt, not a booking or a sale, and needs to be confirmed against the client's actual chats

174 of 971 people, or about 18%, tapped WhatsApp, while only 5 people tapped to call. The likely reason is that international calls are inconvenient for foreign tourists, while typing a chat with their flight details is instant. If your business only counts forms or calls, you will miss your largest group of prospects.

GA4 on this site has no before-and-after comparison, because the date comparison in GA4 was set up wrongly and contact tracking was only installed in late August. These numbers are therefore the baseline for the next comparison.

AI assistants are already the third-largest channel

Sessions by channel, in 28 days
GA4, 21 August – 17 September 2026 · 1,269 sessions in total
Search from Google and other search engines751 · 59%
Direct visits323
AI assistants such as ChatGPT and Perplexity156 · 12%
All other channels combined39
Source: the client's GA4 · "All other channels combined" is the total minus the first 3 channels · GA4 cannot compare with the previous period, so we can say this share is high, but not how much it grew from

This number counts people who asked an AI assistant and then clicked a link it recommended to reach the site. For a transfer business, a question like “what’s the best way to get from the airport to this town?” is easy for AI to answer. A site that clearly explains routes, travel times and how to book has a good chance of being cited.

What we found is that the CDN in front of this site had an option switched on that blocks some AI bots. This contradicts the site’s own robots.txt and llms.txt files, which allow them. Changing this setting has to be done by the client’s account administrator. At the time of writing it had not been changed, which means the 12% happened even though some bots still cannot access the site.

Indexing status in Google

On 29 August, Search Console reported that Google had indexed 152 pages and the sitemap status was successful. Another 63 pages were not yet indexed, mostly for reasons that are normal for a young multilingual site: 19 pages that point to another main page, 21 pages Google had crawled but not yet indexed, 15 pages Google knew about but had not yet crawled, and 5 old Chinese and Vietnamese addresses that are no longer used. The same day, we queued the 5 van pages for priority indexing.

Expected results from here

The numbers above are real results. The chart below shows the keywords that the 10 new pages target. There are no measurable rankings for them yet.

Target · not an actual result
Monthly searches for the keywords the new pages target
Ahrefs, 25 August 2026 · about 4,400 searches a month in total · all keywords KD 0–2
สนามบิน (airport) + town name (Thai)2,000
Town name + airport (English)900
รถตู้ไป (van to) + town name600
เช่ารถพร้อมคนขับ (car rental with driver)600
รถตู้ (van) + town name + กรุงเทพ (Bangkok)300
Search volumes are Ahrefs estimates, not site visits · Town name hidden to keep the business anonymous
Next step What we expect
Re-measure the CTR of the Thai term “taxi + town name” in early October 2026 It currently appears 451 times and gets 6 clicks. We hope the new title with the starting price and review rating will get more people to click
Ask the client’s account administrator to unblock AI bots at the CDN We expect visits from AI assistants to increase
Compare the next GA4 period against the baseline in this article We will see whether WhatsApp taps are growing once we have a full period of data

Checklist for airport transfer and tour businesses

Open your website and answer each point.

Multilingual website

  • Every page uses the correct language code, e.g. “zh” for Chinese and “vi” for Vietnamese, not “cn” or “vn”
  • Links on a page in one language lead to pages in the same language, so German visitors are not sent to English pages
  • Every language has all the main route pages, such as airport to town and town to Bangkok

Content structure

  • Each service group has a hub page, such as airport transfers, town taxi and day trips, and child pages link back to their hub
  • Every article links to at least 2–3 related service pages
  • New pages are based on keywords people actually search and that are not hard to compete for, not on guesses
  • Your page titles on Google give a reason to click, such as a starting price, review rating or no hidden costs

Measurement

  • You count taps on WhatsApp or whichever chat your customers actually use, not just forms
  • GA4 is connected to Search Console, and you make decisions from Search Console rather than external tools
  • You have checked the GA4 channels to see whether visitors are already coming from AI assistants
  • You have checked that your CDN or firewall is not blocking AI bots without your knowledge

If you cannot tick more than half, your website is probably missing a good number of tourists who are searching for routes.


Questions transfer businesses often ask us

Is translating the content enough for a multilingual site? No. You also need to tell Google the correct language codes, and make sure links in each language lead to pages in the same language. In this case the content was fully translated, but the Chinese and Vietnamese language codes were wrong on 18 pages, which you cannot see on the page at all.

Was the 12x growth in clicks all down to SEO work? Not all of it. The site is about 8 months old, and the comparison period was when it was still very new. New sites usually rank better on their own over time, and our work went live about a month after the latest period had already started. What we can say is that the fastest-growing terms were the route searches that we organised pages and links for.

Why track WhatsApp taps separately? Because it is the channel foreign customers actually use to book. In this case 174 people tapped WhatsApp but only 5 tapped to call. Without tracking, you have no idea how many customers the site sends you. But remember that a tap is not a booking, so check it against your actual chats.

Can AI assistants like ChatGPT really send customers? In this case AI assistants sent 12% of sessions in 28 days, the third-largest channel after search and direct visits. What you should do is write route pages that fully answer travellers’ questions, and check that your site is not blocking AI bots by accident.


We do SEO like this for transfer and tour businesses

If you run an airport transfer, flat-rate taxi or tour business in a tourist town and want to know where Google is misreading your multilingual website, start with a free website audit or see our SEO services.

About the author

Kreangsak S.

Kreangsak S.

Southern Whale team · SEO & WordPress Specialist · Web Developer

Over 10 years building websites and doing SEO, focused on WordPress and e-commerce sites with SEO built in from day one, now using AI to ship faster.

LinkedIn ↗Fastwork ↗Chat on LINE ↗

Have a similar project?

Free consultation · we scope the work before quoting