Case study · Wellness · Rawai, Phuket
A spa everyone liked, that nobody could find.
Evergreen Spa had a 4.7-star rating from 72 reviews before Aroc arrived. That reputation is theirs. It was earned over years of doing good work on people who walked past. What they did not have was any way for someone searching online to discover the place existed.
124
searches Evergreen appears in
120 in month two, 23 in month one, none before that
1,980
times shown in results
1,673 in month two, 552 in month one
41
clicks from Google
21 in month two, 8 in month one — still a small number
100
engaged visits
86 in month two, 40 in month one
01 · The constraint
Reputation was never the problem.
Most spas with a visibility problem also have a quality problem. Evergreen did not. The reviews were strong, the customers who found it came back, and roughly half of them were Thai residents rather than tourists — which is the harder audience to win and the one that does not disappear in low season.
The constraint was the earliest one in the Growth System: Discoverable. A business that cannot be found cannot be chosen, and nothing further along the chain matters until that is fixed.
What Aroc did not do: create the reviews, the rating, or the customer base. Those existed. The work was making an existing good business findable by people who did not already know it was there.
02 · What was built
A site that loads fast and books on WhatsApp.
Nine pages: a home page, one for each of the four treatments, plus services, reviews, about and contact. Built mobile first, because 70% of Evergreen’s bookings arrive by message and almost all of its search traffic is on a phone.
Every page opens a WhatsApp conversation in one tap rather than asking for a form submission — matching how the business already takes bookings instead of asking it to change.
Visit evergreenwellnessphuket.comOpens the live site. Check any of it against what is written here.
03 · What happened
First place for three searches, in the first month.
The site went live at the start of July. By the end of the reporting period it appeared in 23 distinct searches, having appeared in none at the start.
| What people searched | Position | Status |
|---|---|---|
| good spa near me | 1 | Page 1 |
| spa rawai phuket | 1 | Page 1 |
| อบสมุนไพร ใกล้ฉัน — herbal steam near me | 1 | Page 1 · Thai |
| ice bath near me | 3 | Page 1 |
| cold plunge sauna near me | 5 | Page 1 |
| sauna rawai | 10 | Page 1 |
| phuket ice bath | 20 | Page 2 |
| rawai spa | 42 | Needs work |
Read the Month 3 report in full: the same report the client received, with people's names removed.
The caveat, in the report's own words:“Being number 1 for a search nobody makes yet is a good sign, not yet a result. The volume comes later.” Three first-place rankings on low-volume searches is a real signal about whether the approach works. It is not yet money in the till, and it was not presented as such.
04 · What did not work
Eight visits. That is a small number.
552 appearances produced 8 visits from Google. Aroc would rather say that plainly than dress it up, and it was the first thing in the client’s report, not a footnote.
The wellness page was answering the wrong question
Shown 191 times. Zero clicks. Google was matching it to people searching “wellness resort in rawai phuket” — people looking for somewhere to sleep. Evergreen is a day spa, so they scrolled straight past. That single search accounted for 256 of the 552 total appearances.
Fixing it will make the headline number go down, and the client was told so before it happens. 200 people who might book is worth more than 550 who never will.
Three pages were competing with each other
For “rawai spa” — an exact-offer search shown 51 times a month — the wellness page, the home page and the massage page all competed. Google resolved it by ranking none of them well: all three sat between position 42 and 55. This was self-inflicted and is being consolidated into one page.
Desktop performance was far worse than mobile
Average position 13 on phones against 46 on computers, with 351 desktop appearances producing zero clicks. Diagnosed in month one and carried into month two. In month three desktop moved from an average position of 31.5 to 12.8, on 641 appearances and 10 clicks, while mobile went from 10.7 to 8.6 on 1,318 appearances and 31 clicks. Desktop is still the smaller share of the clicks.
Enquiries could not be counted at all
Month 1 could report visits but not contacts, because tracking was not installed. Reporting a number Aroc could not stand behind was not an option, so the report said so and the tracking became the first item of Month 2.
05 · Month two
Three times the visibility, and a mistake that cost a month.
Second reporting month, 28 July to 27 August. Google showed the site 1,673 times against 552, across 120 different searches against 23. Clicks went from 8 to 21.
Total visits went down — 160 against 195 — and that is not a contradiction. The visits that disappeared were people who arrived and left immediately. Visits where someone actually engaged with the page more than doubled, from 40 to 86, taking the engagement rate from 21% to 54%.
Contact buttons were tapped 12 times. That measurement did not exist before, and it is the number the previous report could not answer. What Aroc still cannot see is what those taps became — a message, a booking, or nothing. That is stated in the client’s report in the same words.
The mistake, in full
Month one identified that three pages were competing with each other for “rawai spa”. In August Aroc tried to fix it, chose the wrong page to consolidate onto, and added six internal links pointing at it — pushing the signal away from the home page, which was already the stronger page for that search. The problem got worse, not better.
The cost is measurable. Three pages sharing “rawai spa”: 269 appearances, zero clicks, all three at position 44. The same search phrased “spa rawai”, where one page owns it on its own: position 7, and a click. Same business, same customer — page five when three pages share it, page one when one page owns it.
It was fixed on 1 September, and the reason it went unfixed for a month is recorded rather than glossed: the check that would have caught it did not exist until this month. It runs before every report now — any search matching two or more pages is a finding, whether or not it is new.
Still wrong: a third of the audience cannot buy
“Wellness resort in rawai phuket” produced 321 appearances and zero clicks — 35% of everything the site was shown for, and up from 256 last month. Those people want a hotel room. Evergreen is a day spa.
The prediction attached to fixing it: the headline number should fall by 300 to 400 next month. That is the fix working. Judge month three on clicks, not on the total. (Month three, below, shows where that forecast was right and where it was wrong.)
The Thai pages, at their real size
Three Thai-language pages went live on 8 August. In roughly 20 days they earned 34 appearances and one click. That is small and is not being dressed up. What is worth noting is where they sit: the main Thai page ranks at position 5.3, against 19.7 for the English home page. New pages do not normally arrive on page one.
The honest arithmetic: 1,673 appearances producing 21 clicks is a click rate of 1.26%, slightly lower than month one’s 1.45%. The percentages on this page are large because the starting numbers were small. Two months is early — search results take three to six months to settle, so the direction means something and the size of any single number does not, yet.
06 · Month three
Clicks nearly doubled. The forecast was wrong.
Third reporting month, 28 August to 27 September. Clicks from Google went from 21 to 41. Times shown went from 1,673 to 1,980, visits from 160 to 188, and the people behind them from 105 to 152. The home page moved from an average position of 19.7 to 9.7, on 1,040 appearances — onto the first page of Google for the first time.
Every page on the site improved its position and none went backwards. Four pages now average on the first page of Google; in August one did. Most of the clicks arrived in the second half of the month: the changes went live on 1 September, and Google takes a couple of weeks to re-read a site.
| Measure | Month 2 | Month 3 | Change |
|---|---|---|---|
| Times shown in Google | 1,673 | 1,980 | +18% |
| Clicks from Google | 21 | 41 | +95% |
| Click rate | 1.26% | 2.07% | |
| Different searches it appears for | 120 | 124 | +3% |
| Visits | 160 | 188 | +18% |
| People | 105 | 152 | +45% |
| Engaged visits | 86 | 100 | +16% |
| Contact button taps | 12 | 20 | +67% |
The visits and the people differ because one person can come back: 152 people made 188 visits. Google counts a visit as engaged if it lasts more than ten seconds, includes two or more pages, or triggers a contact action, and just over half of the visits now qualify.
Phones and computers both improved. On phones, appearances went from 906 to 1,318, clicks from 12 to 31 and average position from 10.7 to 8.6. On computers they went from 746 to 641 appearances, 8 to 10 clicks and 31.5 to 12.8.
| Week | Times shown | Clicks |
|---|---|---|
| 28 Aug – 3 Sep | 425 | 6 |
| 4 – 10 Sep | 365 | 4 |
| 11 – 17 Sep | 549 | 14 |
| 18 – 24 Sep | 469 | 15 |
| 25 – 27 Sep (3 days) | 172 | 2 |
The forecast, and what it got wrong
Last month’s report said the total number of times shown would fall by 300 to 400 once the hotel-search problem was fixed. It rose by 307.
The fix worked. “Wellness resort in rawai phuket” went from 321 appearances to 52 — a fall of 269, close to the forecast. Its position barely moved (50.2 to 52.2), which is the useful part: Evergreen did not get better at ranking for hotel searches, it stopped turning up in them. What the forecast got wrong was the total. The rest of the site grew by more than that one search shrank.
The mistake was forecasting one part of the site while the rest of it was improving. Predictions in the reports are written per part now, not as one headline number.
Thai pages, and what Google will and will not show
Two new Thai pages went live in the month — Thai massage and Thai ice bath — and both earned clicks straight away. Across four Thai pages, 11 of the 41 clicks came from 277 of the 1,980 appearances. That is a small sample, but they are converting better than the English pages.
By country, Thailand accounts for 1,204 of the 1,980 appearances and 25 of the 41 clicks, at an average position of 9.5.
A new source, at a small size. Visits from AI assistants went from 4 to 27, and 21 of the 27 were engaged. The report attributes them to ChatGPT and records one more from Perplexity. Twenty-seven is still a small number, but it is the fastest-growing source in the data.
A limit on all of this: Google hides any search too few people typed. Of the 41 clicks, only 6can be tied to a named search. The other 35 landed on pages but cannot be traced to what was typed. So the pages show what is working, and the searches show what Evergreen is visible for — and the reports keep the two apart.
Where it ranks, over three months
Searches that name a place improved: “rawai spa” went from 42.2 to 19.5, and “rawai massage” from 43.0 to 13.3. Searches that say “near me” went the other way. Each cell is the average position, with times shown in brackets.
| Search | Month 1 | Month 2 | Month 3 |
|---|---|---|---|
| rawai spa | 42.2 (51) | 32.6 (158) | 19.5 (73) |
| evergreen spa | 6.1 (15) | 5.9 (83) | 3.4 (35) |
| sauna rawai | 9.8 (13) | 8.3 (37) | 6.7 (39) |
| ice bath phuket | 23.0 (3) | 14.8 (19) | 13.9 (40) |
| rawai massage | 43.0 (1) | 24.6 (5) | 13.3 (11) |
| spa rawai | 51.7 (3) | 7.1 (8) | 11.6 (18) |
| ice bath near me | 3.3 (3) | 5.8 (23) | 8.2 (36) |
| wellness resort in rawai phuket | 53.6 (256) | 50.2 (321) | 52.2 (52) |
Still wrong
“Rawai spa” is still split.The home page leads it at position 22.2 on 61 appearances, but two other pages share 67 appearances at positions 41.9 and 41.1, and neither will win. The fix proposed last month — changing a page’s web address — is withdrawn: it cannot be undone, it risks the position the home page just won, and it treats a content problem as a plumbing problem. It is not being done in October, and the client is owed a written proposal with the options and what each one risks.
“Near me” searches are getting worse, not better. “Ice bath near me” went from position 3.3 (3 appearances) to 5.8 (23) to 8.2 (36): shown more often and ranked lower every month. For those searches Google puts the map listings above ordinary web results, so the website is not what serves them.
Two sources fell and the cause is not known. Referral visits went from 14 to 2 and social visits from 21 to 10. Nothing is being invented to explain that; it is being watched.
Nothing here proves a booking.The WhatsApp button was tapped 20 times, up from 12. What those taps became — a message, a booking, nothing — is still not visible, and 41 clicks off a base of 21 is still a small number. Three months is where patterns start to be real, not where they are proven.
07 · What comes next
Discoverable is working. Next is the part people see.
The constraint is moving from Discoverable towards Converting: more of the people who find Evergreen choosing it. October has four jobs and no new pages: building out the business’s Google profile, where most “near me” searches are decided; making the review count update itself; getting three directory listings corrected; and reworking the Thai sauna page.
That last one is a structural fix, not a wording one. The page used the Thai word for herbal steam throughout and never the word people actually type for sauna — so 93 appearances of Thai searches were landing on English pages. The reworked page went live in early October 2026 and was submitted to Google on 7 October. There is no result yet.It will be in next month’s report whichever way it goes.
Reported monthly, in this format, whether or not it went well. Month 1 led with what was not working. Month 3 led with a forecast that was wrong. A case study that only shows the wins is an advertisement.
Want to know which stage is holding you back?
That is what an assessment answers — with the same honesty about what is not working.
Start with Assess · $950