Social Listening Thailand: Search Pantip Only in English and You Miss 4 in 10 Topics

In one line: Thailand’s biggest homegrown site is a forum, Pantip, and it is searched in two scripts — on 11 October 2026 the Latin spelling of six brand names found only 60 of the 104 topics the Latin and Thai spellings found together, so a Thai listening setup that searches in English alone is missing about four topics in ten before it has read a single comment.

Where Thailand’s public conversation lives

Similarweb’s ranking of Thailand’s most-visited websites for September 2026 puts pantip.com at #11. Everything above it is a global platform, a marketplace, a messaging app or an AI assistant: Google, YouTube, Facebook, Shopee, Instagram, X, TikTok, LINE, ChatGPT and Gemini. Pantip is the first site on the list that was built in Thailand for Thais to talk to each other.

It is also a forum, not a feed. Topics are long-lived, people answer each other in nested replies, and a question about a car loan or a phone repair can keep drawing comments for weeks. That shape is exactly what makes it valuable for listening — and what makes it different from the platforms a listening tool is usually built around.

Do Thai listening tools already cover it?

Some do, and it is worth saying so plainly. Wisesight’s ZocialEye, a Thai-built suite, lists Pantip among its sources alongside Facebook, Twitter, Instagram, YouTube, TikTok and news sites. If what you need is a Thai dashboard, buying one is a sensible choice.

The case for pulling Pantip yourself is different:

• You want raw topics and comments in your own warehouse or LLM pipeline, not a dashboard export.
• You cover several markets and want Thailand in the same pipeline as the others.
• Your listening stack is a global suite, and you have not confirmed Pantip is in its source list. Check before assuming.

The rest of this guide is about that second path, using the pantip-scraper-en actor, and about three things we measured that change how you should set it up.

Finding 1: the same brand, two scripts, different topics

Thai posters often write a foreign brand in Thai script — โตโยต้า for Toyota, ช้อปปี้ for Shopee — and Pantip’s search treats the two spellings as different words. To see how much that matters, we searched six brands both ways on 11 October 2026 at 10:46 Bangkok time. Each search returns its first page, ten topics:

The two device brands overlapped heavily. The four service, car and e-commerce brands did not overlap at all. Across all six, the Latin spellings alone surfaced 60 of the 104 distinct topics.

The Toyota rows show what that means in practice. Two of the topics were about the same story — whether Toyota would move production from Thailand to Indonesia:

• โตโยต้าจะย้ายฐานการผลิตทั้งหมดจากไทยไปอินโดนีเซียจริงมั๊ยครับ (“Is Toyota really moving all its production from Thailand to Indonesia?”) — found only by the Thai search.
• Toyota ตอบชัดย้ายจากไทยทำไม่ได้ง่ายๆ (“Toyota makes it clear: leaving Thailand isn’t easy”) — found only by the Latin search.

A monitor with one spelling would have seen one side of the conversation. The fix costs nothing but a longer list: put both spellings, and any nickname people actually use, in keywords.

Finding 2: search is a sweep, not a live feed

Pantip’s search results are not sorted newest-first, and the actor reads the first page per keyword. So a search tells you what has been said about a brand, in whatever order Pantip ranks it — not what was posted this morning. Across our 120 search rows:

That does not make search useless for monitoring; it changes how you use it. Run it on a schedule with dateFrom set to your interval, so each run keeps only topics newer than the last one, and widen the keyword list instead of hoping new topics rank on page one. More variants per brand — both scripts, model names, campaign names — catch more of what is new.

{
  "mode": "search",
  "keywords": ["Toyota", "โตโยต้า", "Grab", "แกร็บ", "Shopee", "ช้อปปี้"],
  "dateFrom": "7d",
  "maxItems": 50
}

dateFrom takes a relative window (12h, 3d, 2w) or an absolute MM/DD/YY date. Each search row comes back with topic_id, title, author, author_id, comment_count, created_at, topic_url, source_query and scraped_at; content is null until you expand the topic.

Finding 3: wait a day before you expand a topic

topic mode fetches the main post and every comment, including nested replies, one row each:

{
  "mode": "topic",
  "topicIds": ["44210691", "44226228"],
  "maxItems": 300
}

Comment rows add comment_id, floor (nested replies are numbered {comment}-{reply}), is_reply, score and content_images. maxItems caps the comments fetched per topic, which is your cost ceiling on a thread that runs into the thousands.

Comments cost more than anything else, so when you expand matters. We fetched four Toyota topics posted in August and September 2026 and lined each comment up against its post time:

Of the 184 comments, 41% arrived within 6 hours, 70% within 24 hours and 80% within 72 hours. Expanding a topic in its first hour captures a fraction of the thread; expanding it after 24 to 72 hours captures most of it in one billed pass. The 24-day tail on the EV topic is the exception, and for a topic you care about that much, a second pass a few weeks later is cheap.

Two more things the same four topics taught us:

• The comment count on a search card is top-level comments only. The cards showed 48, 31, 60 and 17; the topics returned 63, 34, 67 and 20 rows once nested replies were counted. Budget about 1.2× the card number.
• Read the run log for [incomplete-data]. On our first run, one topic’s comment page was refused with HTTP 403 through every retry. The actor reported it as [incomplete-data] instead of returning an empty thread, and a rerun a few minutes later fetched all 67 rows.

A Thailand-specific detail: dates in the Buddhist calendar

Pantip shows dates in the Thai Buddhist calendar, which runs 543 years ahead: a post stamped 23 สิงหาคม 2569 was written on 23 August 2026. The actor converts every date to Gregorian ISO 8601, so you never handle the 2569 yourself.

One detail to know when you sort: search cards only show the day, so a search row’s created_at is Bangkok midnight expressed in UTC — 2026-07-27T17:00:00.000Z means 28 July in Thailand. Main posts and comments carry exact timestamps.

What a daily monitor costs

Three billing events, and the gap between them is the whole budgeting story:

Worked through for 10 brands, both spellings, once a day:

• Search: 20 keywords × ~10 topics = 200 rows × $0.002 = $0.40/day
• Expanding 10 topics: 10 × $0.008 = $0.08/day
• Their comments, at the 46 rows per topic our sample averaged: 460 × $0.003 = $1.38/day

About $1.86 a day, or roughly $56 a month — and three quarters of it is comments. Do your filtering in the cheap pass: use comment_count and the title to pick which topics deserve expansion, then expand them once, a day or two after they appear.

Thailand next to your other markets

If Thailand is one market among several, the Southeast Asia forums guide covers Pantip alongside VOZ, Vietnam’s biggest tech forum, which voz-scraper-en collects the same way. For Singapore, the EDMW monitoring guide shows the equivalent setup on HardwareZone. And for the downstream half — scheduling, scoring comments with an LLM and alerting — the n8n forum sentiment pipeline works with Pantip rows as the input.

Three ways to start

From least effort to most freedom

I run it for you Managed, daily to your sheet You do nothing. From US$299/mo. Use it here Run the free tool Instant, no setup, on this site. Run my actor On Apify, your account Same data, you run it and pay per result. Build your own Sign up for Apify Free account — build the pipeline yourself.

This post contains Apify affiliate links (no extra cost to you). Costs and comparisons are stated honestly, including where Apify is not the best fit.

FAQ

Don’t Thai social listening tools already cover Pantip? Some do. Wisesight’s ZocialEye, a Thai-built suite, lists Pantip among its sources alongside Facebook, Twitter, Instagram, YouTube, TikTok and news sites. If a dashboard is all you need, that is a reasonable buy. This guide is for the other case: you want the raw topics and comments in your own pipeline, next to other markets, or you run a global suite and are not sure Pantip is in its source list. Check that list before assuming.

Why search the Thai spelling if the brand name is in English? Because Thai posters often write the brand in Thai script, and Pantip’s search treats the two as different words. On 11 October 2026 we ran each of six brands both ways. Samsung and iPhone overlapped 8 out of 10 results, but Shopee, Toyota, Grab and Lazada shared no topic at all between the Latin and Thai searches. Across all six, the Latin names alone found 60 of the 104 distinct topics. Put both spellings, plus any common nickname, in the keywords list.

Can Pantip search be used as a real-time feed? Not on its own. Pantip’s search results are not sorted newest-first, and the actor reads the first page, about 10 topics per keyword. In our 120 search rows the median topic was 95 days old, only 21 were from the previous 30 days, and 2 from the previous 7. Use it as a scheduled sweep across many keyword variants with dateFrom set, so each run keeps only what is new, rather than expecting every new topic to rank on page one.

When should I fetch a topic’s comments? About a day after it appears, if you only want to pay once. Across four Toyota topics from August and September 2026, the first comment arrived 6 to 28 minutes after posting, 41% of the 184 comments came within 6 hours, 70% within 24 hours and 80% within 72 hours. One topic was still collecting comments 24 days later. Expanding at hour one captures a fraction of the thread; expanding after 24 to 72 hours captures most of it in a single billed pass.

What does a daily Thai brand monitor cost? Mostly the comments. Searching 10 brands in two spellings is 20 keywords and about 200 topic rows a day, $0.40 at $0.002 each. Expanding 10 topics a day costs $0.08 for the main posts at $0.008 each, and at the 46 comment rows our sample averaged per topic, another $1.38 for comments at $0.003 each. That is about $1.86 a day, or roughly $56 a month, with comments making up three quarters of it.

Chad runs 30+ published Apify actors, including the pantip-scraper-en used here. Every Pantip measurement above — the 120 search rows across 12 queries, the 104 distinct topics, the 184 comments and their timing — was pulled live on 11 October 2026 and can be re-run with the inputs shown.