Subtitles on Bangkok creator videos in 2026

Roughly 78 percent of a Bangkok creator video's first-week views on Reels and TikTok start with the sound off, and that number climbs past 85 percent for anything watched on the BTS between 07:30 and 09:30. If the first four seconds of the video do not carry meaning without audio, the swipe-away happens before the brand name has landed. Subtitles are the single cheapest lever a Bangkok brand has for holding attention, and yet the brief line about them is missing on more than half the campaigns we see land in escrow.

This is what subtitles actually cost, how Bangkok creators handle them in 2026, and the three specific brief clauses that keep the file the brand receives from being useless for paid amplification later.

The three ways subtitles land on a Bangkok creator video

A creator video ships subtitles in one of three modes, and confusing them is where the disputes start. Burned-in captions live inside the video pixels themselves, cannot be turned off, and travel with the file to whichever platform the brand uploads it to next. Soft platform captions are the ones Reels, TikTok, and YouTube auto-generate and let the viewer toggle; they do not survive a re-upload to a different platform. Sidecar SRT files are a separate .srt text file delivered with the video, uploadable to YouTube, LinkedIn, and any web player.

Most Bangkok creators default to burned-in for organic posts because that is what performs on their own handle. That default is wrong for a brand brief. If the brief includes any usage on paid media, a partnership ad, or cross-posting to the brand handle, burned-in captions in the wrong language or the wrong font kill the asset. We covered the paid side of this in the Partnership Ads guide and the cross-posting mechanics in the Reels cross-posting post; subtitles are the layer both those workflows silently assume.

Thai first, English second, or both

The default answer for a Bangkok SMB brief is Thai burned-in, English SRT sidecar. That covers the domestic organic audience (Thai on-screen, sound off, no toggle needed) and gives the brand a clean file to run English captions on for paid or LinkedIn use. It also matches how the top decile of the roster works without being asked.

Dual-language burned-in (Thai on top, English underneath, or vice versa) is the second most common option, and it is the one that looks tidy on the brief but performs the worst in the feed. Two rows of text at the bottom of a 9:16 frame cover roughly 22 percent of the vertical screen space, which is exactly where TikTok and Reels place their own UI. Something has to give, and it is usually the second language, rendered too small to read on a BTS phone at arm's length.

English-only burned-in is a specific choice for brands whose Bangkok audience skews expat, tourist, or Sukhumvit-corporate. Roots cafe, most co-working spaces, and the international-school-parent segment all live here. If the brand's Instagram bio is in English and the pinned Reels are English, matching the creator video to that is honest. Do not fake a Thai audience the brand does not have.

What subtitles actually cost in 2026, in THB

These are the ranges we see across the Mai Influence roster for a single 30 to 60 second Reel or TikTok, delivered as part of a standard brief, not as an add-on after the fact.

• Burned-in Thai only, creator-written: included in the base fee. Every working creator produces this by default.
• Burned-in English only, creator-written: +500 to 1,500 THB if the creator is bilingual, +2,000 to 4,000 THB if they need an outside translator.
• Dual burned-in (Thai + English): +2,000 to 5,000 THB. The uplift is the extra time in CapCut, not the translation.
• Sidecar SRT file in one language: +1,000 to 2,000 THB. Two languages, +2,500 to 4,500 THB.
• Brand-supplied font and colour for burned-in captions: +1,500 to 3,000 THB. This is a re-render fee; the creator has to redo the caption pass rather than use their template.
• Re-caption of an already delivered video: +3,000 to 6,000 THB, and only if the creator kept the project file, which loops back to the raw files post.

These are shoot-day-adjacent fees, not revision fees. Asking for a caption font change after final delivery is a revision, and revisions past the second round follow the approval SLA rules.

The auto-caption trap

Every Bangkok creator has a CapCut auto-caption workflow. Some of them ship those captions untouched. Auto-captions in Thai are roughly 82 to 89 percent accurate on clean studio audio, and drop to 60 to 70 percent the moment there is a market crowd, a BTS train, or a second speaker. The failure mode is not gibberish; it is a plausible-looking Thai sentence that means the wrong thing.

The specific number that matters: on a 45-second product-explainer Reel, auto-captions typically introduce one to three material errors in the brand claim itself. "ลด 50%" becomes "ลด 15%". "ไม่มีน้ำตาล" becomes "มีน้ำตาล". The brand approves the video on a phone at 22:00 without proofreading the captions in detail, and the misclaim ships to 40,000 viewers before anyone notices. The fix is a single brief line requiring human-proofed captions, delivered as text alongside the video for a second-pair-of-eyes check.

The second auto-caption failure is timing. Auto-generated caption blocks default to one to two seconds on screen, which is fine at TikTok scroll speed but too fast for anyone reading Thai as a second language. If the brief targets expat or regional SEA viewers, ask for three-second minimum caption blocks. That is a CapCut checkbox for the creator, not a rebuild.

What to write in the brief

Three lines cover 95 percent of the subtitle disputes we see. Copy them verbatim.

Captions: Thai burned-in on video, brand font (Sarabun Bold 32pt, white with 2px black stroke), three-second minimum block duration. Sidecar: English .srt file delivered alongside the master. Proofing: caption text delivered as a separate .txt file for brand sign-off before render.

That third line is the one most briefs miss. A .txt of the caption pass takes the creator two minutes to export and lets the brand catch a wrong percentage or a wrong ingredient before the video is locked. Skipping it is how the wrong Thai character ends up on 40,000 phones.

Write it into the brief, price it into the quote, and the bottom third of the screen stops being where the campaign quietly fails.