Malaysia TikTok Live reports
Five livestreams from 13 and 15 November 2026. Pick one to see its sales, audience and what ext-scribe captured.
All five livestreams side by side
Sales, audience and Host Score of each session. Colour shows where a session stands against the others in each column. Open a session for the full report.
Sessions
GMV is the TikTok Partner Center export, in RM. The Host Score and its pillars come from the host review and are estimates in part. Flags are claims to check plus silences over 30 seconds. Sessions without a comparable host review show a dash.
| Session | GMV (RM) | GMV vs average (%) | Orders | Views | GMV per hour (RM) | Host Score | Delivery | Selling | Engagement | Compliance | Questions answered (%) | Flags | Report |
|---|
How this report compares with Competitor A
—
| Review item | Area | Competitor A | This report | Status | Where |
|---|
Eucerin Malaysia, TikTok Live, 13 Nov 2026
The live brought in — of GMV from — paid orders and — customers, and reached — viewers.
Sales
Audience
Products
The result
How the score is builtEach pillar score is the mean of its measured parameters. The Host Score is the weighted mean of the pillar scores: Delivery 20%, Selling Quality 35%, Engagement 35%, Compliance 10%. A pillar with no measured parameter passes its weight to the others, so the share of weight actually measured is shown.
—
What to fix first
The three parameters that cost the most points, each with a line from the live. The sample line is an example of the kind of sentence to aim for, written for this report; the host did not say it.
Minute by minute
Bars show seconds of speech in each minute of the replay, red is silence over 5 seconds, dots are minutes with viewer comments. Hover for the clock time.
Answer rate
Share of viewer questions the host answered, answered in part, or did not answer. Judged by a language model.
Top unanswered questions
Questions the host did not answer, or only partly. Many of the same kind point to something to put in the brief.
| Time | Who | Question | Answered |
|---|
All parameters
Parameters that need a brand fact sheet, a must-say list or a run of show are left as not measured, as in the SHISEIDO review. AI marks a model estimate.
| ID | Pillar | Parameter | Value | Target | Score | Status | Note |
|---|
What viewers asked
Comments that ask the host something, and whether the host answered. Judged by a language model.
| Time | Who | Question | Answered | Seconds to answer |
|---|
Longest silences
| At | Seconds | Before | After |
|---|
Claims to check
Lines a language model flagged as possibly wrong or non-compliant. Without a brand fact sheet they are shown, not scored.
| At | Kind | Line |
|---|
Worth repeating
Across the session
Each five-minute window of the replay, in replay time.
| From | Speech (%) | Comments | Confidence | Tone | CTAs | Repetition (%) |
|---|
GMV every 15 minutesGMV (gross merchandise value) is the total value of the orders, in RM. TikTok gives one time for each 15-minute bucket and does not say whether it is the start or the end of the bucket.
—
More metrics
Best-selling products
The five products with the most GMV. The full list is further down.
| Product | GMV (RM) | GMV share (%) | Items | Impressions | CTR (%) |
|---|
Prime timeGMV added up by the hour of the day, in Malaysia time, from the 15-minute buckets. It shows which hours of the live earned the most.
GMV by hour of the day, Malaysia time.
Where the sales came fromImpressions: times TikTok showed the live to someone. Views: entries into the room. Enter-room rate: views divided by impressions. Conversion: TikTok's own rate for that source; Partner Center does not spell out the formula. Show GPM: GMV per 1,000 impressions.
By traffic source. Rates are percentages. Share of GMV is each source's part of the sum over all sources. An empty cell means TikTok gave no figure.
| Source | Impressions | Views | Enter-room rate (%) | Conversion (%) | GMV (RM) | Share of GMV (%) | Show GPM (RM) |
|---|
Product performanceImpressions, CTR, add to cart and GMV are TikTok's figures for the whole live, pinned and unpinned time together. GMV share and Impressions share: the product's part of the live's total, in percent. Pins: how many times the host pinned it. Minutes pinned: time from a pin to the next pin, or to the end of the capture. Share of captured time: minutes pinned divided by the time ext-scribe captured. Impressions per pinned minute: whole-live impressions divided by minutes pinned, so it is a rough guide that also counts impressions from unpinned time.
Figures are for the whole live, pinned and unpinned time together. Pin columns come from the pins ext-scribe captured. Click a heading to sort. Use the filter to see what sold, what did not, and what was pinned.
| Product | GMV (RM) | GMV share (%) | Items | Impressions | Impressions share (%) | CTR (%) | Add to cart | Pinned | Pins | Minutes pinned | Share of captured time (%) | Impressions per pinned minute | In stock |
|---|
Do the totals agree?Partner Center gives GMV in several places. They normally agree. When they do not, this table lists each total.
The same live's GMV as each Partner Center view gives it. The headline figure of this report is the Partner Center export.
| Where the total comes from | GMV (RM) | Share of the headline (%) |
|---|
Watch the live
About this link
- The video is stored privately in S3. The link carries a signature that lets whoever has it watch this one livestream for 7 days. Treat the link like a password and share it only with people who should see the video.
- After 7 days the link stops working. A fresh link has to be made, and this report updated.
- The page works in a normal browser. Use the Transcript menu to switch between the speech models, and tick or untick Speech, Comments and Pins to filter the list.
People watching now
Viewer counts sampled by ext-scribe while the live ran, Malaysia time.
Comments
| Time | Who | Comment |
|---|
Pins
ext-scribe records the product id; the names come from the live's product list in Partner Center.
| Time | Product |
|---|
Joins
| Time | Who joined |
|---|
How good is each transcript?How sure Whisper was about the words, from its own score for each segment, averaged over the speaking time. It is the model's self-rating, not accuracy.
Per model, over the parts transcribed so far. There is no human-corrected text, so word error rate is not measured. These are the checks Whisper's own scores allow.
| Model | Parts | Audio (h) | Words per min | Coverage (%) | Confidence (%) | Low confidence (%) | Suspect segments | Invented phrases |
|---|
Words per minuteWords per minute of audio, counting the whole part. Very different values between models mean one of them dropped or added speech. CoverageShare of the audio with words in it, as a share of the audio length. Low confidenceShare of the speaking time where Whisper's own score was low (below the cut-off the bilabs checks use). Higher means more words to doubt. Suspect segmentsSegments that look wrong: Whisper heard speech in silence, a loop of repeated words, a phrase it tends to invent (such as thanks for watching) or the same line three times.
Do the models agree?How far two models are apart: the word changes (swaps, drops, additions) needed to turn one transcript into the other, divided by the words of the first. It is the same word alignment as a word error rate, but between two models, so it shows disagreement and not who is right.
Measured in one-minute windows, with the same text clean-up and word alignment as the bilabs word error rate tool.
| Model | Compared with | Parts | Words | Word changes | Disagreement (%) |
|---|
Where the models differ most
The twelve one-minute windows with the most word changes for each pair of models. Read both columns to judge which model is right. Times are Malaysia time.
| Time | Models | Changes (%) | First model | Second model |
|---|
How to read these figures
- The headline figures come from the TikTok Partner Center live-analysis export pulled on 16 Nov 2026; the Sales tab comes from the Live Data Screen of Partner Center, pulled the same day. TikTok can still change a live's numbers while it settles them.
- Money is in Malaysian ringgit (RM). Attributed GMV is what TikTok credits to the live, direct sales plus sales that came afterwards. Direct GMV is what was sold inside the live.
- GMV per hour, GMV per viewer and average order value are worked out from the headline GMV, so they move with it.
- All times are Malaysia time (UTC+8). ext-scribe records the computer's clock, which was converted.
- Comments, joins, pins and viewer counts are what ext-scribe saw while the live was open on the capturing computer. Anything it recorded before the live started or after it ended is left out.
What this report does not have yet
- The host review leans on estimates by a language model and leaves out what needs a brand fact sheet, a must-say list or a run of show.
- Names of viewers who commented or joined are shown as ext-scribe recorded them, and everyone who can open this report can read them.
LorealParisMalaysia, TikTok Live, 15 Nov 2026
The live brought in — of GMV from — paid orders and — customers, and reached — viewers.
Sales
Audience
Products
The result
How the score is builtEach pillar score is the mean of its measured parameters. The Host Score is the weighted mean of the pillar scores: Delivery 20%, Selling Quality 35%, Engagement 35%, Compliance 10%. A pillar with no measured parameter passes its weight to the others, so the share of weight actually measured is shown.
—
What to fix first
The three parameters that cost the most points, each with a line from the live. The sample line is an example of the kind of sentence to aim for, written for this report; the host did not say it.
Minute by minute
Bars show seconds of speech in each minute of the replay, red is silence over 5 seconds, dots are minutes with viewer comments. Hover for the clock time.
Answer rate
Share of viewer questions the host answered, answered in part, or did not answer. Judged by a language model.
Top unanswered questions
Questions the host did not answer, or only partly. Many of the same kind point to something to put in the brief.
| Time | Who | Question | Answered |
|---|
All parameters
Parameters that need a brand fact sheet, a must-say list or a run of show are left as not measured, as in the SHISEIDO review. AI marks a model estimate.
| ID | Pillar | Parameter | Value | Target | Score | Status | Note |
|---|
What viewers asked
Comments that ask the host something, and whether the host answered. Judged by a language model.
| Time | Who | Question | Answered | Seconds to answer |
|---|
Longest silences
| At | Seconds | Before | After |
|---|
Claims to check
Lines a language model flagged as possibly wrong or non-compliant. Without a brand fact sheet they are shown, not scored.
| At | Kind | Line |
|---|
Worth repeating
Across the session
Each five-minute window of the replay, in replay time.
| From | Speech (%) | Comments | Confidence | Tone | CTAs | Repetition (%) |
|---|
GMV every 15 minutesGMV (gross merchandise value) is the total value of the orders, in RM. TikTok gives one time for each 15-minute bucket and does not say whether it is the start or the end of the bucket.
—
More metrics
Best-selling products
The five products with the most GMV. The full list is further down.
| Product | GMV (RM) | GMV share (%) | Items | Impressions | CTR (%) |
|---|
Prime timeGMV added up by the hour of the day, in Malaysia time, from the 15-minute buckets. It shows which hours of the live earned the most.
GMV by hour of the day, Malaysia time.
Where the sales came fromImpressions: times TikTok showed the live to someone. Views: entries into the room. Enter-room rate: views divided by impressions. Conversion: TikTok's own rate for that source; Partner Center does not spell out the formula. Show GPM: GMV per 1,000 impressions.
By traffic source. Rates are percentages. Share of GMV is each source's part of the sum over all sources. An empty cell means TikTok gave no figure.
| Source | Impressions | Views | Enter-room rate (%) | Conversion (%) | GMV (RM) | Share of GMV (%) | Show GPM (RM) |
|---|
Product performanceImpressions, CTR, add to cart and GMV are TikTok's figures for the whole live, pinned and unpinned time together. GMV share and Impressions share: the product's part of the live's total, in percent. Pins: how many times the host pinned it. Minutes pinned: time from a pin to the next pin, or to the end of the capture. Share of captured time: minutes pinned divided by the time ext-scribe captured. Impressions per pinned minute: whole-live impressions divided by minutes pinned, so it is a rough guide that also counts impressions from unpinned time.
Figures are for the whole live, pinned and unpinned time together. Pin columns come from the pins ext-scribe captured. Click a heading to sort. Use the filter to see what sold, what did not, and what was pinned.
| Product | GMV (RM) | GMV share (%) | Items | Impressions | Impressions share (%) | CTR (%) | Add to cart | Pinned | Pins | Minutes pinned | Share of captured time (%) | Impressions per pinned minute | In stock |
|---|
Do the totals agree?Partner Center gives GMV in several places. They normally agree. When they do not, this table lists each total.
The same live's GMV as each Partner Center view gives it. The headline figure of this report is the Partner Center export.
| Where the total comes from | GMV (RM) | Share of the headline (%) |
|---|
Watch the live
About this link
- The video is stored privately in S3. The link carries a signature that lets whoever has it watch this one livestream for 7 days. Treat the link like a password and share it only with people who should see the video.
- After 7 days the link stops working. A fresh link has to be made, and this report updated.
- The page works in a normal browser. Use the Transcript menu to switch between the speech models, and tick or untick Speech, Comments and Pins to filter the list.
People watching now
Viewer counts sampled by ext-scribe while the live ran, Malaysia time.
Comments
| Time | Who | Comment |
|---|
Pins
ext-scribe records the product id; the names come from the live's product list in Partner Center.
| Time | Product |
|---|
Joins
| Time | Who joined |
|---|
How good is each transcript?How sure Whisper was about the words, from its own score for each segment, averaged over the speaking time. It is the model's self-rating, not accuracy.
Per model, over the parts transcribed so far. There is no human-corrected text, so word error rate is not measured. These are the checks Whisper's own scores allow.
| Model | Parts | Audio (h) | Words per min | Coverage (%) | Confidence (%) | Low confidence (%) | Suspect segments | Invented phrases |
|---|
Words per minuteWords per minute of audio, counting the whole part. Very different values between models mean one of them dropped or added speech. CoverageShare of the audio with words in it, as a share of the audio length. Low confidenceShare of the speaking time where Whisper's own score was low (below the cut-off the bilabs checks use). Higher means more words to doubt. Suspect segmentsSegments that look wrong: Whisper heard speech in silence, a loop of repeated words, a phrase it tends to invent (such as thanks for watching) or the same line three times.
Do the models agree?How far two models are apart: the word changes (swaps, drops, additions) needed to turn one transcript into the other, divided by the words of the first. It is the same word alignment as a word error rate, but between two models, so it shows disagreement and not who is right.
Measured in one-minute windows, with the same text clean-up and word alignment as the bilabs word error rate tool.
| Model | Compared with | Parts | Words | Word changes | Disagreement (%) |
|---|
Where the models differ most
The twelve one-minute windows with the most word changes for each pair of models. Read both columns to judge which model is right. Times are Malaysia time.
| Time | Models | Changes (%) | First model | Second model |
|---|
How to read these figures
- The headline figures come from the TikTok Partner Center live-analysis export pulled on 16 Nov 2026; the Sales tab comes from the Live Data Screen of Partner Center, pulled the same day. TikTok can still change a live's numbers while it settles them.
- Money is in Malaysian ringgit (RM). Attributed GMV is what TikTok credits to the live, direct sales plus sales that came afterwards. Direct GMV is what was sold inside the live.
- GMV per hour, GMV per viewer and average order value are worked out from the headline GMV, so they move with it.
- All times are Malaysia time (UTC+8). ext-scribe records the computer's clock, which was converted.
- Comments, joins, pins and viewer counts are what ext-scribe saw while the live was open on the capturing computer. Anything it recorded before the live started or after it ended is left out.
What this report does not have yet
- The host review leans on estimates by a language model and leaves out what needs a brand fact sheet, a must-say list or a run of show.
- Names of viewers who commented or joined are shown as ext-scribe recorded them, and everyone who can open this report can read them.
maccosmeticsmalaysia, TikTok Live, 15 Nov 2026
The live brought in — of GMV from — paid orders and — customers, and reached — viewers.
Sales
Audience
Products
The result
How the score is builtEach pillar score is the mean of its measured parameters. The Host Score is the weighted mean of the pillar scores: Delivery 20%, Selling Quality 35%, Engagement 35%, Compliance 10%. A pillar with no measured parameter passes its weight to the others, so the share of weight actually measured is shown.
—
What to fix first
The three parameters that cost the most points, each with a line from the live. The sample line is an example of the kind of sentence to aim for, written for this report; the host did not say it.
Minute by minute
Bars show seconds of speech in each minute of the replay, red is silence over 5 seconds, dots are minutes with viewer comments. Hover for the clock time.
Answer rate
Share of viewer questions the host answered, answered in part, or did not answer. Judged by a language model.
Top unanswered questions
Questions the host did not answer, or only partly. Many of the same kind point to something to put in the brief.
| Time | Who | Question | Answered |
|---|
All parameters
Parameters that need a brand fact sheet, a must-say list or a run of show are left as not measured, as in the SHISEIDO review. AI marks a model estimate.
| ID | Pillar | Parameter | Value | Target | Score | Status | Note |
|---|
What viewers asked
Comments that ask the host something, and whether the host answered. Judged by a language model.
| Time | Who | Question | Answered | Seconds to answer |
|---|
Longest silences
| At | Seconds | Before | After |
|---|
Claims to check
Lines a language model flagged as possibly wrong or non-compliant. Without a brand fact sheet they are shown, not scored.
| At | Kind | Line |
|---|
Worth repeating
Across the session
Each five-minute window of the replay, in replay time.
| From | Speech (%) | Comments | Confidence | Tone | CTAs | Repetition (%) |
|---|
GMV every 15 minutesGMV (gross merchandise value) is the total value of the orders, in RM. TikTok gives one time for each 15-minute bucket and does not say whether it is the start or the end of the bucket.
—
More metrics
Best-selling products
The five products with the most GMV. The full list is further down.
| Product | GMV (RM) | GMV share (%) | Items | Impressions | CTR (%) |
|---|
Prime timeGMV added up by the hour of the day, in Malaysia time, from the 15-minute buckets. It shows which hours of the live earned the most.
GMV by hour of the day, Malaysia time.
Where the sales came fromImpressions: times TikTok showed the live to someone. Views: entries into the room. Enter-room rate: views divided by impressions. Conversion: TikTok's own rate for that source; Partner Center does not spell out the formula. Show GPM: GMV per 1,000 impressions.
By traffic source. Rates are percentages. Share of GMV is each source's part of the sum over all sources. An empty cell means TikTok gave no figure.
| Source | Impressions | Views | Enter-room rate (%) | Conversion (%) | GMV (RM) | Share of GMV (%) | Show GPM (RM) |
|---|
Product performanceImpressions, CTR, add to cart and GMV are TikTok's figures for the whole live, pinned and unpinned time together. GMV share and Impressions share: the product's part of the live's total, in percent. Pins: how many times the host pinned it. Minutes pinned: time from a pin to the next pin, or to the end of the capture. Share of captured time: minutes pinned divided by the time ext-scribe captured. Impressions per pinned minute: whole-live impressions divided by minutes pinned, so it is a rough guide that also counts impressions from unpinned time.
Figures are for the whole live, pinned and unpinned time together. Pin columns come from the pins ext-scribe captured. Click a heading to sort. Use the filter to see what sold, what did not, and what was pinned.
| Product | GMV (RM) | GMV share (%) | Items | Impressions | Impressions share (%) | CTR (%) | Add to cart | Pinned | Pins | Minutes pinned | Share of captured time (%) | Impressions per pinned minute | In stock |
|---|
Do the totals agree?Partner Center gives GMV in several places. They normally agree. When they do not, this table lists each total.
The same live's GMV as each Partner Center view gives it. The headline figure of this report is the Partner Center export.
| Where the total comes from | GMV (RM) | Share of the headline (%) |
|---|
Watch the live
About this link
- The video is stored privately in S3. The link carries a signature that lets whoever has it watch this one livestream for 7 days. Treat the link like a password and share it only with people who should see the video.
- After 7 days the link stops working. A fresh link has to be made, and this report updated.
- The page works in a normal browser. Use the Transcript menu to switch between the speech models, and tick or untick Speech, Comments and Pins to filter the list.
People watching now
Viewer counts sampled by ext-scribe while the live ran, Malaysia time.
Comments
| Time | Who | Comment |
|---|
Pins
ext-scribe records the product id; the names come from the live's product list in Partner Center.
| Time | Product |
|---|
Joins
| Time | Who joined |
|---|
How good is each transcript?How sure Whisper was about the words, from its own score for each segment, averaged over the speaking time. It is the model's self-rating, not accuracy.
Per model, over the parts transcribed so far. There is no human-corrected text, so word error rate is not measured. These are the checks Whisper's own scores allow.
| Model | Parts | Audio (h) | Words per min | Coverage (%) | Confidence (%) | Low confidence (%) | Suspect segments | Invented phrases |
|---|
Words per minuteWords per minute of audio, counting the whole part. Very different values between models mean one of them dropped or added speech. CoverageShare of the audio with words in it, as a share of the audio length. Low confidenceShare of the speaking time where Whisper's own score was low (below the cut-off the bilabs checks use). Higher means more words to doubt. Suspect segmentsSegments that look wrong: Whisper heard speech in silence, a loop of repeated words, a phrase it tends to invent (such as thanks for watching) or the same line three times.
Do the models agree?How far two models are apart: the word changes (swaps, drops, additions) needed to turn one transcript into the other, divided by the words of the first. It is the same word alignment as a word error rate, but between two models, so it shows disagreement and not who is right.
Measured in one-minute windows, with the same text clean-up and word alignment as the bilabs word error rate tool.
| Model | Compared with | Parts | Words | Word changes | Disagreement (%) |
|---|
Where the models differ most
The twelve one-minute windows with the most word changes for each pair of models. Read both columns to judge which model is right. Times are Malaysia time.
| Time | Models | Changes (%) | First model | Second model |
|---|
How to read these figures
- The headline figures come from the TikTok Partner Center live-analysis export pulled on 16 Nov 2026; the Sales tab comes from the Live Data Screen of Partner Center, pulled the same day. TikTok can still change a live's numbers while it settles them.
- Money is in Malaysian ringgit (RM). Attributed GMV is what TikTok credits to the live, direct sales plus sales that came afterwards. Direct GMV is what was sold inside the live.
- GMV per hour, GMV per viewer and average order value are worked out from the headline GMV, so they move with it.
- All times are Malaysia time (UTC+8). ext-scribe records the computer's clock, which was converted.
- Comments, joins, pins and viewer counts are what ext-scribe saw while the live was open on the capturing computer. Anything it recorded before the live started or after it ended is left out.
What this report does not have yet
- The host review leans on estimates by a language model and leaves out what needs a brand fact sheet, a must-say list or a run of show.
- Names of viewers who commented or joined are shown as ext-scribe recorded them, and everyone who can open this report can read them.
NIVEA Malaysia, TikTok Live, 13 Nov 2026
The live brought in — of GMV from — paid orders and — customers, and reached — viewers.
Sales
Audience
Products
The result
How the score is builtEach pillar score is the mean of its measured parameters. The Host Score is the weighted mean of the pillar scores: Delivery 20%, Selling Quality 35%, Engagement 35%, Compliance 10%. A pillar with no measured parameter passes its weight to the others, so the share of weight actually measured is shown.
—
What to fix first
The three parameters that cost the most points, each with a line from the live. The sample line is an example of the kind of sentence to aim for, written for this report; the host did not say it.
Minute by minute
Bars show seconds of speech in each minute of the replay, red is silence over 5 seconds, dots are minutes with viewer comments. Hover for the clock time.
Answer rate
Share of viewer questions the host answered, answered in part, or did not answer. Judged by a language model.
Top unanswered questions
Questions the host did not answer, or only partly. Many of the same kind point to something to put in the brief.
| Time | Who | Question | Answered |
|---|
All parameters
Parameters that need a brand fact sheet, a must-say list or a run of show are left as not measured, as in the SHISEIDO review. AI marks a model estimate.
| ID | Pillar | Parameter | Value | Target | Score | Status | Note |
|---|
What viewers asked
Comments that ask the host something, and whether the host answered. Judged by a language model.
| Time | Who | Question | Answered | Seconds to answer |
|---|
Longest silences
| At | Seconds | Before | After |
|---|
Claims to check
Lines a language model flagged as possibly wrong or non-compliant. Without a brand fact sheet they are shown, not scored.
| At | Kind | Line |
|---|
Worth repeating
Across the session
Each five-minute window of the replay, in replay time.
| From | Speech (%) | Comments | Confidence | Tone | CTAs | Repetition (%) |
|---|
GMV every 15 minutesGMV (gross merchandise value) is the total value of the orders, in RM. TikTok gives one time for each 15-minute bucket and does not say whether it is the start or the end of the bucket.
—
More metrics
Best-selling products
The five products with the most GMV. The full list is further down.
| Product | GMV (RM) | GMV share (%) | Items | Impressions | CTR (%) |
|---|
Prime timeGMV added up by the hour of the day, in Malaysia time, from the 15-minute buckets. It shows which hours of the live earned the most.
GMV by hour of the day, Malaysia time.
Where the sales came fromImpressions: times TikTok showed the live to someone. Views: entries into the room. Enter-room rate: views divided by impressions. Conversion: TikTok's own rate for that source; Partner Center does not spell out the formula. Show GPM: GMV per 1,000 impressions.
By traffic source. Rates are percentages. Share of GMV is each source's part of the sum over all sources. An empty cell means TikTok gave no figure.
| Source | Impressions | Views | Enter-room rate (%) | Conversion (%) | GMV (RM) | Share of GMV (%) | Show GPM (RM) |
|---|
Product performanceImpressions, CTR, add to cart and GMV are TikTok's figures for the whole live, pinned and unpinned time together. GMV share and Impressions share: the product's part of the live's total, in percent. Pins: how many times the host pinned it. Minutes pinned: time from a pin to the next pin, or to the end of the capture. Share of captured time: minutes pinned divided by the time ext-scribe captured. Impressions per pinned minute: whole-live impressions divided by minutes pinned, so it is a rough guide that also counts impressions from unpinned time.
Figures are for the whole live, pinned and unpinned time together. Pin columns come from the pins ext-scribe captured. Click a heading to sort. Use the filter to see what sold, what did not, and what was pinned.
| Product | GMV (RM) | GMV share (%) | Items | Impressions | Impressions share (%) | CTR (%) | Add to cart | Pinned | Pins | Minutes pinned | Share of captured time (%) | Impressions per pinned minute | In stock |
|---|
Do the totals agree?Partner Center gives GMV in several places. They normally agree. When they do not, this table lists each total.
The same live's GMV as each Partner Center view gives it. The headline figure of this report is the Partner Center export.
| Where the total comes from | GMV (RM) | Share of the headline (%) |
|---|
Watch the live
About this link
- The video is stored privately in S3. The link carries a signature that lets whoever has it watch this one livestream for 7 days. Treat the link like a password and share it only with people who should see the video.
- After 7 days the link stops working. A fresh link has to be made, and this report updated.
- The page works in a normal browser. Use the Transcript menu to switch between the speech models, and tick or untick Speech, Comments and Pins to filter the list.
People watching now
Viewer counts sampled by ext-scribe while the live ran, Malaysia time.
Comments
| Time | Who | Comment |
|---|
Pins
ext-scribe records the product id; the names come from the live's product list in Partner Center.
| Time | Product |
|---|
Joins
| Time | Who joined |
|---|
How good is each transcript?How sure Whisper was about the words, from its own score for each segment, averaged over the speaking time. It is the model's self-rating, not accuracy.
Per model, over the parts transcribed so far. There is no human-corrected text, so word error rate is not measured. These are the checks Whisper's own scores allow.
| Model | Parts | Audio (h) | Words per min | Coverage (%) | Confidence (%) | Low confidence (%) | Suspect segments | Invented phrases |
|---|
Words per minuteWords per minute of audio, counting the whole part. Very different values between models mean one of them dropped or added speech. CoverageShare of the audio with words in it, as a share of the audio length. Low confidenceShare of the speaking time where Whisper's own score was low (below the cut-off the bilabs checks use). Higher means more words to doubt. Suspect segmentsSegments that look wrong: Whisper heard speech in silence, a loop of repeated words, a phrase it tends to invent (such as thanks for watching) or the same line three times.
Do the models agree?How far two models are apart: the word changes (swaps, drops, additions) needed to turn one transcript into the other, divided by the words of the first. It is the same word alignment as a word error rate, but between two models, so it shows disagreement and not who is right.
Measured in one-minute windows, with the same text clean-up and word alignment as the bilabs word error rate tool.
| Model | Compared with | Parts | Words | Word changes | Disagreement (%) |
|---|
Where the models differ most
The twelve one-minute windows with the most word changes for each pair of models. Read both columns to judge which model is right. Times are Malaysia time.
| Time | Models | Changes (%) | First model | Second model |
|---|
How to read these figures
- The headline figures come from the TikTok Partner Center live-analysis export pulled on 16 Nov 2026; the Sales tab comes from the Live Data Screen of Partner Center, pulled the same day. TikTok can still change a live's numbers while it settles them.
- Money is in Malaysian ringgit (RM). Attributed GMV is what TikTok credits to the live, direct sales plus sales that came afterwards. Direct GMV is what was sold inside the live.
- GMV per hour, GMV per viewer and average order value are worked out from the headline GMV, so they move with it.
- All times are Malaysia time (UTC+8). ext-scribe records the computer's clock, which was converted.
- Comments, joins, pins and viewer counts are what ext-scribe saw while the live was open on the capturing computer. Anything it recorded before the live started or after it ended is left out.
What this report does not have yet
- The host review leans on estimates by a language model and leaves out what needs a brand fact sheet, a must-say list or a run of show.
- Names of viewers who commented or joined are shown as ext-scribe recorded them, and everyone who can open this report can read them.
Maybelline Malaysia, TikTok Live, 13 Nov 2026
The live brought in — of GMV from — paid orders and — customers, and reached — viewers.
Sales
Audience
Products
The result
How the score is builtEach pillar score is the mean of its measured parameters. The Host Score is the weighted mean of the pillar scores: Delivery 20%, Selling Quality 35%, Engagement 35%, Compliance 10%. A pillar with no measured parameter passes its weight to the others, so the share of weight actually measured is shown.
—
What to fix first
The three parameters that cost the most points, each with a line from the live. The sample line is an example of the kind of sentence to aim for, written for this report; the host did not say it.
Minute by minute
Bars show seconds of speech in each minute of the replay, red is silence over 5 seconds, dots are minutes with viewer comments. Hover for the clock time.
Answer rate
Share of viewer questions the host answered, answered in part, or did not answer. Judged by a language model.
Top unanswered questions
Questions the host did not answer, or only partly. Many of the same kind point to something to put in the brief.
| Time | Who | Question | Answered |
|---|
All parameters
Parameters that need a brand fact sheet, a must-say list or a run of show are left as not measured, as in the SHISEIDO review. AI marks a model estimate.
| ID | Pillar | Parameter | Value | Target | Score | Status | Note |
|---|
What viewers asked
Comments that ask the host something, and whether the host answered. Judged by a language model.
| Time | Who | Question | Answered | Seconds to answer |
|---|
Longest silences
| At | Seconds | Before | After |
|---|
Claims to check
Lines a language model flagged as possibly wrong or non-compliant. Without a brand fact sheet they are shown, not scored.
| At | Kind | Line |
|---|
Worth repeating
Across the session
Each five-minute window of the replay, in replay time.
| From | Speech (%) | Comments | Confidence | Tone | CTAs | Repetition (%) |
|---|
GMV every 15 minutesGMV (gross merchandise value) is the total value of the orders, in RM. TikTok gives one time for each 15-minute bucket and does not say whether it is the start or the end of the bucket.
—
More metrics
Best-selling products
The five products with the most GMV. The full list is further down.
| Product | GMV (RM) | GMV share (%) | Items | Impressions | CTR (%) |
|---|
Prime timeGMV added up by the hour of the day, in Malaysia time, from the 15-minute buckets. It shows which hours of the live earned the most.
GMV by hour of the day, Malaysia time.
Where the sales came fromImpressions: times TikTok showed the live to someone. Views: entries into the room. Enter-room rate: views divided by impressions. Conversion: TikTok's own rate for that source; Partner Center does not spell out the formula. Show GPM: GMV per 1,000 impressions.
By traffic source. Rates are percentages. Share of GMV is each source's part of the sum over all sources. An empty cell means TikTok gave no figure.
| Source | Impressions | Views | Enter-room rate (%) | Conversion (%) | GMV (RM) | Share of GMV (%) | Show GPM (RM) |
|---|
Product performanceImpressions, CTR, add to cart and GMV are TikTok's figures for the whole live, pinned and unpinned time together. GMV share and Impressions share: the product's part of the live's total, in percent. Pins: how many times the host pinned it. Minutes pinned: time from a pin to the next pin, or to the end of the capture. Share of captured time: minutes pinned divided by the time ext-scribe captured. Impressions per pinned minute: whole-live impressions divided by minutes pinned, so it is a rough guide that also counts impressions from unpinned time.
Figures are for the whole live, pinned and unpinned time together. Pin columns come from the pins ext-scribe captured. Click a heading to sort. Use the filter to see what sold, what did not, and what was pinned.
| Product | GMV (RM) | GMV share (%) | Items | Impressions | Impressions share (%) | CTR (%) | Add to cart | Pinned | Pins | Minutes pinned | Share of captured time (%) | Impressions per pinned minute | In stock |
|---|
Do the totals agree?Partner Center gives GMV in several places. They normally agree. When they do not, this table lists each total.
The same live's GMV as each Partner Center view gives it. The headline figure of this report is the Partner Center export.
| Where the total comes from | GMV (RM) | Share of the headline (%) |
|---|
Watch the live
About this link
- The video is stored privately in S3. The link carries a signature that lets whoever has it watch this one livestream for 7 days. Treat the link like a password and share it only with people who should see the video.
- After 7 days the link stops working. A fresh link has to be made, and this report updated.
- The page works in a normal browser. Use the Transcript menu to switch between the speech models, and tick or untick Speech, Comments and Pins to filter the list.
People watching now
Viewer counts sampled by ext-scribe while the live ran, Malaysia time.
Comments
| Time | Who | Comment |
|---|
Pins
ext-scribe records the product id; the names come from the live's product list in Partner Center.
| Time | Product |
|---|
Joins
| Time | Who joined |
|---|
How good is each transcript?How sure Whisper was about the words, from its own score for each segment, averaged over the speaking time. It is the model's self-rating, not accuracy.
Per model, over the parts transcribed so far. There is no human-corrected text, so word error rate is not measured. These are the checks Whisper's own scores allow.
| Model | Parts | Audio (h) | Words per min | Coverage (%) | Confidence (%) | Low confidence (%) | Suspect segments | Invented phrases |
|---|
Words per minuteWords per minute of audio, counting the whole part. Very different values between models mean one of them dropped or added speech. CoverageShare of the audio with words in it, as a share of the audio length. Low confidenceShare of the speaking time where Whisper's own score was low (below the cut-off the bilabs checks use). Higher means more words to doubt. Suspect segmentsSegments that look wrong: Whisper heard speech in silence, a loop of repeated words, a phrase it tends to invent (such as thanks for watching) or the same line three times.
Do the models agree?How far two models are apart: the word changes (swaps, drops, additions) needed to turn one transcript into the other, divided by the words of the first. It is the same word alignment as a word error rate, but between two models, so it shows disagreement and not who is right.
Measured in one-minute windows, with the same text clean-up and word alignment as the bilabs word error rate tool.
| Model | Compared with | Parts | Words | Word changes | Disagreement (%) |
|---|
Where the models differ most
The twelve one-minute windows with the most word changes for each pair of models. Read both columns to judge which model is right. Times are Malaysia time.
| Time | Models | Changes (%) | First model | Second model |
|---|
How to read these figures
- The headline figures come from the TikTok Partner Center live-analysis export pulled on 16 Nov 2026; the Sales tab comes from the Live Data Screen of Partner Center, pulled the same day. TikTok can still change a live's numbers while it settles them.
- Money is in Malaysian ringgit (RM). Attributed GMV is what TikTok credits to the live, direct sales plus sales that came afterwards. Direct GMV is what was sold inside the live.
- GMV per hour, GMV per viewer and average order value are worked out from the headline GMV, so they move with it.
- All times are Malaysia time (UTC+8). ext-scribe records the computer's clock, which was converted.
- Comments, joins, pins and viewer counts are what ext-scribe saw while the live was open on the capturing computer. Anything it recorded before the live started or after it ended is left out.
What this report does not have yet
- The host review leans on estimates by a language model and leaves out what needs a brand fact sheet, a must-say list or a run of show.
- Names of viewers who commented or joined are shown as ext-scribe recorded them, and everyone who can open this report can read them.