Can an AI presenter earn a second booking? Compare it with a phone video.
Newly disclosed evidence
Unresolved
Outcome updated
Historical benchmark context, not commercial forecast outcome
- What we originally said
Platform launch and maker study reported; no external independent blind trial or regulator clearance verified.
- What actually happened
We missed an externally operated blinded creator-preference comparison. New Measure's Avatar Arena reports 542 blind votes in the displayed personal-avatar track for 27 October to 16 November. HeyGen's 17 November technical report referenced the comparison before this Signal appeared.
- What we learned
The comparison narrows our original description of the available evidence. It does not independently reproduce HeyGen's realism percentage or establish viewer retention, customer payments or repeat work.
- What remains uncertain
The leaderboard's votes are not verified as distinct creators. Participants and short clips were self-selected; comparisons between rivals were indirect. New Measure's financial independence was not established, and the separately dated 18 November report has no verified publication hour.
- What we will check next
Check a consented comparison with the same script and audience against a low-cost phone video, then look for attributable retention and cleared repeat client payments within the original 30–90-day window.
- Supporting evidence
- New Measure Avatar Arena leaderboardEvent · Checked Independent evidence · Public accessEvidence limits Externally operated; funding independence not verified; personal-avatar track votes do not prove distinct voters.
- HeyGen Avatar 6 technical reportPublished · Event · Checked Primary source · Public accessEvidence limits Vendor-authored report references the Arena; this is not independent replication of its own realism figure.
Review overdue · no later published assessment recorded · Planned
HeyGen’s Avatar 6 promises more natural gestures. A small business can test that promise without treating views as sales or a presenter’s likeness as free to use.
Explore this signal
Emerging data
A second booking is a different result from a view
Imagine a bakery explaining how to collect a prepaid order. A phone recording and an AI-presented version can use the same script and offer. Count how many viewers return, how many orders are attributable to each version and whether any client commissions another video. A realistic gesture alone answers none of those questions.
View evidence links
| Record | What it proves | What it does not prove |
|---|---|---|
| Viewer retention | How long comparable audiences watched. | That a client paid. |
| Cleared first invoice | One client paid, not that the format is profitable. | That the client will book again. |
| Cleared repeat invoice | Repeat paid work; subtract costs to test margin. | That the work earned a profit after costs. |
What changed on 17 November
HeyGen announced Avatar 6 at 12:07 US Eastern on 17 November, or 01:07 Singapore time on 18 November. The company says the presenter’s gestures respond to speech. Its reported realism preference is a vendor study, not independent evidence that customers stay longer or buy more. A webinar scheduled for 19 November at 02:00 Singapore time had not happened by the research cutoff.
Permission comes before production
Use a consenting adult presenter or your own likeness and voice. Agree in writing on the script, language, audience, channel, permitted duration and how to stop future use. Do not upload customer recordings or imitate another person. Label the synthetic presentation; have a human check the spoken meaning and captions before release. YouTube’s disclosure rules concern its own service; they are not blanket permission to use a likeness.
- Written permission for face and voicePublish only the agreed script, channels and languages.
- Client and presenter approve intended useDisclose synthetic content and check captions.
What a fair comparison costs
Try one short approved tutorial against the phone version. Keep the offer, script, audience and comparison window alike; disclose any difference in placement. Count editing time, translation review, tools and revisions. A hypothetical US$300 job minus US$80 editing, US$35 tools, US$45 translation review and US$25 administration leaves US$115 before overhead and tax. No client payment was verified.
Original forecast • 18 November 2026, 13:05 SGT
Consented AI video could support small, repeat paid production services within 30–90 days, but only when real clients pay and retention beats a low-cost phone video. Recheck at 30 days and through day 90. Stop immediately if permission is unclear. If viewers stay but no client pays, the commercial test has not passed.
- 18 Nov, 01:07 SGTMaker announced the model.
- 19 Nov, 02:00 SGTScheduled webinar, after the 13:05 research cutoff.
- 30–90 daysLook for repeat cleared fees and retention against a comparable phone video.
01 Money Flow
Maker announced Avatar 6 access. Incremental subscriptions or actual creator earnings from it: NOT VERIFIED.
02 Orders/Deposits
No named paid video package contracts or first-customer deposits after launch independently observed.
03 Bottlenecks
Actor consent, voice rights, human translation checks, rendering cost, approval and truthful labelling.
04 Other Beneficiaries
Editors, captioners, translators and rights auditors may receive assignments; paid commissions not verified.
05 Listed Finances And Prices
HeyGen private, NOT publicly quoted. Issuer ARR claims are not audited current cash; company debt, margins, valuation and customer concentration NOT VERIFIED.
06 Job Economics
HYPOTHETICAL US$300 fee less US$185 direct costs = US$115 before overhead; no observed paid job or payback.
07 Insider/Institutional Activity
Private financing or ownership changes specifically tied to Avatar 6: NOT VERIFIED.
08 Hiring
No verified extra video localisation, rights or customer-support roles caused by this launch.
09 Public Procurement
No approved public video procurement tender or agency budget linked to Avatar 6.
10 Foreign-Language Warning
Launch U.S./English. Translation accuracy across other languages and local disclosure requirements NOT VERIFIED.
11 Behaviour And Cleared Payment
No controlled before/after conversion, completed sale or cleared client payment from the new model.
12 Contradictory Readings
Greater realism may increase audience interest or decrease trust; published benchmark is company-selected.
13 Observed Missing Activity
No independently published creator payment ledger in launch; do not infer no one earns money.
14 Commercial Pilots/Approval
Platform launch and maker study reported; no external independent blind trial or regulator clearance verified.
15 Competitive Weaknesses
Phone video, live presenter and competing AI tools can be cheaper, more credible or easier to license.
16 Cross-Border Effects
Multi-country content possible, but local likeness, ad disclosure and voice laws vary; no blanket international permission.
17 Prior Otr Outcomes
Learning page separates view/engagement from paid result. No dated matching OTR avatar-conversion win or loss verified.
18 Awareness And Pricing
Launch public and maker widely markets it; creator willingness-to-pay and any unpriced investment opportunity UNPROVEN.
19 Strongest Disconfirming Facts
Vendor's realism score is not independently reproduced; rights issues or zero conversion negate an earning claim.
20 Clock/Cost/Stop
Prepare one consented test within seven days; hypothetical US$35 tools + human labour. Horizon 30-90 days; stop if permissions invalid or no profitable paid conversion.
Sources and limits
The issuer release preceded the research cutoff by 11 hours 58 minutes. HeyGen’s policy and event listing serve different purposes but share its source family. YouTube explains disclosure on its platform; W3C explains accessible captions. Neither measures HeyGen’s quality or creator income. The OTR learning record is methodology only.
HeyGen • dated Avatar 6 announcement via PR Newswire
HeyGen • moderation and consent policy
HeyGen • scheduled launch webinar, not completed evidence
YouTube • disclosing altered or synthetic content
W3C • captions for accessible media
AI powered research