Voice and an avatar lifted shopping-assistant ratings—but what caused the effect?
In a simulated store, 312 participants rated a text-voice-avatar assistant more highly than a text-only version. Because the experiment changed several features at once, it cannot say which feature mattered or whether sales would change.
The short version
3 key points
- What changed
- The peer-reviewed online experiment recruited 320 people and retained 312 after screening on a simulated e-commerce platform.
- Why it matters
- Shoppers may find richer assistants more engaging, but engagement is not proof of better-value or less pressured decisions.
- What to watch
- Factorial trials that isolate voice, avatar, animation and synchronisation
A strong rating difference in a simulated store
A peer-reviewed experiment reports that people rated a shopping assistant with text, voice and an animated avatar more positively than a text-only version. Researchers recruited 320 participants through Prolific and retained 312 after screening. Participants were assigned between the two interface configurations on a simulated e-commerce platform and completed search, product-evaluation and post-purchase stages.[1]
Who it affects
Shoppers may find richer assistants more engaging, but engagement is not proof of better-value or less pressured decisions.[1]
Keep in mind
The two conditions differ on several design dimensions, preventing attribution to voice, avatar or synchronisation alone.
The platform and journey were simulated; no real purchase, return, spending or retention outcome was measured.
The accepted article is citable and peer reviewed but still awaiting final production edits.
Source analysis · 1 linked source. Original source date: 19 October 2026.
Read the full report · 3 min The evidence, methods, implications and limitations
A strong rating difference in a simulated store
A peer-reviewed experiment reports that people rated a shopping assistant with text, voice and an animated avatar more positively than a text-only version. Researchers recruited 320 participants through Prolific and retained 312 after screening. Participants were assigned between the two interface configurations on a simulated e-commerce platform and completed search, product-evaluation and post-purchase stages.
The richer interface was associated with higher perceived diagnosticity, immersion and satisfaction at all three stages. The largest configuration difference was in evaluation satisfaction: Cohen's d 1.58, with a 95% confidence interval from 1.33 to 1.83. Structural-equation modelling produced indirect-effect patterns consistent with diagnosticity and immersion as pathways, but the paper explicitly says these results do not demonstrate the mechanism.[1]
The bundle cannot reveal which feature worked
The comparison changed voice, avatar, number of modalities, synchronisation and social presence at once. It therefore tests one bundle against another, not the effect of voice or an avatar in isolation. The outcome is self-reported satisfaction in a simulated journey, not completed purchases, returns, spending, decision quality or long-term trust.
For retailers, this is a reason to test richer assistance carefully, especially during product comparison—not evidence to deploy persuasive avatars by default. A useful next trial would separate each design element, pre-register outcomes and measure real behaviour, accessibility, annoyance, disclosure understanding and whether vulnerable users feel pressured. The author reports no external funding and no competing interests.[1]
The distinction that matters
What changed—and what was measured
The experiment compared two complete interface bundles rather than one isolated feature.
Text only
Use a minimal shopping assistant
Baseline satisfaction ratings
Rich bundle
Add voice, animation and avatar cues
Higher reported ratings in simulation
Real commerce
Measure purchases, returns and trust
Not tested
Watch the background, then check the evidence
Illustrative photo · not a video frame
UNESCO · 11 min 27 sec · Background explainer
UNESCO Talks: Interview with 2024 Nobel laureate Professor Daron Acemoglu
YouTube loads only when you choose; playback starts when you press play.
Economist Daron Acemoglu discusses how choices about AI affect workers, inequality and society. Pair the interview with our comparison of task exposure, hiring and skills evidence.
Evidence check: An expert interview from 2024, offering a perspective on policy and work rather than a measured forecast of job losses.
Published 2024-10-14. Source checked 14 October 2026. Check captions on the publisher’s video page.
Read the AI and jobs guidePhoto: Dylan Gillis / Unsplash · Licence
Evidence check
The two conditions differ on several design dimensions, preventing attribution to voice, avatar or synchronisation alone.
The platform and journey were simulated; no real purchase, return, spending or retention outcome was measured.
The accepted article is citable and peer reviewed but still awaiting final production edits.
Reporting basis: Source analysis. Original source date: 19 October 2026.
Continue the story
Read next
Also about multimodal ai
What do people actually ask image-upload AI to do?
An unreviewed Microsoft-led study analysed 42,617 de-identified Copilot image-upload sessions and checked its taxonomy against 23,413 ChatGPT sessions. Users often chained recognition into writing, code and data tasks—but private, model-generated summaries limit what outsiders can verify.
Read the 7 min reportAlso about human-computer interaction
Google adds a live visual avatar to its real-time Gemini model
Google introduced Gemini 3.8 Live with Live Avatar, combining spoken conversation with a real-time visual presence and pointing to a more embodied form of consumer and workplace AI interaction.
Read the 2 min reportExplore the full source and evidence record
What this means for people
- Shoppers may find richer assistants more engaging, but engagement is not proof of better-value or less pressured decisions.
- Retail teams should test accessibility, clarity and behavioural outcomes alongside satisfaction before widening deployment.
Global context
The study was led from China but recruited online through Prolific. Its results concern the tested interface and sample, not every culture, retail category or accessibility context.
What the evidence does not yet show
- The two conditions differ on several design dimensions, preventing attribution to voice, avatar or synchronisation alone.
- The platform and journey were simulated; no real purchase, return, spending or retention outcome was measured.
- The accepted article is citable and peer reviewed but still awaiting final production edits.
What to watch next
- Factorial trials that isolate voice, avatar, animation and synchronisation
- Field experiments measuring purchases, returns, complaints and decision quality
- Accessibility and consumer-protection evidence for persuasive AI interfaces
Living evidence record
Impact record IAI-10PX7MG
Evidence stage
Studied
Confidence
Supported
Reporting basis
Source analysis
Independent or research support
Present
Record status
Monitoring
Last checked
20 October 2026
Source trail
1 direct source across 1 source type.
People impact
Documented in this record.
Uncertainty
Limits and next checks are explicit.
Stages describe the evidence available—not whether a technology is good or bad. See the public method.
Single-source reporting disclosure
This record analyses one direct source. It can establish what Scientific Reports published or reported, but it is not independent corroboration of every performance claim or predicted outcome. The confidence label will change only when broader evidence is added.
Evidence trail
Sources used for this report
Links checked 20 October 2026
Research question, topics and regions
Whether a bundled text, voice and animated-avatar shopping assistant changes perceived usefulness, immersion and satisfaction compared with a text-only assistant
Topics: E-commerce · Consumer experience · Multimodal AI · Human-computer interaction
Regions: China · Online Prolific sample
This report is labelled source analysis. We summarise and analyse source material in our own words; company statements remain attributed claims until independently supported. Translated summaries preserve the meaning of the original source and link back to it. Read our editorial standards.
The Impact Brief · Free
Keep following the stories that matter to you.
Get our weekday briefing on what changed and why it matters. Choose your topics below.
Reader commentary
What does this mean where you are?
Tell us what you have seen at work, in your community or in everyday life. A question is welcome too. No account is needed; we review notes before publishing them.
Explore commentary across the portal →Published reader notes
0No published reader notes yet. You can start the evidence-led discussion above.
Prefer a private correction or response? Contact the newsroom.