Video Marketing

AI Avatar Videos and B2B Trust: When to Use AI and When to Go Custom

Match the tool to the trust: when to use AI avatar video and when to go custom

Why it matters

A 37-point gap between advertisers and viewers

82% of ad executives think young consumers feel positive about AI-generated ads. Only 45% of those consumers do. That's the headline of the IAB's January 2026 study with Sonata Insights, and the gap has grown since 2024.

What the IAB foundResult
Ad executives who think Gen Z and Millennials like AI ads82%
Gen Z and Millennial consumers who do45%
The gap between the two37 points, up from 32 in 2024
Gen Z who call brands that use AI ads “inauthentic”30%
Young consumers who want AI avatars disclosedNearly half
Advertisers naming cost efficiency as AI's top benefit64%

The IAB surveyed 505 US consumers and 104 ad executives. This is consumer research, not a study of B2B buyers. But B2B buyers are consumers too, and in B2B the gap gets expensive. The cheapest-looking shortcut in video right now is an AI avatar reading your script.

The short answer: use avatars where the viewer already trusts you, such as training, onboarding, help and translation. Make custom video wherever a stranger has to trust you: launches, demos, founder messages and customer stories. We make 2D, 3D, AI-generated and mixed-media video, so this isn't an anti-AI page. Each section below includes a video we made that shows the line.

Not sure which videos need custom production?

Talk to the Advids team about which videos on your plan can use AI and which need to earn trust.

Let’s talk

Trust penalty

Buyers trust AI-labelled content less, even when it is identical

Same ad, different label, lower trust.

The Nuremberg Institute for Market Decisions showed people identical ads. Half were told the ad was AI-generated. Half were told people made it. The content was the same. But the "AI" version was rated less natural and less useful, and people were less willing to research or buy the product. NIM also found AI ads were better accepted for innovative high-tech products, so the penalty is real but not universal.

The labels are getting harder to avoid. In the IAB's January 2026 study with Sonata Insights, nearly half of young consumers want AI avatars disclosed, and 30% of Gen Z call brands that use AI ads "inauthentic".

Bloomberg Law: the sources on screen, not a face

Our 2D animation for Bloomberg Law shows the platform across a monitor, a phone and a tablet. A scale weighs "legal" against "regulatory". Then the claims arrive as things a lawyer can check: "over 75,000 global sources", "ALM national & state legal news", "data-driven expert analysis" and "Practical Guidance". No presenter appears. See how we made it.

What works here:

  • Credibility comes from named sources a lawyer can verify.
  • No face on screen is asked to vouch for the product.

The move: Where the buyer's risk is high, put verifiable sources on screen before you put a face there.

Now put that in a B2B setting. Alessandro Cutri, who runs paid acquisition for B2B tech companies, compares a B2B purchase to buying a house: several people are involved and the risk is real. In that kind of decision, the face on screen is making a promise. If buyers suspect the face isn't real, they start wondering what else isn't. Abmatic's piece on trust and credibility makes the same link to conversion.

So does that mean never use avatars? No. It means use them where trust already exists.

Where avatars fit

Avatars work where the viewer already trusts you

Training, onboarding, help and translation: the viewer already chose you.

Think about internal training, customer onboarding, help-centre videos and translations. The viewer already chose you. They want the information fast, and the content changes all the time. That's where an avatar saves real time. Our guide to product training videos covers the formats.

Here's a real example. Wise, the money-transfer company, uses AI avatar video for its people training. On Synthesia's case-study page, Rob Markus, Wise's People and Organisational Development Lead, says production cycles got 20% shorter and each course update takes 20% less time. Synthesia sells avatar video, so read those figures as a vendor's case study.

Wise, via SynthesiaResult
Production cycles20% shorter
Time per course update20% less
EngagementUp 5%

Notice who's watching: Wise's own people. Nobody in that audience is deciding whether to trust Wise. They already work there. That's the sweet spot: lots of updates, an audience that already trusts you, and no promise being made by the face.

SundaySky: avatars paired with customer service

Our promo for SundaySky, an AI video platform, ties each use case to one feature. "Onboarding videos" links to "Personalization": an airline welcome reads "You got your wings!" with "Your tier is {Tier level}". "Customer Service videos" links to "Avatars": an avatar picker, then "Nice to meet you! I'm Chris! Let me walk you through some helpful account features." See how we made it.

What works here:

  • The avatar appears in a customer-service video, talking to someone who already has an account.
  • Personalisation is shown on onboarding, where the viewer has already said yes.

The move: Put the avatar where the viewer is already a customer and wants an answer fast.

That's one column. Now the other: the videos where a stranger has to decide whether to trust you.

Where custom wins

Custom production wins wherever a stranger has to trust you

Brand, launch, demo, founder and customer story: the face is making a promise.

Custom production wins wherever the video has to earn trust from someone who doesn't know you yet. That's your brand video, your launch video, your sales demo, your founder's message and your customer stories.

Video producer Fraser Cottrell says most ads fail because they forget people run on emotion, not logic. A first impression is exactly where that matters most, and it's where a slightly-off avatar does the most damage.

Zoho CRM Plus: real people beside real screens

Our explainer for Zoho CRM Plus puts actors on a plain white set beside real Zoho screens: a customer, and one person for each team. It follows one customer from a campaign email to a chat, a sales call and a support ticket, with hand-off buttons such as "Push To CRM" and "Convert to Ticket" in close-up. A support agent in a headset works the ticket on screen. See how we made it.

What works here:

  • Each team gets a human face next to the app that team uses.
  • Actors play the roles, and nobody is presented as a real customer giving a testimonial.

The move: For a buyer's first look, pair real people with your real product, and never present an actor as a real customer.

One line you should never cross: don't use an avatar as a customer. In August 2024, the US Federal Trade Commission announced a final rule that bans fake testimonials, including AI-generated ones from people who don't exist or never used the product. Real B2B testimonial videos are social proof. A synthetic one is a liability.

So how do you decide for any single video? Ask three questions.

Three questions

A three-question test for any single video

Run every video on your plan through the same three questions.

This test is our recommendation, not a published framework.

QuestionIf yesIf no
1. Does the viewer already trust you?An avatar may fitMake it custom
2. Will the video change often or need many languages?An avatar saves real timeCustom costs little extra over its life
3. Is the face on screen making a promise?Make it customAn avatar may fit

If the first two answers are yes and the third is no, an avatar is fine. Just label it. Otherwise, make it custom.

Jeeves: help content that changes with every upgrade

Our video for 314e's Jeeves, a just-in-time training platform for EHR users, shows help inside the EHR. A search for wound care returns a video and a 20-page tip sheet. The Jeeves Bot answers "How to enter insurance details?" Then it asks "What about creating content during an upgrade?" and shows a screen recording with a webcam presenter turned into new training. See how we made it.

What works here:

  • The viewers are already users: clinicians stuck mid-task.
  • The content has to be remade every EHR upgrade, so speed matters more than polish.

The move: Sort your videos into reference content and trust content. Reference content can be fast; trust content can't.

Let's run it on two videos. A help-centre clip on how to export a report: the viewers are already customers, it changes every release, and nobody's promising anything. Avatar, fine. Your homepage founder video: strangers, it rarely changes, and the face is the promise. Custom, every time.

For the avatar side, guides to end-user training and animated explainers for product training show where reference content lives. And whichever you pick, follow the IAB's advice: use AI to raise the quality of your video, not just to make it cheaper.

Here's the whole test as one table, by video type.

Where to start

Where to start

The short answer: match the tool to the trust the viewer already has.

Video typeUseWhyExample to copy
Internal training, translationsAI avatar, labelledThe audience already trusts you; content changes oftenSundaySky
Help centre, product updatesRun the three questionsFast updates matter; layered files are the other optionJeeves
Brand, launch, sales demoCustomA stranger is deciding whether to trust youZoho CRM Plus
High-risk categories, such as legalCustom, sources on screenAI labels cost trust even on identical contentBloomberg Law
Founder message, customer storyCustom, never an avatarThe face is the promise; fake testimonials are bannedYour own customer

For more on the avatar side, see our AI onboarding videos, or see the software teams we've made video for.

FAQ

Questions about AI avatar videos

Do AI avatar videos hurt trust with B2B buyers?
They can. In NIM's test, identical ads labelled AI-generated were rated less natural and less useful. The penalty matters most where a stranger must trust you.
Which videos are safe to make with an AI avatar?
Videos for people who already trust you and that change often: internal training, onboarding, help-centre clips and translations. Label the avatar, and keep the face from making promises.
Should AI avatar videos be labelled?
Yes. In the IAB's January 2026 study, nearly half of young consumers want AI avatars disclosed. Labelling avoids the bigger trust loss of being found out later.
Can an AI avatar appear in a customer testimonial?
No. The FTC's 2024 final rule bans fake testimonials, including AI-generated ones from people who don't exist or never used the product. Use real AI customer testimonials only.
Is custom production worth the cost for a sales demo?
Usually, yes. A sales demo asks a stranger to trust you, and that's where an AI label costs most. Layered master files keep custom demos current as the product changes.
Can one video mix AI-generated and custom footage?
Yes. AI can speed up backgrounds, translations or reference sections, while real people and real screens carry the promise. Follow the IAB's advice: use AI to raise quality.

About the author

About Jai Ghosh

Video Producer at Advids · LinkedIn

Jai Ghosh is a Video Producer at Advids, where he has planned and produced videos for B2B software and technology companies for 7 years. He works on product demos, explainer videos and brand videos, from the first script to the final cut, for clients from startups to enterprises.

Before Advids, Jai spent over 11 years in freelance media consulting, after starting out in digital journalism. He writes the Advids blog's breakdowns of product and demo videos, sorting real examples by the question each one answers for a buyer.

  • Focus: SaaS product demos, explainer videos and brand videos
  • Experience: 7 years at Advids, over 11 years in media consulting
  • Background: digital journalism

Connect with Jai Ghosh on LinkedIn

Sources

Sources