Okara
Start Free
Back to Blog
Published by Jordan Reese · September 18, 2026 · 19 min read

How HeyGen Grew From a $5 Fiverr Order to $200M ARR

How HeyGen grew through Fiverr validation, watermarked free videos, viral translation demos, fast shipping, SEO, APIs, and enterprise localization

HeyGen did not start with a big launch or a large sales team.

Its first paying customer spent $5.

Founder Joshua Xu had listed an AI-powered spokesperson service on Fiverr. He fulfilled the work manually and used the orders to learn what customers wanted before turning the idea into software.

That small experiment became the start of a much larger growth loop.

HeyGen later launched a self-serve product, gave people a free way to try it, put its watermark on free exports, and built features people naturally wanted to share. Then Video Translate made the product even easier to demonstrate: upload a clip, choose another language, and watch the same person appear to speak it with matching lip movements.

By June 2026, HeyGen said it had passed $200 million in annual recurring revenue (ARR), with more than 30 million users who had created over 118 million videos.

The simplest explanation for HeyGen's growth is this:

HeyGen made the output of the product do the marketing.

A generated video was useful to the person who made it. But when that video was shared, it also became a live demo of HeyGen for everyone who watched it.

That loop started with avatar videos, became much stronger with AI translation, and later expanded into SEO, APIs, and enterprise sales.

HeyGen growth timeline

DateMilestone
December 2020Joshua Xu and Wayne Liang start the company that would become HeyGen, initially called Surreal
2021Surreal serves e-commerce exporters and reports about 10 million photo and video orders
July 29, 2022The self-serve product launches as Movio
November 2022TechCrunch reports a user base in the lower hundreds of thousands and nearly 1,000 paying customers
Early 2023HeyGen reaches about $1M ARR
April 2023Movio rebrands to HeyGen
September 2023HeyGen launches Video Translate; a user demo passes 6 million views
October 2023HeyGen reaches $10M ARR, according to Forbes
November 2023HeyGen reaches $18M ARR and launches a much faster custom-avatar workflow
January 2024A HeyGen-translated version of Javier Milei's Davos speech reaches about 75 million views on X
June 2024HeyGen reports $35M+ ARR and 40,000+ paying business customers; Benchmark leads a $60M Series A
Around October 2025HeyGen says ARR reaches $100M
June 2026HeyGen reports $200M ARR, 30M+ users, and 118M+ videos created

There is one timeline wrinkle worth knowing. HeyGen's founder retrospective says it reached $1 million ARR 178 days after its July 29, 2022 launch, which points to January 2023. Forbes later reported that HeyGen reached $1 million ARR in March 2023. The safest description is early 2023.

Also, ARR is not the same as audited annual revenue. HeyGen is private, so the later revenue figures in this article should be read as company-reported recurring revenue run rates.

1. HeyGen validated the job before building the SaaS

HeyGen began in December 2020 as Surreal.

Xu had spent more than six years at Snap working on advertising systems, machine learning, and camera technology. His basic idea was that AI could remove one of the hardest parts of making a video: recording the person on camera.

Surreal's first customers were Chinese e-commerce exporters.

Those merchants often needed product photos and videos with foreign models. Hiring talent was slow and expensive, so Surreal used synthetic media to create the content instead.

In March 2021, TechCrunch reported that Surreal had already handled around 10 million photo and video orders. That was not 10 million users. It was a count of jobs processed by the service.

The company had technical capability, but Xu still needed to find the right product.

One early attempt was a stylized computer-generated avatar. Customers were not interested.

Xu later wrote:

"Our CG avatar, for instance, lacked a product-market fit. Users wanted authentic videos."

That failure pushed the team toward a much simpler question:

What part of video production are businesses already paying for that AI could make much cheaper or faster?

The answer was spokesperson videos.

2. A $5 Fiverr order gave them better validation than a waitlist

Before building a full SaaS product, Xu searched Fiverr for "spokesperson."

He found 1,811 services.

That told him there was already a market. People were paying humans to record scripted videos for product demos, ads, explainers, and business content.

So HeyGen created its own Fiverr listing.

The service offered multilingual spokesperson footage at a lower price and with a faster turnaround. Xu handled the workflow manually.

The first customer paid $5.

The listing initially did not disclose that the spokesperson was generated with AI. The team later changed the listing to disclose the use of AI, and customers continued to buy.

The important part was not the $5.

It was that a stranger had paid for the outcome.

That gave HeyGen three things before it had built a polished product:

  1. proof that people already paid for this job,
  2. real customer conversations about what they wanted,
  3. early users who could later become SaaS customers.

This is one of the most useful parts of HeyGen's story.

The founders did not ask people if they would use an AI avatar product. They sold the result first.

3. Movio turned the manual service into a self-serve product

Development of the SaaS product began in the second quarter of 2022.

On July 29, 2022, the company launched Movio, the product that would later become HeyGen.

It looked more like a video editor than an AI research demo.

A user could:

  • choose an avatar,
  • type a script,
  • select a voice,
  • use a template,
  • arrange scenes,
  • generate a finished video.

That mattered because most customers did not care how the model worked. They wanted a usable marketing, training, or explainer video.

Movio also launched on Product Hunt. That brought early discovery, but Product Hunt was not the long-term growth engine.

The product itself became the engine.

By November 2022, TechCrunch reported that Movio had a user base in the lower hundreds of thousands and was approaching 1,000 paying customers.

4. The free plan turned every shared video into an ad for HeyGen

This was probably HeyGen's most important early growth mechanic.

Xu and Liang came from consumer products. Xu had worked at Snap, while Liang had worked at Smule and ByteDance.

They brought a consumer growth idea into B2B software.

Movio offered a free tier, but free videos included a visible watermark.

That created a simple loop:

user creates a free video → user shares it → other people see the HeyGen/Movio watermark → some of them visit the product → they create their own videos

Xu later wrote that the watermark "fueled sharing" and that traffic spikes sometimes overwhelmed the service.

Most SaaS products hide inside a company.

HeyGen's product escaped into public feeds.

Every LinkedIn post, TikTok, YouTube upload, landing page, or ad made with the product could also show another person what the product did.

That is much stronger than a normal referral link because the viewer does not have to imagine the value.

They can see the result.

5. TalkingPhoto made the product easier to try and easier to share

Another important growth feature came directly from customer questions.

People kept asking:

Can I use a photo instead of recording avatar footage?

At first, the answer was no.

But the request revealed two problems.

Some users wanted a cheaper way to test the product. Others simply did not want to record footage before they knew whether the result would be useful.

So the team built TalkingPhoto.

A user could upload a still image and make it speak.

TalkingPhoto lowered the effort needed to experience the product, but it also made HeyGen more entertaining.

People could animate family photos, characters, celebrities, or unusual images.

That made the output more likely to spread on social media.

The feature therefore did two jobs at once:

  • it reduced friction for a new user,
  • it created more shareable outputs.

This became a recurring pattern in HeyGen's growth.

The best features were not only useful inside the product. They were also easy to demonstrate outside it.

6. HeyGen focused on getting users to the first generated video

Traffic alone was not enough.

HeyGen found that many new users arrived but did not understand how to use an AI avatar editor.

So from roughly November 2022 through January 2023, the team focused on activation.

The key insight was simple:

The "aha" moment happened when a user watched their first generated video.

HeyGen then tried to shorten the distance to that moment.

The team:

  • added onboarding videos,
  • built pages that could get users to an output without a normal registration flow,
  • created more than 200 templates,
  • kept simplifying the creation process.

This is an important distinction.

HeyGen did not optimize only for signups.

It optimized for completed output.

A signup could forget the product.

Someone who had watched themselves, their photo, or their script turn into a finished video understood the value.

According to HeyGen's retrospective, the company reached $1 million ARR in roughly seven months and became "ramen profitable" soon after. The company later said it was fully profitable by the second quarter of 2023.

7. Video Translate created an even stronger viral loop

The original avatar product still needed some explanation.

Video Translate did not.

HeyGen launched the feature in September 2023.

A user could upload a video of someone speaking, choose another language, and get a version that translated the speech while also changing the voice and visible mouth movements.

The result looked like the same person had recorded the video in another language.

That is an almost perfect social-media demo.

On September 11, 2023, creator Jon Finger posted a test in which he spoke English and used HeyGen to make himself appear to speak French and German.

He did not speak either language.

The post passed 6 million views.

Tom's Guide tested the product around the same time and reported more than 141,000 videos waiting in the processing queue.

The viral loop had become stronger:

person sees impossible-looking translated video → immediately understands the product → tries it on their own video → shares the output → more people discover HeyGen

Unlike a normal SaaS screenshot, the output itself was the proof.

8. The Javier Milei translation showed how powerful a good demo could become

A few months later, the same loop produced an even larger moment.

Argentine president Javier Milei gave a speech in Spanish at the World Economic Forum in Davos in January 2024.

Aaron Slodov, the founder of Atomic Industries, used HeyGen to translate the speech into English while preserving Milei's voice and changing his lip movements.

Slodov posted the result on X.

The translated video reached roughly 75 million views and 240,000 likes, according to The National.

HeyGen did not pay for that distribution.

A user had created a better advertisement for the product than a conventional marketing campaign could have.

The video showed three things instantly:

  • the original content could reach a new language,
  • the speaker still sounded like himself,
  • the mouth movements followed the translated speech.

This is why Video Translate mattered so much.

It was not just a feature.

It was a feature whose value became obvious the moment someone shared the output.

9. Viral self-serve adoption opened the door to enterprise

HeyGen's next step was not to abandon self-serve growth.

It was to build a larger business on top of it.

Forbes reported that HeyGen reached:

  • $1M ARR in March 2023,
  • $10M ARR in October 2023,
  • $18M ARR by late November 2023.

At the same time, HeyGen made it easier to create a personal avatar.

The earlier workflow could require professional footage and take days. A new version released in November 2023 could create a custom avatar from smartphone footage in about five minutes.

That expanded the number of people who could use the product without help.

Then HeyGen started building a more serious enterprise motion.

Larger customers needed things that an individual creator did not:

  • team workspaces,
  • security controls,
  • custom avatars,
  • administration,
  • faster production,
  • APIs,
  • localization at scale.

By June 2024, HeyGen said it had more than 40,000 paying business customers and more than $35 million in ARR.

Benchmark led a $60 million Series A with participation from Thrive Capital, BOND, Conviction, and others.

The company said it was already profitable.

The pattern is important:

self-serve created awareness and product adoption; enterprise captured more value from the customers who needed the product at scale.

10. Localization gave enterprises a clear ROI story

Enterprise AI products often struggle because the value is vague.

Localization gave HeyGen a very concrete use case.

A company that had already recorded a training video, product demo, course, or advertisement could translate it into more markets without recording the speaker again.

That can replace weeks of coordination across actors, studios, translators, editors, and regional teams.

For example, HeyGen says trivago used the product to localize TV ads across 30 markets.

According to the case study, trivago cut post-production time by 50% and saved roughly three to four months.

HeyGen also says Würth Group reduced translation costs by 80%.

These are vendor-published customer case studies, not independent audits, so they should be treated as examples rather than universal benchmarks.

But they show why the enterprise product sold.

The pitch was not simply "use an AI avatar."

It was:

make one video once, then adapt it for every market.

That is a much easier budget conversation.

11. APIs let HeyGen grow through other products

HeyGen had offered an API since the Movio era.

That became more valuable as the product expanded.

Instead of requiring every customer to visit HeyGen's editor, another product could generate avatar videos or translations in its own workflow.

This created a second kind of distribution.

A normal self-serve customer generates videos inside HeyGen.

An API customer can generate videos repeatedly for many end users or internal workflows.

HeyGen later expanded its developer products across video generation, translation, lip sync, real-time avatars, and agent-driven video creation.

That means growth no longer had to come only from people discovering the HeyGen website.

The product could sit inside another product.

12. SEO came later and widened the top of the funnel

HeyGen's early growth was mainly product-led and social.

SEO became a more important layer once the company already had product-market fit.

Webstacks, which worked on HeyGen's website, says it helped HeyGen build an AI Tools Directory around specific use cases such as translation, dubbing, voice generation, and video personalization.

According to Webstacks, the directory attracted more than 25,000 organic users in its first month.

The team also created a reusable system that allowed HeyGen to launch 40+ SEO and SEM landing pages in a matter of weeks.

The strategy fits HeyGen's broader growth model.

Instead of writing only informational blog posts, many pages let the visitor understand or try a specific job.

A person searching for an AI translator, voice tool, avatar, or image-to-video product could land close to the feature itself.

So SEO became another path into the same product-led loop:

search → use case page or tool → first output → share or upgrade

It was an expansion channel, not the original reason HeyGen broke out.

13. HeyGen kept shipping faster than the market was settling

Another part of the story is speed.

By the time Movio became HeyGen in April 2023, the company said it had shipped 32 versions and more than 100 features in roughly nine months.

The team did not wait for perfect data before every decision.

Xu has written that early startups often do not have enough traffic for clean A/B tests. HeyGen preferred to ship, watch how customers behaved, talk to them, and improve the product quickly.

That pace continued as the market changed.

The company moved from:

synthetic models → spokesperson videos → self-serve avatars → TalkingPhoto → video translation → instant avatars → enterprise localization → APIs → video agents

The product kept widening, but the core job stayed similar:

help a person or business create useful video without the cost and friction of a normal shoot.

14. The growth loop in one diagram

HeyGen's growth can be reduced to this:

Useful AI video output
        ↓
User shares the output
        ↓
The output demonstrates HeyGen
        ↓
New viewer tries the product
        ↓
Templates + simple onboarding get them to a first video
        ↓
Some users pay
        ↓
Teams use HeyGen for more videos, languages, and workflows
        ↓
More HeyGen-made output appears in public
        ↺

Video Translate made this loop stronger because the output was surprising enough that people wanted to show it to others.

Enterprise and API usage then added another loop:

Individual or team adopts HeyGen
        ↓
Proves time or cost savings
        ↓
More departments / markets / workflows adopt it
        ↓
Usage expands
        ↓
Higher recurring revenue

15. How HeyGen got from $35M to $200M ARR

HeyGen's later growth became less visible because it was no longer driven by one launch.

The company kept expanding across individual creators, small businesses, enterprise teams, and developers.

In June 2026, HeyGen said it had passed $200 million ARR, doubling from $100 million in about eight months.

The company also reported:

  • 30M+ users,
  • users across 196 countries,
  • 118M+ videos created,
  • adoption across 85% of the Fortune 100.

HeyGen says most of its growth still comes through word of mouth.

Upstarts reported that the company had raised about $74 million in total, used roughly $25 million of it, and remained cash-flow break-even through the first half of 2026.

Those figures come from the company and should not be treated as audited financial statements.

Still, the growth pattern is unusual.

HeyGen did not need hundreds of millions of dollars in funding to reach a reported $200 million ARR run rate.

Its June 2026 announcement said it generated about $2.70 in ARR for every dollar of equity capital raised.

What actually drove HeyGen's growth?

A lot happened between the first Fiverr gig and $200 million ARR, but six things mattered most.

1. It validated willingness to pay before building the full product

The Fiverr experiment answered a better question than a waitlist.

Not "do you like this idea?"

But:

will you pay for this result?

2. It turned product usage into distribution

The free-plan watermark made every public output a possible acquisition channel.

That is especially powerful for creative products because the result naturally leaves the app.

3. It built features that were easy to demonstrate

TalkingPhoto was fun to share.

Video Translate was even better because a person could understand the value in seconds without reading an explanation.

HeyGen's best content was often made by its users.

4. It optimized for the first successful output

HeyGen treated the first generated video as the activation event.

Templates and easier onboarding helped more users get there.

That made the top-of-funnel traffic more valuable.

5. It used self-serve adoption to enter enterprises

Individual users could start without talking to sales.

Once the product proved useful, larger companies could buy security, localization, workspaces, custom avatars, and APIs.

This let HeyGen combine product-led growth with larger contracts.

6. It kept adding new distribution layers

The channels arrived in stages:

Fiverr validation → Product Hunt → freemium sharing → social virality → word of mouth → enterprise sales → APIs → SEO/free tools

HeyGen did not depend on one acquisition channel forever.

It used one layer to reach the next stage.

The biggest lesson from HeyGen

HeyGen's story is often described as an AI story.

It is also a distribution story.

The company found a product where the result could travel.

A watermarked avatar video promoted the avatar product.

A translated video promoted the translation product.

A localized enterprise campaign proved the enterprise use case.

The user did not have to write a testimonial explaining why HeyGen was useful.

The output was the testimonial.

That may be the most important reason HeyGen grew so quickly.

Frequently asked questions

How did HeyGen grow?

HeyGen first validated demand through a manually fulfilled Fiverr spokesperson service. It then launched a self-serve product, used a free tier with watermarked exports to create a viral loop, improved activation with templates and easier onboarding, and gained another wave of growth from viral Video Translate demos. It later added enterprise sales, APIs, SEO pages, and free tools.

What was HeyGen's first customer?

According to co-founder Joshua Xu, HeyGen's first paying customer came through a Fiverr spokesperson gig and paid $5.

When did HeyGen reach $1 million ARR?

HeyGen's own founder retrospective says it reached $1 million ARR 178 days after launching on July 29, 2022, which would place the milestone around January 2023. Forbes later reported March 2023. Because the public sources conflict, early 2023 is the safest description.

How did HeyGen use product-led growth?

HeyGen offered a free tier, placed a visible watermark on free exports, and built shareable features such as TalkingPhoto and Video Translate. When users posted their videos publicly, the outputs acted as product demos and brought new people back to HeyGen.

What made HeyGen Video Translate go viral?

The feature was easy to understand from the output alone. A user could appear to speak another language in their own voice with matching lip movements. Jon Finger's September 2023 demo passed 6 million views, and a January 2024 HeyGen translation of Javier Milei's Davos speech posted by Aaron Slodov reached about 75 million views.

How much revenue does HeyGen make?

HeyGen reported $200 million in annual recurring revenue in June 2026, up from $100 million about eight months earlier. ARR is an annualized recurring-revenue run rate, not audited revenue recognized during a completed financial year.

How many users does HeyGen have?

HeyGen reported more than 30 million users and more than 118 million cumulative videos created in June 2026. The company did not describe the 30 million figure as monthly active users.

Is HeyGen profitable?

HeyGen said it became profitable in the second quarter of 2023. Upstarts reported that the company remained cash-flow break-even through the first half of 2026. These figures are company-reported.

Sources

Tags

Growth stories