Achieving AI-Market Fit
Building AI models and hacking are my strengths, yet they mean little without product-market fit (PMF). This aspect is critical, especially in today's AI boom. We also applied a new concept of AI-market-fit, particularly with video presentation technology.
In technology's fast-paced world, distinguishing demo from user value is tough. Some demos catch your eye, but user value has longevity. We started our company in December 2020, confident that AI could generate quality content like unique product demonstration videos. Our mission was to transform visual storytelling by shifting from camera-based to AI-generated content.


To realize this vision, we divided it into three steps. First, creating a video engine for businesses to facilitate talking head video examples for corporate communications. Then, developing a SaaS product to push technology limits, connect with users, and generate revenue for sustained investment.


After exploring technical feasibility and use cases, we targeted the spokesperson scenario. Searching "spokesperson" on Fiverr revealed 1,811 services, indicating demand if we crafted an innovative product experience. Almost a year ago, we launched a Fiverr gig offering on-demand, multi-language video footage. Initially, we didn't disclose our AI avatars, offering similar services but at a fraction of the cost and with a quick turnaround. I managed the manual process of delivering videos. Our competitive prices and fast delivery led to our first paid customer for $5. Later, we updated our gig to reveal our AI avatars, and feedback remained positive. We gained more clients, understanding use cases and pricing.


This approach validated our AI-market-fit cost-effectively, earning our first customer. These early users evolved into product clients, providing vital feedback for our initial version. For other efficiency tools, platforms like Fiverr or Upwork can reflect demand-supply dynamics, hinting at opportunities in translation, SEO, image creation, video production, and more.
Lessons Learned
- Connect with the right person early, even pre-product. Quickly craft a minimum viable product (MVP) for potential buyers. No payment means no validation.
- After meeting real users, fully listen. It's challenging to accept you may not know what they want. Practice humility. As a tech founder, I had many ideas. Mainly, they didn’t succeed. Our CG avatar, for instance, lacked a product-market fit. Users wanted authentic videos. Listening better would've kept us on track.
- Don't blindly believe in popular tech trends. They don't always align with what users want. Focus on genuine user needs.
- Prioritize the right problems. Validating a need is more important than solving perceived "big" issues.
Product Journey
Launch on Jul 29th
Development began in Q2 2022. On July 29th, 2022, we launched, claiming top spots on Product Hunt. This marked HeyGen's (formerly Movio) journey's start. For comprehensive product launch strategies, see our Step-by-Step Guide to Your SaaS Product Launch Video.


Inventing TalkingPhoto
Customers asked about using photos for avatars. While initially "no," this question revealed the need for low-cost spokesperson videos and quick feature testing pre-complete footage. Hence, TalkingPhoto was born, embraced during beta as it promoted product-led growth (PLG). TalkingPhotos are creative and fun, boosting social sharing, offering new dimensions to create video ads.
Boosting Traffic: Freemium & Watermark
We applied our consumer product growth expertise to B2B with freemiums. Not content with standard referrals, we added a notable watermark in videos, key to bootstrapping. Though unusual, the watermark fueled sharing, yielding a skyrocketing viral coefficient crucial for viral video marketing.


Traffic surged with freemium, augmenting social media buzz. Our system occasionally crashed from massive views—indicative of genuine PMF and viral success.
Amplifying Viral Engine: Featuring in Gen AI Maps
Our traction grew, garnering Generative AI map features. Sequoia Capital mentioned us first on October 24th. From there, mentions increased, amplifying our user acquisition engine through interactive video examples.


Improving User Experience
Initially, many customers didn't activate due to unfamiliarity. From November to January, we focused heavily on onboarding. Realizing the "aha" moment came from viewers seeing their AI-generated video, we built pages bypassing account registration, embedded onboarding videos, and over 200 templates to streamline their process, significantly boosting conversions and gathering favorable G2 reviews. Check out our Best Video Templates for Ads That Convert for effective marketing.


Building the Product
Our startup product approach focused on quick releases—a weekly schedule despite standard bi-weekly industry pace. This accelerated $1M ARR by prioritizing key improvements over peripheral fixes.
We finalize designs weekly, prioritizing fixes by Fridays after Thursday releases. This cycle propels rapid iteration.


Breaking things isn't a setback but an indicator of hitting PMF—a stimulating approach during early AI-Market-Fit. Our MySQL approach emphasizes real PMF moments over premature optimization.
Distributed Teamwork
Initially skeptical of remote productivity, I soon noted team interaction bottlenecks depended on dependencies. Solutions involved embracing async communication tools, time zone overlap meetings, and a flexible team, boosting productivity.
Data Dashboards
Using Metabase, we crafted dashboards early for data insights, ensuring valuable metrics from the start.
A/B Testing Caution
Early startups lack data for A/B testing's statistical efficacy. Shipping fast offers better insight.
Tools
Third-party tools accelerated project progress, e.g., Datadog, Azure Synapse, Zapier, Typeform, and more.
Customer Interaction
A great product requires strong user momentum. By chatting with customers and embracing feedback, we grew alongside them. Even proposing customizations were gauged based on wider user benefit.


Fostering Transparency
"HeyGen Loves" and "HeyGen Hates" capture customer feedback. Transparency keeps issues addressed.


Prioritizing Customer Success
Each day involves checking activity (our key retention measure) in Airtable, aiming for green (positive) indicators. Staying simple prioritizes user value creation.


Reaching 1,000 customers brought scaling considerations.
Learning in SaaS
Learning how to build a SaaS product carried unique challenges. Confidence in adaptability and humility carried immense learning potential, guided by podcasts, newsletters, and competitor products.
Beyond $1M ARR
As of April 26th, HeyGen achieved profitability, just two months past $1M ARR. New projects like HeyGen 2.0 push our vision while expanding our GTM team to drive further expansion.
Sharing our story might aid others on their journey. Building startups is challenging yet rewarding, fostering support for HeyGen.
Start your own journey with HeyGen and explore the possibilities of AI-generated video content. Sign up for free today at HeyGen.