Deep dive into Fiverr
On its way to irrelevance?
1.0 Introduction
Fiverr, the $1 billion company, is being disrupted.
The company was founded in 2010 and is an online marketplace for freelance services.
It gained popularity with its quick gig jobs, priced at $5 (hence the name, Fiverr).
Today, it offers a variety of projects, from microtasks to more complex professional services, priced in the hundreds or thousands of dollars.
I believe that LLMs are one prompt away from solving most of the microtasks, and in the long run, Fiverr is likely to go out of business.
1.1 How does it make money?
There are two categories of revenue:
Marketplace revenue - A cut out of the transaction between the seller and the client; and
Service revenue - Optional services that aren’t related to a transaction taking place.
The first category is quite straightforward. The second category includes advertising (allowing sellers to promote their gigs), subscription (offering various additional features for sellers and buyers), and fees earned on user funds.
The marketplace revenue is barely growing. However, services revenue doubled from $41m in 2022 to $88m in 2024.
2.0 UpWork vs. Fiverr
There’s a common misconception that the freelance battlefield is between UpWork and Fiverr.
In August 2024, I shared my UpWork thesis and briefly touched on why Fiverr isn’t its direct competitor:
The spending per buyer supports this. UpWork’s average buyer (client) spends 16x more.
So, why is this relevant?
3.0 The AI disruption
With the rise of AI, many low-budget microtasks are one prompt away from being solved. So, why would anyone purchase a gig instead of writing the same instructions in a prompt?
Since the peak of the pandemic, the number of active Fverr buyers has been declining. ChatGPT was launched in November 2022.
They’re only showing growth through the services, mainly selling advertising spots to the Fiverr sellers.
This increased their take-rate despite the flat Gross Merchandise Value (“GMV“). With a 36% take rate, Fiverr is one of the greediest platforms out.
A growth of this kind isn’t sustainable. Marketplaces grow by growing the pie, not by taking larger pieces of a shrinking pie.
A few weeks ago, Fiverr unveiled “Fiverr Go”, a visionary AI platform that puts creators front and center. At least, that was their description.
A few videos were released that are crucial to understanding the company's direction.
Fiverr Go allows creators to train AI models based on their own work. Clients can then purchase access to those custom models to generate content.
Currently, Fiverr Go offers:
Generative AI images - allowing the buyers to enter a prompt that generates an image; and
Generative AI audio - allowing the buyers to enter text that will be converted to speech.
They’ll also be available for other services, such as songwriting, graphic design, illustration, copywriting, and digital marketing.
The output is quite expensive, with some voice-overs charging $100 for 30 seconds.
But that’s not all. It also allows training a personal assistant. Basically, a chatbot that responds like the seller.
Fiverr seems to be going all-in with AI, and they probably have no other choice.
In theory, one can train a model to respond to chats and create the work the same way they would, allowing an individual seller to scale up without hiring anyone..
Who pays for this? The sellers. What is the price?
$25/month for the personal AI creation model; and
$29/month for the personal AI assistant.
These are higher prices than the best LLMs, but they come with the opportunity to earn money.
Based on Micha Kaufman (Fiverr’s founder and CEO), the personal assistant converts 57% better than average conversations.
In theory, all of this sounds good. In reality, Fiverr is in serious trouble, and here’s why:
The process of training an AI model seems quite general, and it isn’t really trained as much on the seller’s past data.
With this approach, sellers' output would be as good as the ability to write a good prompt. Let’s assume this is the way to go, and eventually, an individual seller can fulfill an infinite number of orders through AI with little to no involvement.
Why are there thousands of logo designers or copywriters needed when the best few can dominate the entire market?
How would one compete with them? Lower the price? The best sellers would attract many more buyers, and they can lower the prices the most.
These low-priced gigs will become commodities, and buyers will benefit from lower prices.
However, for most sellers, this won’t end well.
All of the above is based on the assumption that Fiverr is still a relevant place to purchase these projects despite the rise of LLMs.
Let’s take this one step further. Why shouldn’t Fiverr create its own models and fully replace the sellers?
But then, how is it different than ChatGPT or Grok?
Hint: It isn’t.
4.0 Alternative data
The sellers and buyers aren’t happy with how everything is evolving, as can be seen by the many posts/comments on the Fiverr subreddit:
At the same time, the demand for microtasks generated by AI has increased since the ChatGPT launch.
The demand for AI voice-overs increased later as better models became available.
Fiverr is competing not only with LLMs but also with companies that specialize in niche services. Here’s the offering of elevenlabs, a text-to-speech and AI voice generator:
For $11/month, it offers 100 minutes of ultra-high quality text to speech.
Despite all of this, the CEO is excited about the company’s future, but after all, that’s his job:
5.0 Historical Financial Performance
Revenue growth has been below 10% over the last two years, and it was due to increasing their take-rate (by charging sellers to advertise more).
The company is still unprofitable, and the number of buyers is decreasing.
This, coupled with the fact that LLMs and niche AI competitors are eating their lunch, brings me to my conclusion that Fiverr is being disrupted.
Anyone betting on Fiverr is betting that:
The pivot to combining AI and sellers outperforms pure LLM outputs, carving out a sustainable edge; and
Buyers prefer purchasing from human beings (instead of using LLMs)
If you found this deep dive valuable, consider sharing it with someone who would. Your support helps grow The Finance Corner and brings more deep dives like this to your inbox.