Fiverr lost a fifth of its buyers to AI. The survivors pay more.

Fiverr's freelance marketplace is shrinking, and the company's own second-quarter numbers show which part of it AI took first. On July 29 Fiverr reported Q2 2026 revenue of $97.8 million, down 10% from a year earlier. The marketplace, the core business where buyers hire people, fell harder: down 15.5%, to $63.1 million. The stock dropped about 17% the same day.

Read one level down and the story sharpens. Fiverr had 2.7 million annual active buyers at the end of June, down 21.9% from 3.4 million a year earlier. A fifth of the buyer base, gone in twelve months. The buyers who stayed spent more, though: $368 each on average, up 15.6% from $318. AI didn't hollow out the whole market evenly. It deleted the cheap, transactional bottom and left a smaller group of buyers paying for bigger work. Fiverr's own guidance says that shift is still speeding up.

What the marketplace number measures

Fiverr reports two revenue lines, and the gap between them is the whole story.

Marketplace revenue is the cut Fiverr takes when a buyer hires a seller. That take rate rose to 28.0% from 27.6% a year ago, so the 15.5% drop in this line isn't Fiverr charging less. It's fewer and smaller jobs flowing through the platform. This is the number that tracks humans getting hired for money.

Services revenue is what Fiverr sells to its own sellers: subscriptions, ads, and tools. It came in at $34.6 million, up 2.0%. That line held while the hiring line fell.

A fifth of the buyers walked

The buyer count is the cleanest read on what AI did. Fiverr went from 3.4 million to 2.7 million active buyers in a year. These are people who spent at least a dollar in the trailing twelve months, so this isn't churn at the margin. It's the base itself contracting.

CEO Micha Kaufman framed it as AI absorbing "high-volume, low-value, transactional tasks." That framing is the company's read, not a measured fact, but it fits the shape of the loss. The buyer who once paid $15 for a logo, a 500-word blog post, a basic translation, or a data-entry batch now types the request into a chatbot or a design tool and gets a good-enough answer in seconds. That buyer doesn't file a brief, doesn't wait two days, and doesn't come back. The work that was easiest to sell was also the easiest for a model to eat.

This is the same pressure showing up across the gig economy. It's the floor described in the median side hustle math: the low end of paid work is where AI lands first and hardest.

The ones who stayed spend more

Spend per buyer went the other way: $318 to $368, up 15.6%. Two things can produce that number, and honesty means naming both.

One is real. The buyers left standing are commissioning larger, more complex projects, the kind where someone still has to manage the AI output, take accountability, and finish the last mile a model can't. Kaufman's own language points here, toward "longer duration projects where AI tools enhance human expertise." The take rate edging up to 28% is weak support for this read.

The other is arithmetic. When you lose your cheapest buyers, the average spend of everyone left rises on its own, even if no single buyer changed their behavior. Some of the 15.6% is almost certainly this composition effect, not a genuine step up in demand. Both forces are probably running at once. The point is that the average is climbing while the headcount falls, and that combination is what a market looks like when its bottom is being cut away.

Where Fiverr's money is migrating

The growing half of Fiverr is the half that charges sellers. Services revenue, now about a third of the total, comes from Seller Plus subscriptions, Fiverr Ads, the AutoDS e-commerce tool, and Fiverr Go, its AI product line. All of that is money Fiverr collects from freelancers, not from buyers hiring them.

So the mix is quietly inverting. As the pool of buyers shrinks, Fiverr leans harder on selling subscriptions and promoted listings to the sellers competing for what's left. For a freelancer, that means the platform's own economics now push you to pay for visibility in a market with fewer clients in it. Worth knowing before you buy the upgrade.

Q3 is guided lower, not flat

None of this is presented by the company as a one-quarter dip. Fiverr guided Q3 revenue to $80 to $88 million, a midpoint around $84 million, down 18% to 26% year over year and well under the roughly $99 million analysts had modeled. It cut full-year guidance to about $364 million at the midpoint, down from around $400 million.

Those are projections, not results, and guidance can be sandbagged. But a company doesn't usually walk its own numbers down by that much unless it expects the trend to continue. On Fiverr's own math, the marketplace bleed accelerates into Q3 rather than bottoming.

If you sell your time, here's the map

The reading here isn't that freelancing is finished. It's that position inside the market now decides everything. The transactional tier, anything a competent buyer can finish in a single AI prompt, is being deleted, on Fiverr and everywhere else. The money that remains is concentrating in fewer buyers paying more for work AI can start and can't finish: complex builds, judgment calls, accountability, and cleaning up what the model got wrong.

That splits sellers cleanly. If your service is a task, you're competing with a free tool that got better this year. If your service is an outcome a client can't get from a prompt, the spend-per-buyer trend is moving toward you. The same divergence runs through Upwork's data, where freelancers who use AI report earning more even as pay for generic AI tasks falls.

The number to watch next quarter is whether spend per buyer keeps rising while buyer count falls. If it does, the market is consolidating into higher-value work, and there's a place to stand. If spend per buyer stalls too, the top is eroding along with the bottom, and the story stops being about position.

Sources

• Fiverr Announces Second Quarter 2026 Results, GlobeNewswire, July 29 2026
• Fiverr Q2 Earnings: Revenue Falls 10% to $97.8M, StockTitan, July 29 2026
• Fiverr tumbles 17% on earnings, revenue miss, and weak guidance, Investing.com, July 29 2026
• Why Fiverr Stock Is Falling Hard Today, The Motley Fool, July 29 2026
• Fiverr Deepens eCommerce Solutions with Acquisition of AutoDS, Fiverr Investor Relations

This is not financial advice.