What Happened to My Fiverr Orders

I pulled my full Fiverr earnings export on 3 September 2026. It covers 2019 onwards, although the first order on the account is dated 17 November 2022. Every figure below comes out of that file, and the amounts are what Fiverr credited to my balance in USD.

The short version: order volume peaked in December 2024, revenue peaked in 2025, and 2026 has been the thinnest stretch since the account started earning.

What the export covers

• 291 earning rows, which resolve to 266 distinct orders from 155 distinct buyers
• $12,084.52 earned in total
• 17 different gigs, five of which account for 86% of the money
• 58 platform charges (Fiverr Ads and Seller Plus) totalling $1,388.47
• 148 withdrawals totalling $10,417.52, averaging $70.39 each

That last line describes the business fairly well. Small orders, cashed out in small amounts.

Earnings by year

• 2022 (from 17 November): $88.00 from 5 orders
• 2023: $156.00 from 12 orders
• 2024: $4,578.24 from 116 orders
• 2025: $6,254.28 from 107 orders
• 2026 (to 3 September): $1,008.00 from 26 orders

2025 was my best year on revenue, but it earned about 37% more than 2024 on nine fewer orders. Average order value went from $39.47 in 2024 to $58.45 in 2025. In 2026 it is back down to $38.77, on roughly a quarter of the volume.

So the revenue peak and the demand peak are not the same event. Fewer, larger orders held the money up through 2025 while the order count was already falling.

When orders actually peaked

December 2024 was the best month on both counts: 39 distinct orders and $1,568.80. November 2024 came second on volume with 23 orders and $501.00. That month also carries the only earnings reversal in the whole file, $357.20 on 14 December, so December was not quite as good as the gross figure suggests.

No month since has passed 16 orders. The best of 2025 was October, with 16 orders and $920.80.

Worth being honest about the shape of this: 2024 was not smooth either. July 2024 was a single order worth $400.00, August and September 2024 had none at all, and the longest gap between order days that year was 74 days. Then November and December delivered 62 orders between them. I have had dry patches before. The difference is that this one has not ended.

What 2026 looks like month by month

• January: 5 orders, $136.00
• February: 4 orders, $144.00
• March: 6 orders, $336.00
• April: 1 order, $24.00
• May: 4 orders, $108.00
• June: 4 orders, $212.00
• July: 1 order, $12.00
• August: 1 order, $36.00

Nothing since 11 August, which was 23 days before I pulled the export.

Against the same window last year (1 January to 3 September), 2025 brought $4,225.68 from 64 orders and 2026 brought $1,008.00 from 26. That is 76% down on revenue and 59% down on order count. On a trailing twelve months, September 2025 to August 2026 produced $3,036.60 from 69 orders, against $6,407.48 from 132 orders in the twelve months before that.

The spacing tells the same story. In 2025 there were 89 separate days with an order on them, an average of 4.0 days apart. In 2026 so far there have been 24 such days, 9.5 days apart on average, with a 35 day gap between 27 March and 27 April.

Which gigs stopped selling

My Java Swing and JavaFX desktop gigs are the biggest earners in the file: $6,992.08 across 166 orders, 57.9% of everything I have made on Fiverr. Split by year:

• 2024: $4,139.64 from 92 orders
• 2025: $2,336.44 from 48 orders
• 2026: $300.00 from 11 orders

That is the clearest line in the export. Those orders were small, self-contained desktop apps delivered against a written spec: a form, a table view, a database connection, a login screen, a student project. It is exactly the shape of task an LLM is good at now. A buyer who paid me $24 for a Swing GUI in 2024 can paste the same brief into a chat window and get something that compiles.

Front-end web gigs (writing and debugging HTML, CSS and JavaScript, converting designs, plus one Laravel gig) held up better: $4,030.80 across 76 orders all time. By year that is $438.60 from 24 orders in 2024, $3,020.20 from 37 orders in 2025, and $548.00 from 14 orders in 2026. That last figure is more than half of everything I have earned this year, on gigs that were a rounding error in 2024.

Browser automation gigs brought in $760.00 across 11 orders, of which $600.00 landed in 2025 and $160.00 came from one order in 2026.

I also opened a gig called "fix ai generated code". It has earned $103.84 in total, across two orders, both in 2025, nothing since. Selling the cleanup of the thing that took the work has not, for me, replaced any of it.

What Fiverr Ads and Seller Plus cost

The first platform charge in the file is Seller Plus on 27 January 2024, and the first Fiverr Ads charge is 1 March 2024. Since then:

• Fiverr Ads: $169.65 in 2024, $398.57 in 2025, $121.89 in 2026 to date ($690.11 total)
• Seller Plus and plan fees: $240.16, $313.20 and $145.00 ($698.36 total)
• Combined: $409.81, $711.77 and $266.89

Set against earnings, those combined charges were 9.0% of 2024, 11.4% of 2025 and 26.5% of 2026. Ads alone went from 3.7% of earnings to 12.1%.

April 2026 is the month that made this concrete. I earned $24.00, paid $16.00 in ads and $29.00 in subscription, and finished the month $21.00 down. The export records what I spent, not what the spend returned, so I cannot show a cost per order from this file. What it does show is that the same promotion budget now sits against a much smaller pile of revenue.

New buyers versus returning buyers

First-time buyers, counted by the year of their first order: 80 in 2024, 51 in 2025, 13 in the first eight months of 2026.

Repeat business is not the problem. 56 of my 155 buyers ordered more than once, and those buyers account for $7,438.12 of the $12,084.52 total, just under 62%. People who have worked with me come back. There are simply very few new people arriving.

How I read the numbers

I am not writing this as an anti-AI piece. I use these tools every day and they have made me faster at the client work I still do. But the shape of the decline in my own data is specific and it is hard to read another way.

The work that disappeared is the work that was fully described in the buyer's first message. A brief that fits in a paragraph, with no existing codebase, no ambiguity and no ongoing relationship, is now a prompt rather than an order. My Java desktop gigs were almost entirely that, which is why they fell hardest and first.

What is still selling is the opposite: debugging somebody's existing code, converting a design someone actually owns, and automation work where the requirements come out in conversation. It is a smaller pool and it is worth less per month than what I lost, but it has not gone to zero.

One thing the export cannot settle: it records what I earned, not what I bid on or how much attention I gave the account. I cannot cleanly separate falling demand from my own focus shifting to direct clients and my own projects. What it can show is that I kept paying for ads all the way through 2026 and still watched new buyer counts fall from 51 to 13.

What I am doing with the account

Keeping it, with the spending trimmed to match what it now returns. The gigs that still convert are the web debugging and design conversion ones, so those are where any promotion budget goes. I am not going to keep paying to advertise desktop GUI work that has produced $300.00 in eight months.

Beyond that, the honest conclusion is a boring one. Fiverr paid for a real part of the last four years and it is now one channel out of several, not the channel. I will pull the same export next year and add the numbers to this post.