Manifesto · on pricing, hours and AIManifesto · on pricing, hours and AI

A €10,000 floor, and why.

This is the long answer to the question we get most — why won't you host small gigs? The short answer: because we think about money the way we think about hours of someone's finite life.

projeto de €10k+ · 10 respostas · €10 para conhecer uma equipa

01

“Good” has gotten slippery.

AI bent the curve in both directions at once — and this is measured, not a mood. Within eight months of ChatGPT, postings for automation-prone freelance work — routine writing, templated code — fell 21% on a major global platform; image-generation tools cut graphic-design postings a further 17%.

The same research found that the jobs which remain in those categories are more complex and pay more. Strategic, integrated, high-trust work did not get cheaper — it got scarcer and more valuable.

So the 2026 question isn't what's a fair hourly rate? It is: am I paying for something a model couldn't do for me, or am I paying someone to operate a model on my behalf? Those are two different products at two different prices.

Data: Demirci, Hannane & Zhu (2023)

Hourly is becoming a trap for both sides.

If someone ships in 8 hours what used to take 40, hourly billing pays them five times less for the same outcome. And measured AI gains are wildly uneven: in the best-known field study, novices got 34% faster while seasoned experts gained almost nothing. An hour has stopped being a comparable unit of value. Fixed scope prices the result; hourly still makes sense where the scope genuinely can’t be fixed yet.

One more data point, against our own hype: in a 2025 randomized trial, experienced developers using AI tools were 19% slower — while believing they were 20% faster. Self-reported AI productivity is not something you can bill by.

Data: Brynjolfsson, Li & Raymond (NBER) · METR (2025)

02

Fair 2026 ranges for a mid-level European team using AI well.

Treat these as floors for real work, not ceilings for cheap work. The ranges are our reading of a moving market, anchored to the published benchmarks under the table — the most arguable numbers on this site, and we'd rather be corrected than vague.

DeliverableRangeNote
Landing page / simple site€500–2,000Down sharply since pre-AI. Commodity work.
Business site (strategy + CMS)€3,000–8,000Roughly unchanged. The thinking didn't get cheaper.
Brand identity€800–3,000You're buying concept and strategy, not vector files.
SEO audit + roadmap€800–2,500Below ~€300, expect largely unedited model output.
SEO retainer€800–2,500 / moUnder ~€600 typically buys automated reporting, not hands-on work.
Digital product / MVP€10,000–30,000Down from €25–60k. Real AI leverage, honestly passed on.
Platform maintenance€2,500–5,000 / moFixes, slices, accessibility and security passes.
Editorial / content retainer€1,800–4,000 / moContinuous publishing and campaign work.

Benchmarks:freelancermap (2026) — DACH avg €103/hMalt (2026) — FR tech €400–700/day

Hourly, if you must benchmarkHourly, if you must benchmark

The global map, as honestly as we can draw it. Price is one factor; timezone overlap, language, retention and handoff cost are the others. Offshore rates are real and can be the right call for well-scoped isolated work — they tend to break on fuzzy scope, complex product and tight review cycles.

  • South / Southeast Asia€10–35 / h5–8h timezone gap
  • North Africa€15–40 / hFR/EN, close to EU hours
  • Latin America€25–70 / hNearshore for US clients
  • Eastern Europe€30–75 / hEU timezone
  • Southern EU€40–95 / hStrong generalists
  • Western / Northern EU€70–120 / hMid-level professionals
  • US / Canada / AU hubs€90–180 / hLocal rates, high cost of living
  • Senior specialists, anywhere€100–250+ / hDeep craft, AI-adjacent engineering
  • Below the regional floorA warningUsually resold model output or hidden subcontracting — ask who does the work

Benchmark:Payoneer (2022) — global avg $28/h, NA ≈ 4.2× S/SE Asia

03

Dinheiro é abstrato. Horas são reais.

€10.000 compram entre 50 e 300 horas de atenção humana qualificada, dependendo de quem contrata. Este é o intervalo honesto. São também pedaços da vida finita de alguém.

50 hrsA little over one focused week of one person.
150 hrsAbout a month of one person, full-time, on you.
300 hrsTwo months of one person — or two weeks of a team of four.
O que custa uma hora, por quem a dáO que custa uma hora, por quem a dá
  • Senior W-EU specialist (10+ yrs)€100–150 / h65–100 hrs at €10k
  • Mid-level W-EU professional€70–90 / h110–140 hrs at €10k
  • Mid-level S-EU professional€50–70 / h140–200 hrs at €10k
  • Eastern EU / nearshore€30–60 / h165–330 hrs at €10k

O comprimento da barra são horas compradas por €10.000 — o mesmo dinheiro, uma quantidade diferente da semana de alguém. Nenhum dos extremos é a resposta honesta por si só: a questão é qual deles consegue levar o seu projeto.

04

Então porquê um mínimo?

Porque abaixo de cerca de €10.000 não há espaço suficiente para uma equipa fazer a parte que faz um projeto funcionar: compreender o problema, discutir a forma, construí-lo adequadamente e ainda estar lá quando falha. Abaixo do mínimo está a comprar horas; acima dele está a comprar um resultado.

Um mínimo é também a única forma honesta de gerir um directory que não cobra percentagem. Não ganhamos nada com o tamanho do seu projeto — publicar é grátis, candidatar-se é grátis, e €10 de cada lado marca a reunião. Se alojássemos trabalhos de €800 teríamos de monetizar volume, e os incentivos inverter-se-iam discretamente.

Porquê €30 depois do lançamento?

Uma apresentação de 30 minutos utiliza cerca de meia hora de tempo profissional de cada lado. €30 fica no limite inferior da faixa de €60–120 por hora indicada acima. Durante o lançamento reduzimos este sinal de compromisso para €10 por lado, até 31 de dezembro de 2026.

Se o seu projeto é genuinamente mais pequeno, isso não é um fracasso — é uma forma diferente de trabalho, e as plataformas na página de panorama são feitas para isso. Preferimos indicá-lo para lá do que aceitar o seu dinheiro mal.

05

The sources.

The studies and datasets this page leans on. Where a number is ours, we say so in place; where it is measured, this is who measured it — and each entry names its year, because this market moves fast enough that the date is part of the data.

  1. Demirci, Hannane & Zhu (2023)Who Is AI Replacing? The Impact of Generative AI on Online Freelancing PlatformsHarvard D3 Institute / CESifo working paperPostings for automation-prone freelance work fell 21% within eight months of ChatGPT (−17% for image work after image-generation AI); the jobs that remain are more complex and pay more.
  2. Brynjolfsson, Li & Raymond (NBER)Generative AI at WorkNBER Working Paper 31161 · 5,179 support agentsAI assistance raised productivity 14% on average — +34% for novices, close to zero for the most experienced workers.
  3. METR (2025)Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer ProductivityRandomized controlled trial · arXiv:2507.09089Experienced developers using AI tools were 19% slower on their own mature codebases — while believing they were 20% faster.
  4. freelancermap (2026)Freelancer Study 2026DACH · 5,412 freelancers surveyed + 340k platform profilesAverage rate €103/h (IT ≈ €100/h) — the first year-on-year decline since the study began.
  5. Malt (2026)Malt Tech Trends 2026Europe · 1M+ registered freelancersTech freelancers in France average €400–700 per day.
  6. Payoneer (2022)Global Freelancer Income Report2,000+ freelancers · 122 countries — the most recent global regional dataset, and dated accordinglyGlobal average $28/h; North American rates ≈ 4.2× South/Southeast Asian rates for comparable work.

Escreva o brief. A pontuação é gratuita e as regras são públicas.

Receberá uma pontuação de 100 e uma lista simples do que falta, antes de alguém o ver e antes de qualquer dinheiro se mover.

Agora abertoAberto

Duas portas à entrada. Entre pela que é sua.