There is no published price list, and we would be misleading you if there were. What a project costs depends on your rubric, your acceptance criteria and the shape of your data. What we can be precise about is the structures we use, what drives the number inside each and how the terms are enforced once agreed.
Most engagements combine two, and change shape over time as volume becomes predictable.
Metered live against a prepaid balance, drawing down as tasks are accepted. The most common structure while scope is still moving.
Pilots, variable volume, exploratory work
One scope, one fee, invoiced against agreed milestones. Suits a dataset build with a clear finish line.
Defined deliverables with a deadline
For work that cannot fairly be counted in units: exploratory research, complex review, embedded support of your team.
Research, advisory, embedded work
A dedicated named team at an agreed monthly figure. Usually follows a successful per-task phase once volume is predictable.
Steady ongoing volume
Priced per assignment or per day where collection happens on location. Travel is covered by the project.
On-location capture
Only where a bespoke pipeline has to be built first: a custom tool, a recruited language cohort, a dedicated hub. Quoted as a line item, never folded into the unit price.
Bespoke builds
A range with its workings shown, so you can see which assumption to argue with. It is a planning tool, not a quote.
A planning range, not a quote. Change the assumptions and watch what actually drives the number.
Order-of-magnitude planning figures for scoping conversations, not a quote and not an offer. Real pricing depends on your rubric, your acceptance criteria and the shape of your data, which is why we scope it with you rather than publish a rate card.
The most common worry about outsourced annotation is not price, it is losing control. Here is exactly where control sits at each stage.
A conversation, not a form. We would rather tell you a project is a poor fit for us at this stage than six weeks in.
The terms that usually sit in a PDF nobody enforces are implemented in the platform itself. Both sides see the same clock, the same balance and the same thresholds, at the same time.
Each delivery arrives with a countdown you can see in the portal, configured per contract. If it expires without a decision, the batch is deemed accepted automatically. So contributors aren't left unpaid indefinitely and your team has a predictable SLA in both directions. The clock is enforced in software, not by a rep chasing you.
You reject with a reason tied to the frozen rubric. That routes the work back to the original contributor inside a cure window, so it gets corrected rather than silently rewritten by someone else. If your rejection rate crosses a configured threshold, a calibration trigger fires. Usually a sign the rubric needs clarifying rather than the workers.
It varies by engagement, and there is no single published rate card. Depending on what you are running we price per accepted task, per project, per hour, per delivered batch, per field assignment or as a monthly retainer for a dedicated team. Most engagements combine two of these. A pilot priced per task, then a retainer once volume is steady. What fits your work is agreed with our team before anything starts, and the agreed structure is what the portal then meters against.
Sometimes, and it depends on how much has to be built before the first task can run. Work that reuses an existing pipeline usually has none. Work that needs a bespoke rubric, a custom tool, a recruited and trained language cohort or a dedicated onsite hub carries a one-off setup or onboarding fee covering that build. Where a fee applies we tell you what it covers and what it buys before you commit, and it is quoted as a line item rather than folded into the unit price.
Per accepted task work runs on a prepaid balance that draws down as tasks are accepted, visible in the portal in real time with a low balance warning before it becomes a problem. Deemed accepted work is billed the same way, because we do not treat an expired clock as free labour. Project, hourly, retainer and field arrangements are invoiced on the schedule in your contract, monthly or against milestones, rather than metered from a balance. Both settle through the same commercial agreement.
Yes, and most clients do. Engagements commonly start per task while scope is still moving, then shift to a project fee or a monthly retainer once the work is predictable and a stable team is in place. Changing the structure is a contract amendment agreed with our team, not a support ticket and it takes effect from an agreed date rather than retroactively.
Yes. Set an allow list or a deny list per project and it's enforced at task assignment, not just documented in a contract. Presets exist for common compliance regimes and a platform-wide sanctions layer applies underneath your settings regardless.
No. Clients never see contributor identities or pay rates. That separation is structural, not a setting. You see the work, the quality metrics and the delivery timeline. We handle the workforce.
Dataset packaging exports your accepted deliveries with SHA-256 checksums via signed download URLs, with every export logged. You can also pull work programmatically through the API with an issued key.
Tell us the work and we will come back with a structure, a number and a timeline.