sales-skills/sales-prolific

Prolific Platform Help

Prolific (prolific.com) is a developer-native online research-participant panel — it recruits, screens, and pays vetted participants from a 13M+ global pool (130+ countries) for surveys, online tasks, and AI human-data collection at scale. Participants pass ID + live-video + government-document verification and Prolific runs active bot / AI-answer detection + a manual data-quality team, hence its research-grade data integrity reputation.

It's the bring-your-own-instrument panel: you host the survey/task in your own tool (Qualtrics, Google Forms, a custom app, a prototype) and Prolific supplies the humans. Unlike the recruit-only interview marketplace Respondent (/sales-respondent), Prolific's edge is a self-serve public REST API with an instant Token key — plus HMAC-signed webhooks, an MCP server, and an open-source CLI — so recruitment can sit in a real pipeline. It's also the recruitment backbone usability tools plug into (e.g. Useberry recruits via Prolific).

Two things to say almost every time:

• Real people answering ≠ demand. A large, vetted sample is strong signal for problem/behavior research — but "I'd pay" is still not a purchase; keep the why and take a go/no-go to a real behavior test (/sales-idea-validation, /sales-funnel).
• The sticker reward is not the cost. A platform fee rides on top (best-effort 42.8% corporate / 33.3% academic) — budget reward + fee and run the cost calculator first (prolific.com/pricing).

Step 1 — Gather context

If references/learnings.md exists, read it first for accumulated platform knowledge.

Ask only what you can't infer:

1. What do you need?
   • A) Recruit / target — filters, screeners, representative samples, eligible-count checks, quotas
   • B) Study lifecycle — create → cost → publish → monitor → review/approve submissions
   • C) Pay / bonuses — approve, reject, pay bonuses, understand the platform fee + minimum pay
   • D) Automate / integrate — the REST API, webhooks, the MCP server, or the CLI in a pipeline
   • E) AI human data — annotation/eval Batches or original-data Collections via the AI Task Builder
2. Where does the task live? (Prolific supplies participants; your survey/task/app is external.)

Skip-ahead: if the user wants to compare research/recruit tools across the market or the validate-before-building method, that's /sales-idea-validation — route in Step 2.

Step 2 — Route or answer directly

| If the user's question is about… | Route to | |---|---| | Comparing research/recruitment/idea tools across the market, or the validate-before-building method | /sales-idea-validation {question} | | The data-quality-first peer panel — Sentry screening + a support-provisioned API (no self-serve token/webhooks) | /sales-cloudresearch {question} | | Recruiting B2B / hard-to-reach professionals for moderated interviews (a recruit-only marketplace) | /sales-respondent {question} | | An all-in-one recruit + run + repository research suite | /sales-great-question {question} | | Running a usability / IA test (card sort, tree test, prototype tasks) on a built artifact | /sales-uxtweak or /sales-lyssna {question} | | Running a real behavior demand test (smoke-test page, waitlist, pre-sale) after research | /sales-idea-validation or /sales-funnel {question} |

When routing, give the exact command: "This is a {domain} question — run: /sales-idea-validation {original question}"

Otherwise, answer Prolific-specific questions using Step 3.

Step 3 — Prolific reference

Read references/platform-guide.md for the full reference — the capability/automation-surface table (API vs webhook vs MCP vs UI), pricing/plan gates + the fee and minimum-pay math, the Study → Submission → Participant data model, filters/representative samples, the AI Task Builder (Batches vs Collections), integrations, and quick-start recipes. Read references/prolific-api-reference.md for the verbatim API surface (Token auth, base URL, study & submission objects, filters, webhooks + HMAC verification, cost calculator, CLI, MCP). Answer using only the relevant section — don't dump the full reference.

Step 4 — Actionable guidance

• Say the caveat: a vetted sample is not demand. Prolific gets real, verified people to answer (a strong signal — genuine behavior/language) but a stated "I'd pay" is not observed demand. Keep the why; take the go/no-go to a real behavior test (pre-sale, smoke test) → /sales-idea-validation.
• Budget reward + platform fee, and use the cost calculator. Total ≈ reward + platform fee (best-effort 42.8% corporate / 33.3% academic), and Prolific enforces a minimum pay rate (best-effort £6/$8/hr, recommended £9/$12/hr — higher for specialized AI/coding/STEM work). Call POST /api/v1/study-cost-calculator/ with reward + total_available_places to get the VAT-and-fee-inclusive total before publishing. Present all figures as best-effort — confirm at prolific.com/pricing.
• Fix "participants don't fit the brief" with filter/screener design, not the panel. Prolific is ID/video-verified, so off-brief responses are a targeting/screener issue: apply prescreening filters (GET /api/v1/filters/, then a select/range payload), check the eligible count before launch, add custom-screening questions with attention checks, and use quotas to balance the sample. Report genuinely bad submissions for data-quality review.
• Guard against AI-generated open-ended answers. The top data-quality threat now is humans pasting ChatGPT into free-text. Mitigate with behavior-specific prompts, timing/attention checks, and copy-paste detection in your own instrument — then reject or request a return on failing submissions, and reserve bonuses for quality.
• Run the API pipeline — create → publish → approve — self-serve and instant. Create a Researcher token in account settings (no expiry, full permission) and send Authorization: Token <token> to api.prolific.com/api/v1. Flow: POST /studies/ (external_study_url with {{%PROLIFIC_PID%}}, reward in cents, filters, completion codes) → publish via POST /studies/{id}/transition/. A submission pays only once APPROVED — gate on your quality check, then auto-approve valid completions (a completion-code auto-approve action, or POST /submissions/{id}/transition/ APPROVE / bulk-approve). Use webhooks (participant.submission.created, payment.processed, …) and verify the HMAC-SHA256 signature (X-Prolific-Request-Signature over timestamp + body, base64, constant-time) before trusting the body; add a status-filtered reconcile poll (GET /api/v1/studies/{id}/submissions/) so a missed event never leaves a submission unpaid.
• For AI human data, pick Batches vs Collections. Batches = you upload data (CSV), participants evaluate/label it (annotation, model eval, pairwise preference); Collections = participants produce original data (images/files/text). Both attach via data_collection_method + data_collection_id — reach for these for RLHF / preference / eval rather than hand-rolling a task.
• CLI + MCP for exploration/CI. The open-source CLI (prolific-oss/cli, PROLIFIC_TOKEN) and MCP server (docs.prolific.com/_mcp/server) suit scripting; use the raw API for production pipelines.

If you discover a gotcha or tip not in references/learnings.md, append it there with today's date.

Gotchas

Best-effort from research (2026-07) — pricing (platform fee 42.8% corporate / 33.3% academic, min pay ~£6/$8/hr, rec. £9/$12/hr), panel size ("13M+"), and API/webhook specifics all move; verify at prolific.com/pricing and docs.prolific.com.

• The reward is not the total. A platform fee sits on top (best-effort 42.8% corporate / 33.3% academic) and a minimum pay floor is enforced (best-effort £6/$8/hr, rec. £9/$12/hr — low pay also drags quality). Run the cost calculator before publishing so the fee + VAT don't surprise you.
• A Researcher API token has no expiry and full permission. Treat it like a password — never expose it client-side, scope it to a workspace, and rotate if leaked.
• Verify the webhook signature, and don't rely on delivery alone. HMAC-SHA256 over timestamp + body, base64, constant-time compare against X-Prolific-Request-Signature. Also reconcile with a status poll — treat webhooks as best-effort and payments/approvals as the source of truth.
• A vetted sample ≠ demand. 1,000 verified people answering "I'd buy this" is stated preference at scale, not a purchase — take the WTP go/no-go to a real behavior test.
• AI-generated free-text is the new data-quality threat. ID/video verification stops bots, not a human pasting ChatGPT — add attention/timing checks and copy-paste detection in your own instrument.
• "Participants don't fit the brief" is a filter/screener problem, not a bad panel. Prescreen with filters, check the eligible count before launch, add screening questions + attention checks, quotas.
• Prolific supplies people, not the survey. You host the task (Qualtrics/Forms/your app), pass {{%PROLIFIC_PID%}}/{{%STUDY_ID%}}/{{%SESSION_ID%}} in the URL, then return a completion code.

Related skills

• /sales-idea-validation — The validate-before-building method + the full research/recruit/validator tool landscape (a research "yes" is not demand). Install: npx skills add sales-skills/sales --skill sales-idea-validation -a claude-code
• /sales-cloudresearch — The data-quality-first peer panel: the same recruit class, but its edge is Sentry data-quality screening and a support-provisioned X-API-KEY (no self-serve token, no webhooks/MCP — poll to reconcile), where Prolific's is an instant token + HMAC webhooks + MCP + CLI. Pick CloudResearch when screening is the priority, Prolific when a self-serve pipeline is. Install: npx skills add sales-skills/sales --skill sales-cloudresearch -a claude-code
• /sales-respondent — The recruit-only B2B/professional interview marketplace (Prolific's edge is broad online samples + a self-serve API; Respondent's is hard-to-reach professionals + a partner-provisioned API). Install: npx skills add sales-skills/sales --skill sales-respondent -a claude-code
• /sales-great-question — The all-in-one recruit + run + repository suite (Prolific is recruit + collect only; you bring your own instrument and repository). Install: npx skills add sales-skills/sales --skill sales-great-question -a claude-code
• /sales-useberry — Prototype-first usability testing that recruits its participants via Prolific — pair them when you need usability testing on top of Prolific's panel. Install: npx skills add sales-skills/sales --skill sales-useberry -a claude-code
• /sales-funnel — Build the smoke-test / fake-door landing page that turns a research hypothesis into a real demand test. Install: npx skills add sales-skills/sales --skill sales-funnel -a claude-code
• /sales-do — Not sure which skill to use? The router matches any sales objective to the right skill. Install: npx skills add sales-skills/sales --skill sales-do -a claude-code

Examples

Example 1: "How do I create a Prolific study and auto-approve submissions from my own backend?" (developer/automation)

User says: "I want to launch studies and pay people through Prolific's API without clicking around the dashboard." Skill does: Explains the self-serve flow (Step 4's API pipeline) — Researcher token + Authorization: Token, POST /studies/ → publish transition, subscribe to participant.submission.created webhooks, verify the HMAC signature, and APPROVE valid completions (approval releases payment) with a reconcile poll so nothing is left unpaid. Result: The user runs the whole study lifecycle from their backend with the payment gate correct.

Example 2: "Half my Prolific responses look low-effort or AI-written"

User says: "I'm getting open-ended answers that read like ChatGPT — is the panel bad?" Skill does: Reframes it as a data-quality/instrument issue, not a bad panel (Prolific is ID/video-verified with bot/AI detection). Prescribes behavior-specific prompts, attention + timing checks and copy-paste detection in your own survey, tighter prescreen filters with an eligible-count check, and rejecting / requesting returns on failing submissions while reserving bonuses for quality; report genuinely bad submissions for Prolific's data-quality review. Result: The user cleans the sample by fixing screening + attention checks instead of switching tools.

Example 3: "What will an n=300 study actually cost on Prolific?"

User says: "I want 300 participants at a 10-minute survey — what's the real bill?" Skill does: Explains total ≈ reward + platform fee (best-effort 42.8% corporate / 33.3% academic) with an enforced minimum pay (best-effort £6/$8/hr, recommended £9/$12/hr), and shows the POST /study-cost-calculator/ call (reward in cents + total_available_places) that returns the VAT-and-fee-inclusive total in cents before publishing. Presents all figures as best-effort — confirm at prolific.com/pricing. Result: The user budgets reward + fee up front and sizes the study to their spend.

Troubleshooting

"My API calls 401 / where's my API key?"

Symptom: Requests fail auth. Cause: Prolific uses a Authorization: Token <token> header (not Bearer, not two-header), and the token is a Researcher token you create in account settings. Solution: Create a Researcher token at app.prolific.com/researcher/tokens/, send Authorization: Token <token> against https://api.prolific.com/api/v1/. The token has no expiry and full permission — keep it secret and workspace-scoped.

"My webhook fired but I'm not sure it's really from Prolific / I missed some events"

Symptom: Untrusted or missing webhook deliveries. Cause: Webhook bodies must be signature-verified, and delivery is best-effort. Solution: Recompute HMAC-SHA256 over X-Prolific-Request-Timestamp + body with your secret, base64-encode, and constant-time compare against X-Prolific-Request-Signature. Also run a status-filtered reconcile poll (GET /studies/{id}/submissions/) so a dropped event never leaves a submission unreviewed or unpaid.

"A participant says they weren't paid"

Symptom: A completed submission wasn't rewarded. Cause: A submission only pays once it's APPROVED (RESERVED → ACTIVE → AWAITING REVIEW → APPROVED); left in review it stays unpaid, and rejected/returned/timed-out/screened-out submissions don't pay. Solution: Review the submission status via GET /studies/{id}/submissions/, transition it to APPROVED (or bulk-approve), and use a bonus payment for extra compensation. Automate "valid completion → approve" but keep a manual review path for disputes.