Outsource
Content moderation outsourcing for listings, reviews and user reports
Outsource non-graphic content moderation to a trained team in your tools: listings, reviews, profiles and user reports. $5 per person-hour, QA included.
$5 per person-hour for standard tasks.
Content moderation outsourcing hands the daily review of user content to a trained team that applies your policy. Deepen AI takes on non-graphic moderation: marketplace listings, reviews, profiles, comments and user reports, worked inside your own moderation tools. We do not review graphic or violent material.
What the work is, and who buys it
Buyers are marketplaces, classifieds and review sites, community apps, and software products where users post content. The queue grows faster than the trust and safety team. Most of it is routine: spam, scams, prohibited items, fake or off-topic reviews, phone numbers posted in public. The hard part is applying the policy the same way every time, and noticing new abuse patterns before they spread.
Moderation buyers also ask where user data goes, and they should. Our people work inside your tools, on accounts you control and can switch off. Deepen AI holds SOC 2 Type II, ISO 27001 and TISAX, and the certification scope covers the Deepen AI services delivery team. Every moderator is a known, vetted member of the Deepen AI team who completes security training, not an anonymous crowd worker. Details are on the security page.
What our team does
- Review new listings against your prohibited-items and listing rules, then approve, reject or escalate
- Check user reviews for spam, copied or incentivized text, off-topic content and personal details
- Work the user-report queue and apply the action your policy sets
- Review profile names, bios and usernames against your community rules
- Tag posts, articles and listings by topic for search and recommendations
- Flag unclear items and repeat offenders for your team to decide
- Log policy gaps, meaning cases your policy does not answer yet, for your weekly review
We do not review graphic or violent material. If your queue might contain it, tell us in the pilot form and we will agree routing rules with you. If an item like that reaches our queue anyway, the moderator stops and escalates it to your team.
How it runs
- You share the policy and examples. Send your moderation policy, the actions available in your tool, and 20 to 50 past decisions, including a few your own team argued about.
- We agree scope and train. Our ops team sets the people, hours and coverage window in your pilot plan. Moderators train on your past decisions, and we build a gold set of items with known-correct actions for you to confirm.
- We deliver, with a named team lead and QA sampling. Moderators work the live queue. The team lead handles escalations, keeps a decision list for gray areas, and brings new patterns to your contact.
- You get a weekly report. Items reviewed by action, QA sample results, escalations, new abuse patterns and policy questions.
Illustrative example: one marketplace listing
Illustrative only. The listing and policy are invented.
- Input, listing: "Brand new unlocked smartphone, sealed box, $150. Text me at +1 555 010 4471. No in-app payments."
- Input, your policy: 4.2 No contact details or off-platform payment requests in listings. 6.1 Escalate likely scams to Trust.
- Output, action: Reject
- Output, reason code: 4.2, off-platform contact and payment
- Output, escalation: sent to your Trust queue under 6.1. The price is far below similar listings, and the same number appears in two other listings today.
- Output, report note: third listing this week with the phrase "no in-app payments". Suggest adding it to your keyword filter.
- Who did what: one moderator decided and escalated. The team lead checked the escalation. This item was not in this week's QA sample.
How we keep decisions consistent
In moderation, quality mostly means consistency: the same item gets the same action whoever reviews it. Moderators train on your past decisions, and gold items with known-correct actions are mixed into live work, so each person's consistency is measured on real conditions. QA reviewers re-check a sample of decisions against your policy and the gold set. They do not re-check every decision. Disagreements go to the team lead and into the next training round, and each gray-area ruling is added to the decision list. The weekly report shows how many decisions were sampled and the acceptance rate on that sample. See how we check quality.
Pricing scope
Non-graphic content moderation and tagging is a standard task at $5 per person-hour, with a named team lead and QA sampling included. We price by the hour, not per item, so a policy change that slows review for a week does not reopen the agreement. Work starts with a two-week prepaid pilot: people × hours per week × 2 weeks × $5, with at least 20 hours per person per week. After the pilot, we invoice weekly in arrears. Moderation in a language other than English is quoted after our ops team reviews the task. See the pricing page.
When to use the enterprise team instead
Moderating user content and labeling data to train a perception model are different jobs. If you need camera or sensor data annotated, for example to teach a model what it is seeing, the Deepen AI enterprise team scopes and prices that work per project. See enterprise AI data services.
Send your policy and a sample of past decisions. We email you a pilot plan with a quote.
FAQ
Can you provide 24/7 content moderation?
Tell us the hours and time zones you need in the pilot form. Our ops team states in your pilot plan which hours the team can cover, before you pay anything. We do not promise coverage we have not planned.
What content do you not moderate?
Graphic or violent material. If your queue might contain it, tell us first so we can agree routing rules that keep it with your team. If an item like that reaches our queue anyway, the moderator stops and escalates it. We also do not take phone support.
What should we look for in a content moderation services provider?
Ask how moderators are trained on your policy, how consistency is measured, who decides gray-area cases, and where your user data goes. Our moderators work inside your tools on accounts you control, a named team lead handles gray areas, and QA results appear in a weekly report.
Do your moderators make the final call on bans and appeals?
They apply the actions your policy gives them. Account bans, appeals and anything the policy does not cover can stay with your team. You decide where that line sits, and we write it into the guidelines.
How is content moderation outsourcing priced?
Non-graphic moderation in English is $5 per person-hour, with a named team lead and QA sampling included. We price by the hour, not per item. Work starts with a two-week prepaid pilot, with at least 20 hours per person per week.