<!-- How to Build a Dating Trust and Safety Team | https://whitelabeldating.com/guides/trust-safety-team-building/ | Run by High Intent Media, which also runs Dating Partners and DatingAPI. -->

# How to build a trust and safety team for a dating site or app

> A dating trust and safety team writes the rules, reviews profiles and reports, handles appeals and police requests, and keeps records. Start with one named accountable person and a moderation function that can remove illegal content quickly, which Ofcom expects of every service. Add written policies, targets, training and quality checks as risk and volume grow.

## What does a trust and safety team do on a dating service?

It writes the rules, applies them to profiles, photos, messages and reports, and answers for the results. On a dating service that means fake profiles, scams, harassment, reports of harm after a date, and police requests.

The Trust and Safety Professional Association (TSPA) describes twelve functions in a large company, from content policy to threat research. A dating brand that runs its own software can fold them into four jobs.

| Job | What it covers | In a small team |
|---|---|---|
| Policy | The rules, how to interpret them, and the wording members see | The accountable lead |
| Operations | Daily review, appeals, training, quality checks and vendor management | Moderators and a team lead |
| Tools and data | Review queues, automated flags, measurement and reports | A developer or analyst, part time |
| Legal and police response | Legal requests, urgent escalations and regulator contact | The lead, with outside legal advice |

Our guide to the [content moderation workflow](/guides/content-moderation-workflow/) follows a report through these jobs. This page is about the people.

## What do regulators expect of the team?

They expect someone to be accountable, and trained people to do the review work. The detail grows with the size and risk of the service.

| Source | Applies to | What it expects |
|---|---|---|
| Ofcom illegal content codes | All services in scope | A named individual accountable for illegal content safety, reporting and complaints. A moderation function that can review suspected illegal content and take it down swiftly |
| Ofcom illegal content codes | Large services, and smaller services with several risks | Written statements of responsibilities, a code of conduct, compliance training, internal content policies, performance targets, rules for prioritising, enough resourcing, and training and materials for moderators |
| EU Digital Services Act, Article 12 | All services offered in the EU | A contact route for members that does not rely solely on automated tools |
| EU Digital Services Act, Article 20 | Online platforms above small size | Complaint decisions taken under the supervision of appropriately qualified staff, and not solely by automated means |
| EU Digital Services Act, Article 15 | Services above small size | A yearly report that describes the training and assistance given to moderators |

Ofcom describes a large service as one used by more than 7 million people in the UK each month on average. A smaller service counts as "multi-risk" when it is assessed as medium or high risk for two or more kinds of illegal harm. A dating service that scores that way for both fraud and harassment would qualify, so check your risk assessment against the second row.

Our guides to the [Online Safety Act](/guides/online-safety-act-dating/) and the [DSA for dating sites](/guides/dsa-dating-compliance/) cover each regime. This page is general information, not legal advice.

## How many people does a dating trust and safety team need?

Enough to review your real volume within your target times, during the hours you promise, in the languages your members use. We have not found a reliable published ratio of moderators to members, so work it out from your own figures. The TSPA suggests the questions:

- **Volume.** How many items need a person each week: new profiles, photos, reports and appeals?
- **Review time.** How long does an average item take, and how long does a hard one take?
- **Turnaround.** How quickly must each kind of item be decided?
- **Hours.** Does cover need to be all day every day, weekdays only, or as needed?
- **Languages and markets.** Which ones must the team read and understand?
- **Depth.** Does the work mean reading private messages and investigating, or checking single photos?

Multiply volume by review time, add the hours of cover, and you have a first staffing figure. Then check it against risk. The Digital Trust and Safety Partnership lists using risk assessments "to determine allocation of resources" among its best practices, and Ofcom's codes include a measure on resourcing moderation for riskier services.

One practical floor applies to every team: no single person should be the only one who can remove content or answer an urgent report.

## Should moderation be in house or outsourced?

Keep the hardest cases with your own employees, and use a vendor only for work that is high in volume and settled in policy. The TSPA sets out four staffing models and their trade-offs.

| Model | Best for | Trade-off |
|---|---|---|
| Employees | High-risk and high-priority cases, and police escalations | Full control and visibility, but hard to scale for a lean company |
| In-house contractors | Temporary work, or policies that are still changing | Medium control, usually on fixed terms, with some management overhead |
| External vendor | Long-running review at scale under standard procedures | Easy to scale, but you have no hand in hiring the vendor's staff |
| Volunteers | Community-run spaces | Low cost, but consistency is hard and wellbeing support is thin |

Volunteers rarely suit dating, because the work involves private messages.

Mixed models are common. LOVOO, a dating app, says in its 2025 transparency report that it runs in-house moderation, and that images its automated checks cannot clearly classify go to an external review provider.

If you use a vendor, ask:

- Who trains your staff on my rules, and how are rule changes passed on?
- Which hours and languages are covered?
- How is accuracy checked, and can I see the results?
- What wellbeing support do reviewers get?
- Where is member data viewed and stored?

## How should you train moderators and check quality?

Train before a rule goes live, test with practice cases, and review a sample of real decisions.

- **Materials.** The TSPA lists policy language, enforcement guidelines, examples and screenshots of the review tool. It warns that more information than necessary can overwhelm a reviewer or slow them down.
- **Practice tests.** New moderators make decisions on mock examples before they touch live queues.
- **Shadowing.** LOVOO describes onboarding that covers the product and moderation, detailed training in content moderation and security, weekly team meetings and side-by-side practical sessions.
- **Specialist training.** In 2021 Match Group said its brands would work with experts, including RAINN, to train specialised safety teams for the most serious breaches, with the aim of more empathetic responses to reports of harassment, violence and abuse.
- **Quality checks.** The Digital Trust and Safety Partnership lists assessment methods that evaluate policies and operations for accuracy. In practice a lead re-reviews a sample of decisions, and appeals that succeed are fed back to the moderator.

Keep a record of who was trained and when. Under the Digital Services Act, services above small size must describe moderator training in their yearly report. Our guide to the [trust and safety tooling stack](/guides/trust-safety-tooling-stack/) covers the queues and review screens.

## How do you protect the people doing the work?

Limit what they have to see, give them control over how they see it, and make support routine. The Digital Trust and Safety Partnership lists investing in the "wellness and resilience of teams dealing with sensitive materials" as a best practice.

The TSPA describes review tool features built for this:

- Blurring photos by default, so the moderator chooses when to view them.
- Converting images to black and white to reduce the impact of graphic material.
- Changing audio frequency for the same reason.

Beyond tools, rotate people between heavy queues and lighter work, let a moderator hand a case on without having to explain, and do not judge anyone on speed alone. If you outsource, ask the vendor how it does the same.

## Who decides the hardest cases?

A named senior person does, and every moderator should know who that is. Ofcom's codes expect a named individual accountable for illegal content safety, reporting and complaints, and that person sits at the top of the escalation path.

Send these to the senior level straight away:

- Anything suggesting a member is under 18.
- Threats of violence, and reports of assault or stalking.
- Suspected organised fraud.
- Requests from the police or a regulator.

The Digital Services Act adds a legal trigger. A hosting service that becomes aware of information suggesting a crime involving a threat to someone's life or safety must tell the police or judicial authorities promptly. Our guide to [law enforcement requests](/guides/law-enforcement-requests/) covers what to keep and how to respond.

## Do you need your own team on a white label platform?

No. On a white label network the platform employs the moderators, because members are shared across brands and one team has to apply one set of rules.

Dating Partners, our sister company run by the same group as this site, handles moderation, fraud screening and support for the dating websites on its network. It lists four layers of member checks: identity checks, AI scam detection, human review of every profile and ongoing behaviour monitoring. Partners do not need their own moderation team. Dating Partners is for websites only. Apps run through DatingAPI, also part of our group, which runs safety and moderation as part of its platform. Our [safety and moderation](/platform/safety-and-moderation/) page describes what the platform does.

A white label does not fit if you want to hire your own moderators or write your own rules. For that you need your own software.

Ask any provider:

- Who is the named person accountable for safety?
- During which hours, and in which languages, do people review reports?
- Which work is done by employees, and which by a vendor?
- How are moderators trained, and how is accuracy checked?
- If a member or the police contact my brand directly, where do I send it?

## Common questions

### Can customer support staff also do moderation?

In a small team they often do, but treat them as two skills. Moderators need the content policy, practice cases and quality checks, and the hardest cases should go to a named senior person.

### Do moderators need to speak every language on the site?

They need to cover the languages your members write in. Very large platforms must report their moderators' language skills under the Digital Services Act, which shows how seriously regulators take it.

### Can AI replace a moderation team?

No. Automated tools sort and flag, but the Digital Services Act says complaint decisions must be supervised by qualified staff and not made solely by automated means.

### Who should the head of trust and safety report to?

To someone senior enough to change the product. Ofcom's codes expect a named individual to be accountable for illegal content safety, reporting and complaints.

### Is there a standard ratio of moderators to members?

We have not found a reliable published one. Size the team from your own volumes, review times, opening hours and languages.
