> ## Documentation Index
> Fetch the complete documentation index at: https://docs.repdata.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Research Defender in Research Desk

> How SEARCH, ACTIVITY, and REVIEW protect your surveys before respondents enter your instrument.

Every respondent entering a Research Desk project passes through **Research Defender** fraud and quality checks before reaching your survey. This happens automatically for Buy Sample, Manual, and Automated sources — no separate integration required.

## Respondent flow

```
Supply Sources → SEARCH → ACTIVITY → REVIEW → Your Survey
```

Legitimate respondents experience minimal friction. Fraudulent respondents are detected, terminated, and blocked from your survey instrument.

For API-level integration details, see the [Research Defender API documentation](/research-defender/index).

## SEARCH — Digital fingerprinting

SEARCH runs in the background as respondents travel from their supply source into Research Desk. It combines proprietary technology and third-party anti-fraud services to identify e-commerce and ad-tech fraud patterns.

### Thresholds

| Setting                      | Value                                  |
| ---------------------------- | -------------------------------------- |
| **Recommended Threat Score** | 29                                     |
| **Typical range**            | 19–40                                  |
| **Lower score**              | Blocks more traffic (tighter security) |
| **Higher score**             | Blocks less traffic (looser security)  |

### Common termination reasons

| Code | Reason                                                   |
| ---- | -------------------------------------------------------- |
| 02   | Duplicate entrant (IP or digital fingerprint)            |
| 03   | Emulator usage                                           |
| 04   | VPN usage (combined with other fraud signals)            |
| 05   | TOR network detected                                     |
| 06   | Public proxy server                                      |
| 07   | Web proxy service                                        |
| 08   | Web crawler activity                                     |
| 09   | Internet fraudster                                       |
| 10   | Retail/ad-tech fraudster                                 |
| 11   | Subnet spoofing                                          |
| 12   | Recent abuse                                             |
| 13   | Duplicate survey group                                   |
| 14   | Navigator Webdriver (Selenium, Puppeteer, Playwright)    |
| 15   | Developer tools open                                     |
| 16   | WebRTC IP detection                                      |
| 17   | T-Mobile proxy                                           |
| 18   | MaxMind minFraud flag                                    |
| 19   | Batch fraud entrants                                     |
| 20   | ChatGPT/Claude emulator                                  |
| 21   | Previous explicit bad behavior                           |
| 22   | Obfuscation mechanisms (incognito, virtual environments) |

<AccordionGroup>
  <Accordion title="Full SEARCH termination code reference">
    SEARCH evaluates dozens of fraud signals in real time. Termination codes appear in your traffic report and Defender export. Elevated Defender terms during soft launch may indicate targeting that attracts professional survey takers, or specs (LOI/IR) that do not match the actual survey experience.
  </Accordion>
</AccordionGroup>

## ACTIVITY — Ecosystem activity tracking

ACTIVITY measures a respondent's survey-taking rate across the market research ecosystem. High volume correlates with poor data quality.

| Setting                   | Value               |
| ------------------------- | ------------------- |
| **Recommended threshold** | 50                  |
| **Typical range**         | 30–100              |
| **Lower threshold**       | Blocks more traffic |
| **Higher threshold**      | Blocks less traffic |

## REVIEW — Open-end quality check

REVIEW presents a standalone web page with an open-end question before your survey. It uses:

1. **AI response detection** — Flags patterns consistent with AI-generated text
2. **Contextual analysis** — Flags answers irrelevant to the question asked
3. **Text analytics** — Scores grammar, length, profanity, copy/paste, and engagement
4. **Honeypot** — Hidden field visible only to bots
5. **DevTools detection** — Flags respondents with inspector panels open

| Setting                   | Value               |
| ------------------------- | ------------------- |
| **Recommended threshold** | 25                  |
| **Typical range**         | 20–70               |
| **Lower threshold**       | Blocks more traffic |
| **Higher threshold**      | Blocks less traffic |

Open-end questions used in REVIEW are maintained in the **Quality** library in Research Desk navigation. Do not modify library content.

## Monitoring Defender outcomes

During fieldwork, watch **Defender terms** (security outcomes) alongside completes and terminates:

| Signal                             | Likely cause                                  | Action                                  |
| ---------------------------------- | --------------------------------------------- | --------------------------------------- |
| High Defender terms, low completes | Professional survey takers blocked            | Review targeting; check LOI/IR accuracy |
| Sudden Defender spike              | Possible fraud attack or bad supplier traffic | Pause source; contact support           |
| Elevated REVIEW terms              | Open-end quality failures                     | Review REVIEW question relevance        |

See [Managing Fieldwork](/research-desk/guides/05-managing-fieldwork) and [Troubleshooting](/research-desk/support/troubleshooting).

## Adjusting thresholds

Defender thresholds are configured at the account level by Rep Data. Contact [clientsuccess@repdata.com](mailto:clientsuccess@repdata.com) if you need threshold adjustments for a specific study design.

## Related guides

* [Best Practices](/research-desk/guides/08-best-practices)
* [Exporting and Reporting](/research-desk/guides/12-exporting-and-reporting) — Defender export
* [Research Defender API](/research-defender/index) — Direct API integration