Flag Management & Exclusion

Flags from speeder, straight-liner, AI agent, and bot detection are all managed in one place: the Flags list on the monitoring screen. This page explains the flag lifecycle and how to exclude confirmed fraud from tabulation.

Reviewing flags

1. Open the project monitoring screen
2. Select the Flags tab
3. Filter by detection type (speeder / straight-liner / AI / bot)
4. Click a row to see the full raw response

Flag states

Each flag has one of three states:

• pending — the default state immediately after auto-detection
• confirmed — a reviewer has determined it is fraud
• dismissed — a reviewer has determined it is a legitimate response

Confirm / Dismiss buttons

Each row has two action buttons:

• Confirm → moves the flag to confirmed
• Dismiss → moves the flag to dismissed
• You can change the state back and forth at any time

Batch operations are also supported: select multiple rows to confirm or dismiss them all at once.

Exclude fraud toggle

The GT and cross tabulation screens both have an Exclude fraud toggle.

• Toggle ON: responses with any confirmed fraud flag are excluded from tabulation
• Toggle OFF: all completed responses are included
• pending and dismissed flags are not excluded — only confirmed fraud is

The toggle is independent of the review workflow, so you can preview "what the results would look like" with fraud excluded at any time.

Detection notifications

When a new fraud flag is detected, an email notification is sent to project administrators. Notifications are throttled to avoid spamming mailboxes.

Raw data export

Exported raw data (CSV / Excel) also includes the detection category and flag state as columns, so you can apply your own filtering rules in external tools if needed.