Auto Escalation
2 min read
Auto-escalation is the Pro feature that triggers Panic Freeze automatically when an attack spreads across multiple device fingerprints faster than per-fingerprint lockouts can contain it. It’s the answer to distributed botnet attacks: when one bot trips a lockout, the next bot in the rotation just takes over. Auto-escalation watches for that pattern and freezes the entire checkout until you can intervene.
How It Works #
Auto-escalation maintains a separate counter that tracks distinct fingerprints that have hit velocity thresholds within a sliding window. When the counter crosses the trigger threshold, Panic Freeze activates automatically.
| Default Setting | Value |
|---|---|
| Trigger threshold | 3 distinct fingerprints |
| Window | 10 minutes |
| Panic Freeze duration | 15 minutes (standard) |
So: if 3 different fingerprints all hit velocity lockouts within any 10-minute span, auto-escalation fires. That’s a strong signal of a distributed attack, since legitimate traffic rarely produces 3 simultaneous high-velocity decliners.
The Geo-Diversity Safeguard #
Before firing, auto-escalation runs the geo-diversity safeguard to confirm the pattern looks like an attack rather than legitimate viral or flash-sale traffic. See Geo-Diversity Safeguard (Pro) for the full logic. In summary: if the declining fingerprints are spread across ≥10 countries with no single country dominating, the safeguard blocks the escalation.
This prevents the worst-case failure: auto-freezing the store on Black Friday because a viral hit spiked decline counts naturally.
What Happens When It Fires #
- Panic Freeze activates with a 15-minute duration
- An admin banner shows the freeze with “Auto-escalated” as the trigger
- Slack and email alerts fire (if configured) with the
auto_escalatedevent type - The Card Testing event log records the escalation with the contributing fingerprints
- The freeze can be cancelled manually like any Panic Freeze
Tuning the Threshold #
The 3-fingerprint default is conservative — it triggers on relatively small distributed attacks. Tune up or down depending on your store’s profile:
| Threshold | When to Use |
|---|---|
| 2 fingerprints | Very aggressive — recommended only after observed false negatives at default |
| 3 fingerprints (default) | Balanced |
| 5 fingerprints | Loose — recommended if you have high natural decline diversity (e.g. retry-heavy gateway, large geographic customer base) |
The window can be adjusted between 5 and 30 minutes; the default of 10 minutes works for most stores.
Configuration #
Settings: TrustLens → Settings → Modules → Card Testing → Auto-Escalation.
| Setting | Default | Description |
|---|---|---|
| Auto-escalation enabled | On | Master toggle |
| Fingerprint trigger threshold | 3 | Distinct fingerprints that hit velocity within the window |
| Window duration | 10 minutes | Sliding window for counting fingerprints |
| Geo-diversity safeguard enabled | On | Block escalation if declines look geographically distributed |
When Auto-Escalation Won’t Fire #
Even with auto-escalation on, the system won’t fire if:
- Geo-diversity safeguard determines the pattern looks legitimate
- Fewer fingerprints have triggered than the threshold requires
- Panic Freeze is already active (no duplicate freezes)
- Card-Testing Defense module is disabled
Alerting #
The auto_escalated event fires Slack and email alerts to all configured recipients. The alert includes:
- Trigger time
- Number of contributing fingerprints
- Top decline codes seen in the window
- Geographic distribution summary
- Direct link to the Card Testing page
This is one of the highest-priority alerts the system produces — if you’re going to wake up for any TrustLens event, this is the one.
After the Escalation #
Use the freeze time to:
- Review the Attack History tab — see the contributing fingerprints, decline codes, hourly chart
- Check whether the attack continues after the freeze lifts (often it shifts to a different vector)
- Consider tightening per-fingerprint velocity thresholds if the attack used the slow-paced pattern
- Document the attack pattern for post-incident review
Auto-escalation is reactive, not preventive. If you’re seeing repeated auto-escalations, the underlying attack pattern is recurring — adjust the per-fingerprint defenses to catch attacks earlier rather than waiting for the distributed signal.
Auto-Escalation vs Manual Panic #
Both result in the same Panic Freeze state. The differences:
| Aspect | Manual Panic | Auto-Escalation |
|---|---|---|
| Trigger | Admin clicks button | System detects distributed attack |
| Reaction time | Depends on admin presence | Within seconds of pattern detection |
| Audit reason | Manual / [admin username] | auto-escalation |
| Geo-diversity safeguard | Doesn’t apply | Required to fire |
Best practice: leave auto-escalation on, treat it as a backup. Use manual panic for known-bad situations where you have specific evidence the system hasn’t picked up yet.