Six Customer Segments

TrustLens divides the 0–100 trust score into six segments. The segments are how you’ll think about customers day-to-day — segment is what shows on the orders list, on the dashboard, and in automation rule conditions. This guide describes each segment in detail, when customers move between them, and the operational treatment recommended for each.

---

The Six Segments #

Segment	Score Range	Color	Default Treatment
VIP	90–100	Green	Reward, fast-track, protect from friction
Trusted	70–89	Blue	Normal processing — your reliable base
Normal	50–69	Gray	Normal processing — neutral, average customers
Caution	30–49	Yellow	Monitor; consider optional friction; no auto-block
Risk	10–29	Orange	Hold for review; require verification before fulfilling
Critical	0–9	Red	Manual approval or block

The thresholds have no Settings screen; developers can change them with the trustlens/segment_thresholds filter. The defaults match how TrustLens is documented throughout the admin UI.

You don’t have to memorize these. The Customers page includes a built-in “What do the segments mean?” legend — a collapsible glossary of all six segments with their score bands and a short description of each — and inline ? tips on the Card-Testing Defense page explain terms like “fingerprint” right where you encounter them.

---

VIP (90–100) #

Who’s In It #

The top tier. Customers reach VIP through one of two paths:

• Allowlisted. Toggling allowlist on locks any customer at exactly 100 and assigns VIP. This is a kill-switch — useful for customers you never want flagged, regardless of behavior.
• Earned. A customer can organically reach the 90–100 range through long tenure (+15 age bonus), high completed-order count, low refund rate, clean chargeback history, and no abuse signals across modules.

What VIPs Get #

• Green badge wherever they appear in the admin
• Card-Testing Defense bypass by default — established customers skip velocity checks at checkout (configurable; default on). This isn’t tied to the VIP segment itself: it covers anyone with at least the Full-confidence Order Threshold of orders (default 3) who isn’t in Risk or Critical, so most VIPs qualify
• Excluded from many automation rules by default (you typically condition rules on “segment is not VIP”)
• Still scored on every refund — VIP doesn’t suppress the Returns signal, so an unusual refund spike still shows up

Recommended Treatment #

• Reward loyalty: dedicated VIP discount tier, early access, free shipping
• Never apply friction (verification holds, manual review) automatically
• Treat VIP refund requests as automatic approvals unless something extreme is happening
• Watch for VIPs dropping out of the segment — that’s a strong leading indicator something has changed

How Customers Leave VIP #

Either (a) you remove them from the allowlist, or (b) they accumulate enough negative signals to drop below 90 organically. The most common organic drops: a filed dispute, a sudden refund spike, or being detected as linked to a risky account. The allowlist insulates against all of these.

---

Trusted (70–89) #

Who’s In It #

Reliable repeat customers. Typically multiple completed orders, low single-digit refund rate, account tenure of three or more months, no abuse signals. Most healthy stores find 20–40% of their scored customer base lands here.

Recommended Treatment #

• Normal processing — no friction, no special handling
• Eligible for standard loyalty programs and email marketing
• Customers who frequently move between Trusted and VIP are good candidates for proactive allowlisting

How Customers Leave Trusted #

Either drift upward into VIP (with continued clean behavior plus accumulating age bonus) or drift downward into Normal (a single refund or dispute can drop a Trusted customer 5–20 points depending on severity).

---

Normal (50–69) #

Who’s In It #

The default segment. Three populations live here:

1. Genuinely average customers. Some refunds, some orders, nothing exceptional in either direction.
2. New customers below the Full-confidence Order Threshold. Anyone with fewer than 3 orders (configurable default) carries the “Limited local history” system signal. Most modules wait until they reach the threshold, so they usually sit at or near 50 — but a confidence-weighted return penalty, a linked-account penalty or the account-age bonus can still move them, including out of Normal.
3. Recovered customers. A customer who was previously in Caution and is rebuilding trust through clean orders may land in Normal.

Recommended Treatment #

• Normal processing
• The largest segment in most stores — should be the default in automation rules
• No automatic friction; no allowlisting needed

Distinguishing Sub-Populations #

To tell apart “average customer at 55” from “new customer at 50,” check the signal breakdown on the customer detail page. The “Limited local history” signal is the giveaway for sub-threshold customers.

---

Caution (30–49) #

Who’s In It #

The early-warning segment. Customers here have at least one meaningful negative signal — an elevated return rate (above 25%), a coupon-then-refund pattern, a country mismatch, a burst of new shipping addresses. Nothing dispositive yet, but worth watching.

Recommended Treatment #

• Don’t block or hold orders by default in Free
• Optional: surface these customers in a weekly review queue
• In Pro, common automation patterns: fire a webhook alert (for example to Slack), flag the customer for review or add a note to their profile, or send an email to a manual review inbox
• Pay attention to direction — Caution trending toward Risk needs faster intervention than Caution trending back toward Normal

What Drives Customers Into Caution #

Most commonly:

• Return rate crossing 25% (up to -10, scaled by sample-size confidence)
• A first refund where the customer also used a first-order coupon
• Shipping address velocity (changes too frequently)
• Country mismatch between billing and shipping
• One linked account, no further signals

---

Risk (10–29) #

Who’s In It #

Multiple negative signals adding up. Either one severe signal (a lost dispute, a card-testing device link, a critical-tier return rate) or several moderate signals compounding. The behavior pattern is consistent with active abuse.

Recommended Treatment #

• Hold orders for review — common automation action in Pro
• Require verification (Pro automation action on order triggers) — holds the order for checks without blocking the customer
• Manual moderation before issuing further refunds
• Tighter coupon eligibility (block first-order coupon use)
• In Pro, automation rules typically fire alerts when a customer enters Risk

What Drives Customers Into Risk #

• Return rate crossing the high threshold — 40%, or 30% when that is 2× your store’s return rate (up to -20, scaled by confidence)
• Wardrobing pattern (2+ delivered orders fully refunded with item-return evidence, up to -8 or -12 added to the return penalty)
• Two or more refunded coupon orders (-15 to -25 from Coupons module, plus -10 if a first-order coupon was involved)
• Linked to an account in Risk or Critical (-10 from Linked Accounts by default; more if the linked account is blocked)
• A filed but pending dispute
• Compound effects across two or three modules

The Reversibility Question #

Risk-tier customers can return to Normal if they place several clean orders without further negative events. The account-age bonus continues to accumulate, and as more orders dilute the return rate, the Returns signal weakens. This recovery typically takes 3–6 months of clean behavior.

---

Critical (0–9) #

Who’s In It #

Confirmed multi-module abuse. Typically:

• Very high return rate (60%+) plus repeated full refunds after delivery
• A lost dispute plus existing return abuse
• Fraud-ring members linked to multiple already-Critical accounts
• Card-testing-attack-tied fingerprints with subsequent customer activity

Internally these customers often score significantly below 0 (e.g. -45) before being clamped to 0. The clamp means you can’t distinguish “bad” from “very bad” by raw score in this range — look at the signal breakdown for the magnitude.

Recommended Treatment #

• Block at checkout (requires the master checkout-blocking toggle enabled)
• Manual approval required for any future orders
• No coupon eligibility
• Refund denials supported by the signal breakdown as evidence
• In Pro, generate a Dispute Evidence Report if a chargeback is filed — the report bundles the trust score, all signals, full order history, and behavioral analysis for processor submission

The False Positive Question #

Critical is the tier where false positives matter most. Before blocking a Critical customer, always:

1. Open the Customer Detail page and read every signal
2. Confirm the signals are real (e.g. the linked-accounts link is to a customer you’d also flag)
3. Check whether the customer is actually a known high-value customer who triggered an unusual edge case

If you find a legitimate customer in Critical, allowlist them — the allowlist locks them at 100 and prevents future false positives. Don’t just unblock; that doesn’t fix the underlying score.

---

Segment Transitions #

When a customer crosses a threshold, TrustLens fires the trustlens/segment_changed action with the old and new segments. Pro automation rules subscribe to this to trigger segment-change behaviors.

Promotion (Upward Transitions) #

From	To	Typical Trigger
Caution	Normal	Refund rate falls back below 25% as new clean orders dilute the rate
Normal	Trusted	Clean-order bonus growing (+10 at 5 clean orders, +15 at 10) combined with the tenure bonus
Trusted	VIP	1-year tenure bonus pushes score above 90, or allowlist toggled on

Demotion (Downward Transitions) #

From	To	Typical Trigger
VIP	Trusted	A refund or coupon abuse signal in a non-allowlisted VIP
Trusted	Normal	A dispute, or refund rate spiking
Normal	Caution	Return rate crossing 25% threshold
Caution	Risk	Return rate crossing 40%, or a second negative signal
Risk	Critical	Dispute filed, or linked-account fraud-ring detection

---

Segment Distribution Targets #

There’s no universal “right” distribution — it depends on your industry, return policy, and customer base. But a healthy store usually looks roughly like:

Segment	Healthy Range	If Significantly Higher
VIP	2–10%	You may be over-allowlisting; check for over-broad rules
Trusted	20–40%	Excellent — strong repeat base
Normal	40–70%	Mostly fine; can indicate min-orders threshold is too high
Caution	5–15%	Tune thresholds if very high — may be over-sensitive
Risk	1–5%	If > 5%, you have either real abuse or false positives — investigate samples
Critical	< 1%	If > 2%, suspect a fraud ring, a leaked coupon, or a wardrobing community

The Dashboard’s segment distribution row shows your current percentages. Comparing them against these ranges is a quick health check.

---

Industry Variations #

Industry	Common Distribution Quirks
Apparel / Shoes	Higher Caution share (often 15%+); raise return-rate thresholds since high return rates are baseline for the industry
Electronics	Lower refund rates but higher dispute exposure; chargeback signals dominate
Beauty / Cosmetics	Coupon abuse dominant; new-customer fraud rings common
Home goods	Lower velocity; long-tenure VIPs common
Digital / subscription	Refund signals weaker; chargeback and card-testing dominate

---

Per-Segment Automation Patterns (Pro) #

Common Pro automation rules per segment:

Segment	Common Rule
VIP	If order placed, tag the customer “VIP” for fulfillment fast-track
Trusted	Usually no rules — normal processing
Normal	Usually no rules
Caution	If order placed, add a note to the customer profile; the Daily Digest notification sends ops a daily summary
Risk	If order placed, hold for manual review; fire a webhook alert (e.g. to Slack)
Critical	If order placed, hold + alert + require manual unhold; block customer if dispute filed

See Automation Overview for the full rules engine.