Review queue

When a workflow assigns a REVIEW decision to an incoming order or return, it lands in the Review queue in the Pinch dashboard. This is where your CS or fraud-ops team makes the final call. Items that hit ALLOW or DENY automatically bypass this queue.

The Review page has two top-level tabs:

• Orders — orders flagged for Reseller Abuse at checkout time (Suspected Abuse Type: Reseller Abuse). Used when you want to manually clear high-value or limited-release orders before fulfillment.
• Returns — return requests flagged for return abuse (Wardrobing, Empty Box, etc.). The historical primary use of the Review queue.

Both tabs share the same structure (OPEN / CLOSED / BLOCKED sub-tabs, same column shape, same row actions); the only difference is what events feed them.

The three tabs

The Review queue has three tabs across the top:

Tab	Contents
OPEN	Returns waiting on a human decision. This is your working queue.
CLOSED	Returns that were REVIEW-flagged and then manually approved or denied. The full audit trail of every reviewer decision.
BLOCKED	Returns that were REVIEW-flagged and then manually denied. The audit trail of every reviewer denial.

The page title shows the count of returns in the currently-active tab — e.g., Order List (1253) on the Closed tab.

A time range selector in the top right narrows the visible rows: Today, 7D, 30D, 3M, 6M, 1Y, Go Live.

Columns

Every row shows the same set of fields:

Column	What's in it
Order ID	Your platform's order identifier (#4396125). Click to drill into the order.
Return ID	Your RMS's return identifier (109149204). Click to open in your RMS.
Refund Amount	What the customer is requesting back.
Order Amount	What they originally paid.
Customer	Email address. Click to open the customer profile.
Country	Two-letter country code.
Suspected Abuse Type	The pattern Pinch detected — Wardrobing, Empty Box, Copycat, etc.
Abuse Risk	0–100 score (colored: low / medium / high) — the same Pinch Assessment that appears on the customer profile.
Decision	The current state — Pending, Approved, or Denied. (Only visible on the Closed and Blocked tabs.)

On the OPEN tab, each row has two action icons on the right:

• Green checkmark — Approve the return
• Red X — Deny the return

Clicking either records the decision, removes the row from OPEN, and adds it to CLOSED or BLOCKED with the reviewer's identity and timestamp logged.

AI Analysis

Underneath each row is a short paragraph tagged AI: Analysis. This is Pinch's plain-language explanation of why the return was flagged. The model surfaces the strongest evidence, both for and against the abuse hypothesis. Real examples from the queue:

"Mixed signals. The current return has clear fit-based explanations on both items and was initiated just 3 days after purchase, which is consistent with a normal sizing miss, but the customer also shows three suspect returns clustered in March around the same dress/accessory/heel set."

"Red flag detected: this return fits a wardrobing-like pattern more than a standard defect claim. The item is a single occasion-style dress returned after just 4 days, and this customer has 14 returns across 17 orders with repeated fashion-item returns but no disposition evidence confirming damage."

"Mixed signals. The flagged return itself is a fast 5-day return on an $89.90 novelty heel, but the stronger story is a repeat comparison shopper with 17 orders and 13 returns across many categories rather than a concentrated high-value abuse burst..."

A few rules of thumb when reading AI Analysis:

• "Red flag detected" → strong evidence for abuse. Default to deny unless you have offsetting context.
• "Mixed signals" → genuinely ambiguous. Use your judgment plus the customer profile.
• "Standard return with..." → looks normal on its own; the flag came from cumulative behavior.

Clicking into a return

Click the Order ID or Return ID link in any queue row to open the Return Details page. This is a dedicated page with everything needed to evaluate the return — no need to leave the queue, open the customer's profile separately, or cross-reference with your order management tool. The page is organized into five panels.

Order panel

The top-left panel summarizes the order this return is filed against:

• Channel — Online or Instore pill badge
• Timestamp — when the order was placed
• Customer name, email, phone
• Shipping carrier / return method (icons; N/A if not provided)
• First Order date (with a relative age like "over 1 year ago")
• Last Order date (with a relative age)

The First/Last order dates frame the customer's relationship with your brand — a return from someone whose first order was a week ago reads very differently than one from a 2-year repeat buyer.

Location Details panel

Every billing and shipping address Pinch has on file for this customer. Each address row has:

• The address as a single line
• A Street View link
• An icon indicating address type

A Map toggle in the top-right of the panel switches to a map view of all addresses. Useful for spotting customers shipping to forwarder addresses, mules, or many different addresses in a short window.

Pinch Assessment panel

The risk panel that drives the decision. Dark-navy background so it stands out from the other panels.

• Score — 0–100, with a gradient bar visualization (green → yellow → red)
• Suspected Abuse Type — Wardrobing, Empty Box, Copycat, Missing Item, or N/A
• Triggered Workflow — pill badges listing every workflow branch that matched this return. Often more than one (e.g., "Except Admin Users, Influencers, BORIS" plus "High Return Rate Low Value Customer Email").
• Key Risk Indicators — the variables that drove the score, with their current values. Examples on a real customer:

Variable	Value
customer_value_by_email	86.90
orders_placed_from_this_ship_add_in_last_6_months	3.00
num_email_per_ship_address_in_last_180_days	1.00
return_due_to_size_ratio_by_ship_address	0.29
return_rate_by_value_by_email	0.59
customer_value_by_ship_address	96.90
customer_value_by_ip	71.15

• Social — any social profiles Pinch matched to the customer's email or phone (or "No social media profiles were found")

Key Risk Indicators are configurable per merchant

The variables that appear in this panel — and the thresholds that flag a customer — are configured for your account during onboarding. They can be edited any time to match your policy. Talk to your customer success manager to add, remove, or re-tune which indicators surface here.

Items panel

The full line-item breakdown of the order, with a summary header and a return-status pill for each return associated with the order.

The header row summarizes:

• Total items in the order (e.g. "7 items (7 units)")
• Order value ("$400.30")
• Returned units across all RMAs ("2 units returns in 1 RMAs ($129.80)")
• Value kept ("$270.50 kept")

Each return on the order appears as a section with the Return ID, date, and a decision badge (Approved / Denied / Pending Review). Inside each return section:

• Item title (e.g., "IN THE WILD LEOPARD DENIM")
• LINE_ITEM identifier
• Category (BOTTOMS, TOPS, SANDALS, Apparel & Accessories, etc.)
• Return Reason — what the customer said when initiating the return (e.g., "Right item, but in the wrong size", "Item was too big")
• Price × quantity = line total

Items the customer kept are listed below the returned ones — useful for understanding the kept-vs-returned ratio on a single order.

Customer History panel

A row of KPI cards summarizing this customer's entire history with you:

Metric	What it means
True Customer Value	Net lifetime value — revenue minus refunds minus loaded return cost
Total Order Value	Gross revenue from this customer
Total Return Value	Cumulative refunded amount
Returned Items	Number of items they've returned
Ordered Items	Number of items they've ever ordered
Return Rate	Items returned ÷ items ordered, as a percentage

Below the KPIs, two tabs:

• Order History — full table of every order Pinch has on file for this customer, with columns for order ID, date, type, decision, customer info, shipping/billing, IP, device ID, Pinch ID, amount, return amount, damaged return
• Customer Graph — visual graph of every identifier (email, phone, billing, shipping, device IP, device fingerprint) linked to this Pinch Identity Number, useful for spotting multi-account abuse

Right-side controls on the History panel:

• All Orders — toggle between only-returns and all orders
• Pinch ID — filter by linked identifiers
• 1Y time range chip
• Filters — opens a drawer for decision / amount / damage status filters

Ask AI

A floating Ask AI button in the bottom-right opens a chat scoped to this specific return and customer. The assistant has visibility into everything on the page — Pinch Assessment, Key Risk Indicators, order history, graph. Useful for:

• "Why is this customer's score so high?"
• "Has this customer triggered an abuse workflow before today?"
• "What's the strongest single reason to deny this return?"
• "Are there other customers linked to this device?"

The chat closes when you navigate away.

Acting on the return

The approve / deny actions live both on the queue row (inline) and on the Return Details page (via the page header). Either path produces the same outcome — the decision is logged with the reviewer's identity, timestamp, and any Review note left.

Workflow for a reviewer

A typical operator hits the Review queue twice a day, in this order:

1. Filter to the most recent time range (Today or 7D) — fresh returns matter most because customer expectations are timing-sensitive.
2. Sort by Abuse Risk descending — work the riskiest ones first.
3. For each row: read the AI Analysis. If unambiguous, decide inline. If mixed, expand the row and check Customer Information and Recent Orders.
4. For genuinely hard calls, click Open Order Details to see the customer's full history and graph of linked identities.
5. Leave a Review note before approving or denying so the audit trail is meaningful.

The closed and blocked tabs are useful weekly — sample 20 rows from each and verify the AI Analysis matched your team's eventual call. If you find systematic mismatches (e.g., Pinch is flagging too many normal returns as Wardrobing), report them to your customer success manager so the model and workflow can be tuned.

What happens after you decide

Action	What it does downstream
Approve	The return moves to the CLOSED tab. Pinch updates the customer's history. Your RMS reads the customer tag Pinch wrote on the Shopify customer and proceeds with the refund flow.
Deny	The return moves to the CLOSED tab. The same tag-writeback mechanism tells your RMS to refuse the return label or refund.
Leave a Review note	Saves alongside the decision. Visible in the audit trail and to future reviewers.

Once a decision is recorded, it's final from the queue's perspective. To reverse it, you'll need to use your RMS's standard refund-override path; Pinch doesn't re-open a closed review.

See also

• Workflows — where REVIEW outcomes are configured
• Customers — the customer profile linked from each row
• Playbooks — automated follow-ups for customers in risk segments