Return Intelligence
Return abuse, explained — not just flagged. Score every return and refund 0–100, see exactly why, and decide with your team in control.
- Explains every score: each signal, its points, and the raw values
- Confidence is scored separately — low confidence always routes to a human
- No automatic declines — ever, at any score or setting
- Free 12-month Historical Return Leakage Audit
- Pre-ship scoring flags risky buyers before you ship
- Your customers' data is never shared across merchants
The problem
Returns are unavoidable. Return abuse isn't.
Roughly a fifth of online sales come back, and a meaningful share of those returns aren't honest — wardrobing, serial returning, extreme bracketing, “item never arrived” claims against signed deliveries, refunds on merchandise that never ships back, and the double hit of a refund followed by a chargeback. Shopify's native rules handle windows, fees and eligibility — they were never designed to spot behavioural patterns across a customer's history.
- Manual detective work. Your agents export CSVs to figure out whether this customer has done this before.
- No history at the moment of decision. The call gets made in the return queue — where the customer's past 12 months aren't visible.
- Refund and chargeback. You refund in good faith, then the same order arrives as a chargeback. You pay twice.
- Fear of false accusations. Accusing a good customer costs more than the refund — so most merchants just approve everything.
Return Intelligence estimates risk; it does not determine guilt.
Start with a number, not a trial
A free Historical Return Leakage Audit
The moment you install, Return Intelligence imports up to 12 months of your orders, returns, refunds, fulfilments and chargebacks — then hands you a report on money that already left your business: return rate by value, refunds issued without a completed return, repeat returners, per-customer pattern flags, the most frequent return reasons, and an estimated potential exposure.
Potential exposure, not guaranteed savings. The estimate is 100% of refunds issued without a completed return, plus 30% of refunded value from repeat-return customers whose refunds exceed half their lifetime spend. We show you the formula, because a number you can't audit is a number you shouldn't trust. It runs on Shopify's Bulk Operations API in a background worker, so you can close the tab.
Live from the moment you install
How it works
1. Connect
Install from the Shopify App Store and approve only the permissions the app needs. There's nothing to configure to get value.
2. Import & learn
Up to 12 months of history is imported and every customer's abuse-relevant profile is computed.
3. Score in real time
Every new return request is scored within seconds, with a full explanation — and every new order is scored against the buyer's history before you ship.
4. Decide, with context
Your team works a prioritised queue and clicks a decision. Everything is logged, and savings are measured.
The scoring engine
Every score comes with its receipts
Most tools give you a number. Return Intelligence gives you the number, the signals that produced it, how many points each contributed, the raw values behind them, the confidence level, and the exact rule version used — stored permanently, for every score.
Risk is a 0–100 score in four merchant-adjustable bands: Low (0–29, approve normally), Moderate (30–59, a light touch), Elevated (60–79, a human should look), and High (80–100, prioritise). Confidence is a separate number, based on how much history exists — a 70 on a customer with two orders isn't a 70 on a customer with forty. Below your confidence threshold, every case routes to a human regardless of score. The app would rather say “I don't know enough” than guess.
Nine signals, every one explainable
What the score is actually made of
Return frequency
Share of the customer's orders that came back, with a bump for 3+ returns in 90 days. Only fires once a customer has at least 3 orders — no judging first-timers.
Refunds vs lifetime spend
Refunded total as a proportion of everything they've ever spent, banded at >30% / >50% / >80%.
Repeated claims
Prior damaged, not-received and wrong-item claims, with an extra escalation when this request repeats the same claim type — the single most predictive pattern in practice.
Bracketing (category-aware)
Multiple variants of one product, most coming back — heavily discounted for size-driven categories like footwear and apparel, where buying three sizes is normal shopping.
Refund without return
Prior refunds where the merchandise never physically came back — the clearest hard-loss pattern.
Return immediately after delivery
A return opened within hours of the real carrier delivery scan (threshold configurable, default 24h).
Multiple accounts per address
Other accounts sharing a hashed address — deliberately capped low. Households legitimately share addresses, so it can support other signals but never drive a decision alone.
High-value exposure
Request value above your threshold (default $300, configurable), weighted higher when the buyer has almost no history.
Chargeback / refund double-dip
A chargeback filed on an order that already refunded, and customers with that pattern in their past — the one merchants complain about most.
Structurally enforced in code
The guarantees behind every score
No auto-decline exists
There is no automatic-decline option at any score or setting. Declines are human decisions citing your published policy — it simply isn't in the product.
Deterministic, not a black box
Same inputs, same output, every time. It's an explainable rules engine, not a proprietary model that drifts.
Versioned
Every score records the engine version and rule-set version that produced it.
Never blocks Shopify
If a score can't be computed, the case is marked “score unavailable” and your normal Shopify workflow continues untouched.
No protected characteristics
Scoring uses order, return, refund, delivery and chargeback behaviour. Nothing else — no demographic data, ever.
A VIP exemption, not a blank cheque
Trusted-tag orders get a 60% risk discount — a discount, not a zero — so a compromised VIP account still surfaces at extreme totals.
Pre-ship protection
The best return to prevent is the one you haven't shipped yet
Return Intelligence scores every new order against the buyer's own history — return ratio, refund-to-spend, prior claims, refunds without returns, chargeback history — the moment it comes in.
When a buyer looks risky, the order is tagged RRAI:risky-buyer (and RRAI:watchlist if you've flagged them), risk metafields are written to the order for staff and automations to read, and an audit note records what was tagged and why. Optionally, a releasable hold on fulfilment for High-band buyers lets a human look before it ships — one click in Shopify Admin releases it. It's off by default; you opt in. A hold is a pause, not a cancellation — and it's the only pre-ship action the app can take.
Built for the person working the queue
The day-to-day review workflow
Dashboard
Open cases, refund value at risk, and recorded savings — plus a double-dip alert (“don't refund again”), a high-risk queue link, and an ingestion-health card so failures surface in-app, not weeks later.
Prioritised review queue
Sorted by score, highest first. Filter by state and risk band, bulk “mark reviewed”, and export cases or customers to CSV in one click.
Case detail: “Why this score”
Every signal with its point contribution, plain-English explanation and raw value, plus confidence, engine and rule version, and a Re-score now button for when data arrives late.
Six one-click decisions
Approve · Approve as store credit · Request evidence · Require tracked return · Mark reviewed · Decline under policy — each with a free-text reason kept in the audit trail.
Customer risk profile
Full history, watchlist auto-tagging of future orders, and a one-click “legitimate shared household” suppression that's recorded in the audit trail.
Monitor & Assisted modes
Start in Monitor mode — the app scores and explains but never touches Shopify returns. Turn on Assisted mode when you're convinced, and approvals/declines sync to Shopify on an explicit click.
Evidence collection
Ask for proof without accusing anyone
Click Request evidence and the app mints a private upload link plus a ready-to-send message you can paste into email or your helpdesk. The page the customer sees is calm and neutral — it asks for photos of the product, packaging and shipping label, and never mentions fraud, risk or scores. Up to 6 photos, 5 MB each (JPG, PNG, WEBP, HEIC), on a single-purpose link that expires after 14 days.
Duplicate-photo detection. Every uploaded image is fingerprinted. If the same photo shows up on a different claim — the classic recycled-damage-photo trick, and increasingly AI-generated images reused across claims — the case shows a warning naming the other order. The recommended response is to request fresh evidence, not to accuse.
Chargeback evidence pack
One click, one PDF, one dispute response
Every case prints or saves as a PDF with everything a dispute response needs: the order, case status, value and customer; the risk assessment with every signal explanation, confidence and engine version; the returned items with reasons; and customer-submitted evidence photos inline with upload timestamps.
At its heart is a full chronological timeline — order placed → fulfilment created → every carrier scan → return requested → refunds issued → chargeback filed → every staff decision with actor, timestamp and reason. When you're contesting an “item not received” claim against a signed delivery, the evidence is only as good as the paper trail. This is the paper trail.
It lives inside Shopify, not beside it
Shopify-native by design
Embedded admin app
Built with Shopify's official framework (Polaris + App Bridge), so it looks and behaves like Shopify Admin.
Risk card on the order page
A native admin block showing score, band, confidence, recommendation and top signals — right where your team already works.
Order & customer metafields
risk_score, risk_band, confidence, recommended_action and top_signals in an app-owned namespace, merchant-readable and usable in automations.
Automatic order tags
RRAI:elevated-risk, RRAI:high-risk, RRAI:risky-buyer and RRAI:watchlist — filterable in Shopify and usable in Flow.
Shopify Flow trigger
“Return risk scored” fires after scoring completes, with order, customer, score, band, confidence and recommended action — closing the gap left by Shopify's own pre-score trigger.
Works with your returns stack
Shopify self-serve returns and third-party return portals keep running exactly as they are. This is an intelligence layer, not a replacement.
Built privacy-first, because fraud tools have to be
Privacy, trust & compliance
Personal data minimisation
Email, phone and address are never stored in readable form — only as one-way salted hashes used purely for matching. Names are first name plus last initial; full details stay in Shopify.
Per-store salt
Hashes are unique to your store, so nothing can be correlated across merchants.
No cross-merchant data sharing, ever
Your customers' behaviour stays yours. Others market shared “global fraud ledgers” — we've deliberately chosen not to build one.
GDPR / CCPA handling built in
Data requests produce a report of exactly what's held; customer redaction wipes identity data while keeping de-identified statistics; uninstalling triggers full deletion.
Full append-only audit trail
Every score, decision, override and signal suppression is recorded with actor and timestamp — including who decided and why.
Fair-by-design
No protected or demographic characteristics. Household-sharing is capped and one-click suppressible, bracketing is category-aware, and no automatic declines exist. Nothing hidden behind a black box.
We report what we saved you — conservatively
Measured savings & honest accuracy
Prevented refund
A refund declined under your published policy after review — recorded with the basis of the calculation.
Store-credit conversion
A cash refund turned into store credit, with the cash retained.
Recovered merchandise
A return that actually came back after your team required a tracked return.
Your false-positive proxy
Approved-after-elevated-flag rate — if we're crying wolf, this number tells you. We show you the unflattering metric on purpose.
Override rate & review time
How often your team disagrees with the recommendation, and the average time from scoring to decision.
Per-product return rate & digest
Units returned ÷ units sold, so you can find the product that's actually the problem — plus a weekly digest email, switchable off.
Your policy, your thresholds, your call
Tune it to your store — and preview before you save
Choose Monitor or Assisted mode, set an individual weight slider (0–2×) for all nine signals — zero disables a signal entirely — move the band cutoffs, set your high-value threshold and minimum confidence for automation, pre-fill category exceptions where multi-size buying is normal, and manage VIP tags and integration toggles.
Preview impact before you save. Adjust any setting and the app dry-runs the new rules against your open cases, then tells you exactly how many would change band and in which direction — before anything is saved. Rule versions are append-only: saving creates a new version and re-scores open cases, while old scores keep pointing at the exact rules that produced them, so a change you regret is rolled back by re-saving the previous values.
Real behaviours, handled fairly
When we say “this is fine” — and when we raise a hand
3 sizes, returns 2 → LOW
Bracketing is discounted for size-driven categories. This is normal shopping, and the explanation says so.
Shared household address → suppressed
The address signal is capped low and can never drive a decision. One click marks it a legitimate household and suppresses it permanently.
Brand-new customer → routed to a human
Confidence drops on thin history, so the case goes to your team rather than being guessed at.
VIP with an unusual return
The trusted-tag discount applies, but the case still surfaces if the total is extreme.
Serial returner → signals stack
Return-frequency and refund-to-spend fire together, with the actual ratios shown.
Third damage claim in six months → ELEVATED
The repeat-claims signal escalates because this claim repeats the pattern.
“Never arrived” on a signed delivery → HIGH
Repeat-claims + refund-without-return + high-value stack into an urgent review — never an auto-decline.
Refund, then a chargeback → double-dip alert
The double-dip signal fires, the dashboard alerts, and the case warns your team not to refund twice.
Who it's for
Built for the stores that feel this most
Return Intelligence is an ideal fit for apparel, footwear, beauty, accessories and consumer-electronics brands doing $1M+ in annual GMV with 50+ return or refund requests a month — teams handling returns by hand today, running generous policies they want to enforce fairly rather than tighten, and already using Shopify self-serve returns or a portal.
It serves the owner (exposure and savings), the support agent (queue and decisions), the returns manager (policy, thresholds, exceptions), finance (chargeback evidence and the savings ledger), and the analyst (CSV exports, product return rates). It's not a returns portal, shipping-label or exchange-automation tool — it's the intelligence layer that sits behind those.
Other tools flag. We explain.
Why Return Intelligence is different
It explains, it doesn't just flag
Every signal, its points and its raw values, in plain English — not a bare number or a mystery label.
No auto-decline — it can't
Automatic declining isn't a toggle we leave off; it doesn't exist in the product. Your team always decides.
A free 12-month audit, formula shown
Start with your own leakage number, with the exact exposure formula printed — no gated trial.
Confidence separate from risk
Low confidence always routes to a human, so thin-history customers are never guessed at.
Your data is never pooled
No shared cross-merchant fraud ledger. Your customers' behaviour stays yours.
Preview-impact, versioned rules
Dry-run any rule change against open cases before saving, with append-only versions you can roll back.
Every plan includes every feature — you pay only for scored-request volume
Priced on return volume, not store size
Audit · Free
Full 12-month leakage audit plus 25 scored requests/month. Find out what you lost — no card.
Growth · $99/mo
250 scored return requests/month. Best for $1–5M GMV stores.
Pro · $249/mo
1,500 scored requests/month — higher volume with automation.
Plus · $599/mo
Unlimited scored requests. Multi-store, high volume, priority support.
FAQ
Questions merchants ask before installing
Do you replace my returns portal, Loop, ReturnGO or Shopify self-serve returns?
No. Return Intelligence is an intelligence layer that runs alongside whatever you use. Your customer-facing return flow doesn't change.
Will this automatically decline my customers?
No — and it can't. There is no automatic-decline capability in the product at any score or setting. Declines are made by your team, citing your published policy.
What do my customers see?
Nothing, unless you request evidence. If you do, they see a neutral page asking for photos of the product, packaging and label. Scores and risk bands are never shown to customers.
Will it flag my honest customers?
It's built to avoid that: bracketing is discounted for size-driven categories, shared-address matching is capped and one-click suppressible, low-confidence cases always go to a human, and we report our own false-positive proxy.
How much history do you need?
It works from day one and improves as history accumulates. Until then, confidence stays low and cases route to your team rather than being guessed at.
What data do you store about my customers?
Order, return, refund, delivery and chargeback behaviour. Email, phone and address are stored only as one-way salted hashes; names as first name plus last initial. Full details stay in Shopify.
Is my data shared with other merchants?
No. Never.
What happens if I uninstall?
Processing stops immediately and your store's data is deleted in line with Shopify's data-deletion flow.
Can I tune the scoring?
Yes — every signal's weight, all band cutoffs, the high-value threshold, the confidence floor, category exceptions and VIP tags. And you can preview the impact of any change before saving it.
Does it work with Shopify Flow?
Yes — a dedicated “Return risk scored” trigger that fires after scoring, plus risk metafields and order tags any Flow workflow can read.
Does it help with chargebacks?
Yes, in two ways: it detects refund-plus-chargeback double dips and alerts you, and it produces a one-click evidence pack with the full order/delivery/decision timeline for your dispute response.
Find out what last year's returns really cost you
Free 12-month audit. No card. Installs in minutes, and nothing changes for your customers — Monitor mode by default, and no automatic declines, ever. Uninstall deletes your data.
Return Intelligence is a Shopify embedded app that scores every return and refund request from 0–100, shows the exact signals behind each score in plain English, and recommends an action — while your team always makes the final call. It reads up to 12 months of your own order, return, refund, fulfilment and chargeback history and turns it into an explainable risk assessment at the moment of decision.
It's an intelligence layer, not a returns portal: it runs alongside Shopify's native returns and any third-party portal without replacing them. There are no automatic declines anywhere in the product, scoring is deterministic and versioned — no black box — and your customers' data is never shared across merchants. Start with a free 12-month Historical Return Leakage Audit: a number, not a trial.
Dig deeper on Return Intelligence
Give your store an unfair advantage
Join hundreds of thousands of merchants running Synova. Start free, upgrade when you're ready.