Fraud-prevention roundups rank enterprise tools on one leaderboard and quote “contact sales” pricing. A small store needs a different comparison: what the built-in tools already cover, the two fundamentally different jobs paid tools do, and the volume at which each job is worth paying for.
The two jobs: scoring vs liability shift
Scoring tools (Sift, SEON, Kount, Stripe Radar) evaluate each order and tell you how risky it looks — you still decide, and you still eat the chargeback if you're wrong. Chargeback-guarantee tools (Signifyd, Riskified, NoFraud) approve or decline the order and absorb the loss if an approved order turns out fraudulent — you're buying the liability shift, not just the score. Different products, different math. Never compare them on one leaderboard.
What your platform already gives you (the free layer)
Before spending anything: Shopify's native fraud analysis triages every order into low/medium/high risk with the signals exposed (AVS, CVV, IP geolocation) — route medium-risk to manual review, don't auto-decline. Stripe Radar's standard screening is built into Stripe. AVS + CVV + 3D Secure at checkout cost nothing extra and stop the crudest fraud outright. Most small stores never outgrow this layer properly configured — the failure is usually unconfigured rules, not missing vendors.
The tools compared
| Tool | Model | Price signal | Best fit |
|---|---|---|---|
| Stripe Radar (for Fraud Teams) | Scoring | $0.02 per screened transaction (checked 2026-10-11) | Stripe stores wanting per-order screening |
| SEON | Scoring (modular) | Pay-as-you-go, modular — viable at low volumes | Small teams wanting device fingerprinting without enterprise contracts |
| NoFraud | Guarantee | Free tier available | Smallest stores wanting guarantee protection at $0 to start |
| Signifyd | Guarantee | Custom/volume pricing | Growing stores where fraud losses exceed the per-order fee |
| Riskified | Guarantee | Custom/volume pricing | Revenue-focused stores optimizing approvals |
| Sift | Scoring | Custom/volume pricing | Scaling teams with review operations |
| Kount (Equifax) | Scoring | Custom/enterprise | Enterprise, multilayered fraud operations |
Small-store buyers consistently land on the same shortlist: SEON for transparent pay-as-you-go device fingerprinting and fast deployment, NoFraud's free tier as the $0 starting point for guarantee-style protection, and Signifyd/Riskified once volume justifies the sales conversation. The G2 discussion threads for this category keep returning to the same buying criteria for small stores — transparent pricing, quick time-to-value, and real support during onboarding — which is exactly where “contact sales” enterprise tools fail the test.
The volume decision: when a paid tool earns its keep
The setup: $30,000/month online, ~1,000 transactions, 0.5% fraud rate = $150/month in fraud losses.
Scoring at $0.02/transaction: $20/month. It pays for itself by stopping a single $50 fraudulent order — before counting the chargebacks it prevents.
Guarantee products: worth it when your fraud losses exceed the vendor's per-order fee at your volume. Below ~$50k/month, the free tier (NoFraud) or the built-in platform tools usually cover you; $50k–$500k is where SEON-style modular tools earn their keep; above that, the guarantee vendors' economics start working.
The setup sequence: rules before vendors
- Turn on everything free first: AVS, CVV, 3D Secure, and your platform's native fraud analysis.
- Set a review band, not a block list. Route 1–2% of orders to manual review; auto-holding more than 2–3% means your rules are too tight and you're losing good revenue.
- Measure for two weeks: fraud rate, false-decline complaints, review workload. Numbers first, vendor second.
- Add one paid layer only if the numbers justify it — scoring for screening efficiency, or a guarantee product if fraud losses exceed its fee.
- Re-check quarterly. Fraud patterns shift; the tool that earned its keep in Q1 may be redundant by Q3.
What success looks like: fraud rate under control, manual review steady at 1–2% of orders, no legitimate-customer complaints about declines, and a paid tool (if any) whose monthly cost is visibly smaller than the fraud it stops.
A note on steps: this is a buying-decision guide with a setup sequence — the numbered sequence above is the actionable part. It's a configuration procedure, not a repair, so there are no device settings paths; the “settings” are your platform's fraud rules.
FAQ
Do I need a fraud tool if I use Shopify Payments?
Probably not at first — Shopify's native fraud analysis plus AVS/CVV/3D Secure covers most small stores. Add a paid tool when your measured fraud losses exceed what the tool costs.
What's the difference between Signifyd and Sift?
Signifyd is a guarantee product — it absorbs the loss on approved orders that turn out fraudulent. Sift is a scoring product — it tells you how risky an order looks, and you keep the liability. Buy the guarantee when fraud losses justify the fee; buy scoring when you need screening efficiency.
How much manual review is normal?
Around 1–2% of orders for a healthy setup. Above 2–3% auto-held, your rules are too aggressive; below 1% held with a rising fraud rate, they're too loose.