Spam catch rates are the wrong test for payment fraud
If your board asks which email tool stops business email compromise, the honest answer depends on the attack. A spoofed CEO asking for gift cards, a lookalike supplier domain sending new bank details, and a real supplier whose mailbox has been taken over are three different detection problems. A product can be excellent at the first and close to blind on the third.
That's why many buyers come away with the wrong answer from Abnormal vs Proofpoint comparisons that lean on malware and spam benchmarks. Payment-fraud emails usually carry no attachment and no malicious link, and often pass authentication. They're plain, polite and arrive in an existing thread. What catches them is context: who normally emails whom, what a supplier usually asks for, and whether the language or timing is out of character.
This comparison judges Abnormal, Proofpoint and Microsoft Defender for Office 365 Plan 2 on three scenarios:
- Executive spoofing: someone poses as your CEO or CFO, usually from a free webmail account or a display-name trick, and asks for an urgent transfer.
- Vendor impersonation: an attacker registers a lookalike of a real supplier's domain and sends updated banking details.
- Compromised supplier mailbox: the supplier's real account is taken over, and the fraud request comes from a fully authenticated address, often as a reply in a genuine invoice thread.
We haven't run lab tests on these products. The assessment rests on how each is built, what the vendors document, and how those designs line up against common attack patterns. Features and packaging change, so verify current details with each vendor.
Looking to streamline vendor RFP selection?
Access pre-negotiated tier discounts through our independent institutional network.
Secure email gateway vs API: why the architecture decides what each tool can see
The secure email gateway vs API question matters more for BEC than for almost any other threat, because it decides what data the detection engine can use.
Proofpoint: gateway first, API added
Proofpoint's core Email Protection is a secure email gateway. You point your MX records at Proofpoint, and it filters inbound mail before it reaches Microsoft 365 or Google Workspace. That's a strong position for stopping bulk phishing, malware and lookalike domains before anyone sees them. The catch: a pure gateway only sees mail crossing the perimeter. It doesn't see internal mail between employees, and it has no mailbox history to learn normal behavior from. Proofpoint has narrowed that gap with post-delivery removal (Threat Response Auto-Pull), an API-based deployment option, and Supplier Threat Protection, which flags suppliers whose accounts look compromised. Check which of these are included in the tier you're quoted.
Abnormal: API-only, behavior first
Abnormal is the clearest example of API-based email security. It connects to Microsoft 365 or Google Workspace through the provider's APIs, so there's no MX change. It reads inbound, outbound and internal mail and builds a baseline for each employee and each external sender, suppliers included. Suspicious messages are pulled from inboxes after delivery. That's the trade-off: it acts after delivery rather than before, so a user could occasionally open a message before it's removed. In exchange, it can tell when a known supplier suddenly asks to change payment details from an unfamiliar location or in an unusual tone.
Microsoft Defender for Office 365 Plan 2: native, configuration-dependent
Defender runs inside Exchange Online, so it sees every message, internal mail included. Plan 2 adds anti-phishing impersonation protection, mailbox intelligence, Safe Links, Safe Attachments, Threat Explorer, automated investigation and response, and attack simulation training. Its BEC defenses work, but they lean on configuration. User impersonation protection covers a defined list of protected users, domain impersonation covers the domains you add, and mailbox intelligence learns each user's frequent contacts. It doesn't build a dedicated risk profile for each supplier the way Abnormal does.
Scoring the three payment-fraud scenarios
The table reflects our assessment by design, not measured catch rates. "Strong" means the architecture is built for the scenario. "Partial" means it works with tuning or extra modules. "Weak" means the product depends on signals this attack usually doesn't produce.
| Scenario | Abnormal | Proofpoint (gateway + supplier modules) | Defender for Office 365 Plan 2 |
|---|---|---|---|
| Executive spoofing from webmail or display-name tricks | Strong: learns executives' normal sending patterns without a configured list | Strong: mature impersonation classifiers at the perimeter | Strong if tuned: needs protected users and mailbox intelligence switched on |
| Lookalike supplier domain | Strong: compares against known supplier relationships | Strong: lookalike detection before delivery, plus DMARC tooling | Partial: domain impersonation covers listed domains; unlisted suppliers rely on general phishing models |
| Compromised supplier mailbox (fully authenticated) | Strong: supplier-level behavioral baselines are central to the product | Partial to strong: Supplier Threat Protection adds cross-customer signals, but it's a separate module | Weak to partial: mail passes authentication from a known contact; detection relies on content cues |
| Internal account takeover asking finance to pay | Strong: sees internal mail and unusual sign-in behavior | Partial: a gateway doesn't see internal mail; needs the API deployment | Partial: sees internal mail; account risk depends on identity-protection licensing |
The compromised supplier mailbox should drive your decision. It's the hardest to catch and usually the costliest, because the request looks routine and sits in a real conversation. Take a hypothetical: a 300-person manufacturer gets a reply to a three-week-old invoice thread from its packaging supplier. The reply says the supplier has changed banks and attaches a letter on letterhead. SPF, DKIM and DMARC all pass. No link, no malware. A gateway relying on authentication and reputation has nothing to block. Defender may flag the urgency, but the sender is a frequent contact, so mailbox intelligence treats the message as more trustworthy. A behavioral engine that knows this supplier has never sent banking changes, never written from that network and never replied at 2 a.m. local time has the best chance.
None of the three replaces a payment control. If an attacker phones from a spoofed number to "confirm" the email, no email product sees it.
What it costs and where the price really comes from
None of these vendors publishes simple price lists for every tier, and real pricing varies by region, reseller, seat count and bundle. The figures below are typical 2026 market ranges per user per month, in USD. Treat them as a starting point and confirm with the vendor or a local partner.
| Product | Typical 2026 range (per user/month) | Main cost drivers |
|---|---|---|
| Defender for Office 365 Plan 2 | Roughly $4–$6 standalone; included in Microsoft 365 E5 | Whether you already hold E5 or a security add-on; regional Microsoft pricing; admin time to tune policies |
| Abnormal | Roughly $3–$9 depending on modules and volume | Seat count, add-ons (account takeover, supplier risk, mailbox triage), contract length, minimum seat commitments |
| Proofpoint | Roughly $3–$7 for core gateway bundles; $6–$12+ with post-delivery and supplier modules | Bundle tier, supplier-threat and DMARC modules, archiving and DLP add-ons, managed services |
Licensing is the smaller line item. These costs are easier to miss:
- Analyst time. Defender is cheapest on paper if you already own E5, but impersonation lists, quarantine policies and investigations take hours from someone who knows the product.
- Overlap. A Proofpoint gateway plus Abnormal on top of Defender means three detection layers. Some organizations justify that. Many are paying twice for spam filtering.
- Insurance terms. Social engineering and funds-transfer fraud cover is often sublimited, and many carriers ask about email security and payment verification controls during underwriting. Ask your broker how your carrier treats them.
Looking to streamline vendor RFP selection?
Access pre-negotiated tier discounts through our independent institutional network.
Red flags to test in demos and contracts
Vendors all demo the same thing: a dramatic CEO-fraud email getting caught. Ask for the harder cases.
- Ask for a supplier-compromise demo using authenticated mail. If the vendor can only show lookalike domains, it hasn't shown you the scenario that matters most.
- Run a proof of value on your own tenant. API tools can usually scan historical mail during a trial and show what they would have flagged. Check false positives too. A tool that quarantines legitimate remittance advice will get switched off by accounts payable within weeks.
- Check where your data is processed and stored. API products read the content of every mailbox. Under GDPR in Europe, the UK GDPR, India's Digital Personal Data Protection Act, and data-protection laws in the Gulf states, Brazil and elsewhere, you'll need the hosting region, subprocessors and transfer mechanisms. Regional hosting options differ by vendor and change, so get them in writing.
- Ask about remediation timing. For post-delivery tools, ask how quickly messages are pulled and what happens if the mail provider throttles the API.
- Get the module list in writing. Supplier risk features are often add-ons. Make sure the quote includes what you saw in the demo.
- Watch for auto-renewal and seat floors. Multi-year terms with uplift clauses are common. Negotiate a seat floor that matches realistic headcount.
Verdicts: Microsoft-only shops, gateway users and lean finance teams
Microsoft-only shops
If you're on Microsoft 365 E5, or Business Premium with a Plan 2 upgrade, start by configuring the Defender you already pay for. Add every executive and finance approver to protected users. Add your top suppliers' domains to domain impersonation. Switch on mailbox intelligence and apply the strict preset policy to high-risk users. If you make large or frequent supplier payments, or you've already had a supplier-compromise incident, add an API layer. In the Abnormal vs Proofpoint choice for a Microsoft-only shop, Abnormal is usually the better fit: no MX change, it works alongside Defender, and supplier compromise is a core strength rather than an add-on.
Organizations already running a gateway
If Proofpoint already fronts your mail and handles phishing and malware well, don't rip it out over BEC alone. Close the gaps: confirm post-delivery removal and Supplier Threat Protection are licensed and tuned, and consider the API deployment for internal-mail visibility. If supplier-compromise emails still get through, run an Abnormal vs Proofpoint bake-off on that one scenario. Some organizations keep the gateway for perimeter hygiene and add Abnormal for behavioral detection. It costs more but covers both stages.
Lean finance teams
With a two- or three-person finance function and no security analyst, pick the tool that needs the least tuning. Abnormal's behavioral approach suits that, if the budget allows. If not, Defender Plan 2 with preset policies on is a reasonable floor. Either way, process matters more:
- Verify every bank-detail change with a call to a number already on file, never one taken from the email.
- Require two-person approval above a set threshold.
- Hold first payments to changed accounts for 24 to 48 hours.
- Use payee-name checks where your bank offers them, such as Confirmation of Payee in the UK and Verification of Payee in the EU. Similar checks exist or are emerging in other markets.
Decision rule: pull 12 months of supplier payments and count how many suppliers you pay by bank transfer and how many changed bank details. If hundreds of suppliers change details regularly, budget for a behavioral API layer. If it's a few dozen stable suppliers, tune Defender, tighten callback verification and revisit in a year. Before renewing any cyber policy, ask your broker or a licensed insurance adviser how your social-engineering cover responds to these controls.