Last updated 23 October 2026. Buy one anti-phishing product and something still gets through. A gateway that catches a fake DocuSign link will happily deliver an invoice-change request from a hacked supplier mailbox, because there is nothing in it to block. Below, the layers in the order an attack meets them, and for each one the hole it leaves.
The short answer
Authenticate your own mail. Filter inbound mail and web traffic. Put admins and finance on security keys or passkeys. Then train people to report, because a domain registered this morning, a QR code inside a PNG, or a phone call from a “supplier” all land in front of a person before any list knows about them. The Verizon 2026 Data Breach Investigations Report found the human element in 62% of breaches.
Phishing protection solutions, layer by layer
| Layer | What it stops | What still gets through |
|---|---|---|
| Email authentication (SPF, DKIM, DMARC) | Mail that forges your exact domain | Lookalike domains, display-name tricks, mail from a real but hacked account |
| Inbound email filtering | Known malicious links, senders and attachments | Newly registered domains, links that turn bad after delivery, QR codes drawn inside images, plain-text requests with no link at all |
| Web and DNS filtering | Visits to sites already listed as phishing | Fresh sites that no list has seen yet, and anything opened on a personal phone |
| Multi-factor authentication | A stolen password used on its own | One-time codes typed into a fake login page, and push approvals pressed under pressure. Phishing-resistant methods close this gap |
| Endpoint protection | Malware delivered by an attachment or download | A password typed into a fake page, and payments made after a convincing email or call |
| Trained, reporting staff | Lures that passed every filter, once someone recognises and reports them | Whatever nobody reports. Report rate and time to report are what to measure |
1. Email authentication
SPF lists the servers allowed to send for your domain, DKIM signs each message, and DMARC tells receiving servers what to do when a message fails both. Large mailbox providers now expect it: since 1 February 2024, Google’s sender guidelines require SPF or DKIM from every sender to Gmail, and SPF, DKIM and DMARC from anyone sending more than 5,000 messages a day. Authentication protects your own domain from being forged. None of it helps against rnicrosoft.com or yourcompany-payroll.com: those domains pass their own SPF and DKIM perfectly well.
2. Inbound email filtering
Gateways and API filters in Microsoft 365 or Google Workspace score senders, links and attachments on reputation and detonate files in a sandbox. Their blind spot is time. A domain registered at 9am has no reputation at 10am. A link can point to a harmless page when the filter checks it and to a login clone an hour later. And a two-line email asking accounts to “use our new bank details from Monday” has no link or file to inspect. QR codes are awkward too: the URL is pixels in an image, and the person scans it with a phone the filter never sees.
3. Web and DNS filtering
Browser and DNS blocklists turn a click on a known phishing site into a block page. Known is the catch. They also only see company laptops on managed networks; the QR code scanned on someone’s own phone goes straight past.
4. Multi-factor authentication, and why the kind matters
Any MFA beats none, but not every kind resists phishing. CISA’s fact sheet on phishing-resistant MFA describes the common attack: a fake login page collects the username, the password and the code from the authenticator app, all at once. It also lists push bombing, where an attacker sends approval prompts until someone accepts. CISA calls FIDO and PKI-based methods the gold standard. A security key or passkey is bound to the real domain, so the browser simply refuses to use it on a lookalike. Start with administrators and whoever can approve payments.
5. Endpoint protection and patching
EDR and patching deal with the macro document and the exploit. Credential phishing and payment fraud never run code on the laptop, so there is nothing for them to see.
6. People who recognise and report phishing
The reader sees a new lure before any blocklist does. Two figures from the 2026 DBIR show why this layer needs testing beyond email: phishing was the initial access vector in 16% of breaches, and in simulations the median click rate on voice and text messages was 40% higher than on email.
What helps here is practice and an easy way to report. Phishing simulation sends safe test lures across email, SMS and voice, shows a short lesson to anyone who clicks, and tracks who reports. A one-click report button lets the security team pull the real message out of other inboxes while the attack is still running. Track report rate and minutes to first report, not just clicks. If you want an outside team to run this for you, see our managed phishing simulation service. For a vendor comparison, see the best phishing simulation tools.
A starting checklist for a small security team
- Publish SPF and DKIM for every domain that sends mail, including the marketing and billing tools that send as you. Then add DMARC, for example
v=DMARC1; p=none; rua=mailto:dmarc@yourdomain, and move top=quarantineonce the reports show only your own senders. - Check which link and attachment scanning your Microsoft 365 or Google Workspace licence already includes; it is often switched off.
- Give administrators and finance staff security keys or passkeys first.
- Add a report button to every mailbox. Name the person who reads the reports, and how fast.
- Run a phishing simulation each month on more than one channel, and track report rate over time.
- Bank-detail changes get confirmed by calling a number you already had, never one from the email.
FAQs
What is phishing protection?
Phishing protection is the set of controls that stop phishing messages from arriving, stop people from acting on the ones that do, and limit the damage when someone does. In practice that means email authentication, email and web filtering, phishing-resistant MFA, endpoint protection, and staff who recognise and report phishing.
What is a phishing protection program?
The controls above plus the routine that keeps them working: who reviews DMARC reports, who reads user reports and how fast, how often staff are tested, and how payment changes are confirmed. Without named owners, most of these layers decay within a year.
Is an email filter enough to stop phishing?
No. Filters rely on what is already known to be bad, so new domains, QR codes in images and requests with no link at all often get through. Phone and text attacks never pass the email filter in the first place.
Does MFA stop phishing?
It stops a stolen password from working on its own. Codes from an app or text message can still be phished through a fake login page that asks for the code too. FIDO security keys and passkeys resist this because they only work on the real site’s domain.
Where does phishing simulation fit?
It tests and trains the last layer: the people who see whatever the filters miss. A good programme covers email, SMS and voice, teaches at the moment of the click, and measures how often and how fast people report.
