Microsoft Defender email security: what the latest results mean for your business

What the latest comparison shows

Microsoft has published its fifth consecutive quarterly email security benchmark, covering May to July 2026. It compares how email protection services perform and explains how the findings are influencing Defender’s development.

In the comparison of secure email gateways — services that filter incoming email — Defender missed 221 high-severity threats for every 1,000 protected users. That was 55.4% fewer than the next-best gateway vendor assessed.

Microsoft measures missed threats rather than simply counting blocked messages. Its reasoning is that different customer environments receive different amounts of malicious email, making headline blocking totals harder to compare. These are Microsoft-published figures, using Microsoft Defender data, so they should inform a purchasing discussion rather than settle it on their own.

Strong results do not mean every threat is stopped

Microsoft reports that missed threats have increased across several reporting periods, including for Defender. It links this with attackers using AI to research targets, personalise messages and produce more convincing impersonations.

For a small business in London or Manchester, the practical point is straightforward: an email reaching someone’s inbox is not proof that it is safe. Protection needs to keep assessing messages after delivery, as well as checking them on arrival.

During the benchmark period, Defender caught an average of 92% of post-delivery malicious messages. Microsoft says it reassesses delivered emails as fresh information about threats becomes available, allowing it to address risks that were not clear during the initial checks.

Where additional email security helps

The report also looks at integrated cloud email security services: additional tools that work alongside existing email protection. Their largest contribution remained filtering promotional and bulk messages.

Their reported contribution to catching malicious email rose from 0.13% in the previous quarter to 0.30%, while the spam figure increased from 0.28% to 0.52%. Microsoft describes the additional gains for these categories as comparatively modest.

This distinction matters when deciding where to spend a limited IT budget. Reducing inbox clutter can be useful, but it is not the same outcome as stopping dangerous messages. Ask suppliers to explain which problem an extra service would solve for your business.

What Microsoft has changed

Microsoft highlights three developments informed by benchmarking, customer feedback and threat data. A new Promotions folder in Outlook helps separate marketing and bulk messages while keeping them accessible.

Defender’s revised machine-learning and AI models also consider language-related clues, including a message’s topic. In a four-week research period, Microsoft observed roughly two-thirds fewer missed threats and nearly one-fifth fewer legitimate messages incorrectly flagged. Those findings relate to that observation period, not a guarantee for every organisation.

Microsoft has also introduced protection against malicious instructions embedded in emails for AI systems to follow. It says this detects and isolates such instructions before delivery, helping protect Copilot, AI agents and other systems that process inbox content.

What to do next

Ask your IT provider to explain your current checks before delivery and your ability to detect and address threats afterwards. Review how suspicious messages are reported and handled, rather than relying only on a total of blocked emails.

Before buying another security layer, request evidence of its likely additional value in your environment. If your business uses AI to read or act on emails, include protection against malicious AI instructions in that review.