Open the Microsoft Defender portal on a Monday morning and the number at the top of the dashboard looks authoritative — a percentage score, a trend line, a list of things to fix. It is worth reading, but only once you know what it counts and what it leaves out entirely.
What Secure Score actually measures
Secure Score is a measure of an organisation’s security posture, shown in the Microsoft Defender portal. It is built from a list of recommended improvement actions, and each action carries points: complete an action and the number goes up. That is the whole mechanism, and it explains both what the score is good for and where it misleads.
Two consequences follow. The score tracks configuration choices, not incidents — nothing in the number reflects whether anything has been attacked, contained or recovered. And the denominator moves: Microsoft adds, retires and reweights improvement actions over time, so a tenant that changes nothing at all can still watch its score drift. Read the figure as the share of currently recommended actions that have been addressed, not as a percentage of safety.
The detail behind the number is where the real value sits. Each improvement action describes what it changes, the exposure it addresses, and often the licence level needed to enable it. That list is a workable backlog; the score is only its summary.
Why a higher number is not automatically safer
Points reward actions, not consequences. An action marked “resolved through third party” earns its points even though Microsoft cannot verify the control, and a control nobody monitors or a feature whose alerts no one acts on still counts. Meanwhile an account holding standing global administrator rights with no phishing-resistant method behind it is a larger exposure than many low-risk hygiene items, yet the score does not rank by risk alone, and some exposures never appear as an action at all.
The score is also blind to operations. It cannot tell you whether alerts are triaged, whether restores have been tested, or whether you could produce a record of who accessed a particular site last quarter. Those are the questions an auditor, a customer or an incident will actually ask. Use Secure Score to direct effort; do not use it to declare the work finished.
A triage order that suits a small team
Where one person runs IT alongside everything else, work the improvement list in risk order rather than point order.
- Identity and MFA first. MFA on every account, including admin, service and shared accounts; Conditional Access policies that block legacy authentication and constrain risky sign-ins; no standing global administrator roles where just-in-time access will do; and a pass over guest accounts. Identity is the layer everything else assumes is intact, and it is where access to personal data is actually controlled for PDPA purposes. Where there is no internal owner for this, managed security services covering endpoint, email and PDPA obligations are the usual alternative.
- Email and endpoint controls second. Anti-phishing and impersonation protection, Safe Links and Safe Attachments, blocking auto-forwarding to external domains, and a review of Defender for Office 365 configuration against what is licensed rather than what was set once at migration. On devices, confirm onboarding and confirm that antivirus and EDR policies are applied rather than silently disabled.
- Lower-impact hardening last. Password rules, app consent settings, sharing defaults and similar items are genuine improvements, but each is worth comparatively little next to the identity and email work above. Do them once the first two tiers are demonstrably in place.
Two habits make the order stick. Verify a sample in the tenant before marking an action complete, because a green tick in the portal is not the same as a setting that is actually applied. And record what changed and why, since the next review starts from that baseline.
Where the score stops, and what a review adds
Secure Score is scoped to Microsoft 365. It says nothing about your other SaaS applications, your network or your backups. It will not tell you whether your DLP policies match the way staff actually handle personal data, and it will not show whether the SharePoint permissions inherited from years of project work still make sense. Those gaps need a review that reads the tenant rather than the score.
That is the gap the M365 Security Audit Package on sakal.shop is built for. At S$2,180 including 9% GST (S$2,000 before GST), it covers Microsoft Secure Score analysis, an MFA and Conditional Access review, a SharePoint permissions audit, Defender for Office 365 configuration, DLP policy assessment and a PDPA compliance check — the identity, email and data areas above, examined against your environment instead of a generic checklist.
Start with the improvement list you already have: sort it into the three tiers, finish identity before touching the rest, and keep a record of what you changed and why. If you want the score interpreted against your actual tenant — with the SharePoint, DLP and PDPA gaps named and ordered by risk — the M365 Security Audit Package linked above is the practical next step.
Disclosure: Code n Rob is written by the founder of Sakal Network, which runs sakal.shop.

