What E5 actually adds, when the upgrade pays for itself, and why a uniform rollout is usually the most expensive answer.
Microsoft 365 E5 adds security, compliance, analytics and telephony components on top of E3, for roughly twenty dollars more per user per month on Microsoft's list. The upgrade pays for itself when you would otherwise buy two or more of those components separately and you decommission what they replace. It does not pay for itself when it is rolled out uniformly to a population that never touches the premium components — which is the most common way E5 is deployed.
Updated
The third column is the one that decides it: who genuinely needs the component, not who is currently licensed for it.
| Component | What it can replace | Who actually needs it |
|---|---|---|
| Defender for Endpoint P2 and Defender for Office P2 | A third-party EDR and mail security product | The whole estate, if you adopt it. This is the component that most often justifies E5 on its own — but only if the existing tool is actually decommissioned. |
| Entra ID P2 | Privileged Identity Management and risk-based conditional access | Administrators and privileged roles. Reviews routinely find that only a minority of licensed P2 seats use a single P2 feature. |
| Purview eDiscovery Premium and Insider Risk | External eDiscovery tooling and manual compliance work | Legal, compliance and HR. A small group, and one where the per-seat cost is irrelevant next to the cost of an external review. |
| Power BI Pro | A separately purchased Power BI Pro seat | Whoever builds or consumes reports. Easy to count, and often already purchased separately — which means it is being paid for twice. |
| Teams Phone | A PBX or a separate calling plan | Only where telephony actually moved to Teams. An unused Teams Phone entitlement is the clearest sign an E5 rollout was licensed but never adopted. |
Read the middle column as a commitment, not a feature. E5 only becomes cheaper than E3-plus-add-ons if the replaced products are actually switched off. An estate running Defender alongside its previous endpoint tool has bought E5 and kept the cost it was meant to remove.
How many of those five components would you buy anyway? At two or more, E5 is usually the cheaper instrument. At one, an E3 base with that single add-on almost always wins. At zero, the question is not E3 versus E5 — it is why anyone is holding E5 at all. The answer has to come from per-service-plan usage, because every other signal (headcount, department, job title, last sign-in) will tell you the population is active, and that was never the question.
Per-user, per-service-plan evidence instead of a department-wide guess.