5 to 15 percent of paid seats showed no activity, and nobody had looked

The strongest position in a Microsoft renewal is knowing your own usage better than the account team does. In the internal reviews we led, the gap between licenses assigned and licenses used was the recurring story: idle seats, E5 users living entirely inside E3 features, and add ons billing twice for capabilities the suite already carried.

Prepared by Redress Compliance · August 15, 2026 · Microsoft advisory. 30 to 40 usage reviews led, 2024 to 2026.

• Start 9 to 12 months before renewal, because the sequence is reclaim, observe the lower baseline for a quarter, then lock the reduced count into the negotiation.
• Measure activity, not assignment: last activity date per service, sign in recency, and feature specific use are the signals that separate active from idle.
• Work the tiers in order, from the never activated seats nobody will miss to the judgment call downgrades that need the feature data behind them.
• Sweep the duplicates: standalone security and voice add ons bought before the suite covered them, still billing beside the bundle that includes them.
• Document every decision in the matrix, so the renewal team walks in with a defended count rather than a spreadsheet of assertions, feeding the renewal playbook and the runbook baseline.
• Make it quarterly, since the estate drifts continuously and an annual heroic audit always arrives to find a year of accumulated waste.

Every Microsoft estate is audited regularly. The only question is by whom. The account team audits it at every renewal, through the lens of its own telemetry and its own interests, and arrives with a view of your demand that is never conservative. The auditor's version arrives less often but with more paperwork. The internal usage review is the third audit, the one you run yourself, and it is the only one of the three whose findings work in your favor. Estates that skip it are not unaudited; they have simply outsourced the audit to parties who profit from the answer.

What the self audit finds is remarkably consistent, because the causes are structural. Licenses are assigned at onboarding by default and removed at offboarding by exception, so the idle pool grows to 5 to 15 percent on pure administrative drift. Suite tiers are chosen by role category in a procurement cycle and never revisited against actual feature use, so a fifth to a third of E5 seats belong to users whose working life happens entirely inside E3. Add ons purchased in one budget year survive into the suite consolidation of the next, so 10 to 20 percent of tenants pay twice for the same capability. None of this is anyone's mistake, which is why it is everyone's cost.

The discipline that converts these findings into money is sequencing, and this is where most reviews fail. A review run ninety days before renewal produces a beautiful report about waste that is now embedded in the baseline, because the count Microsoft prices is the count on the day of the quote. The review that works starts 9 to 12 months out: reclaim the tiers, hold the lower count for an observed quarter, and walk into the negotiation with the reduced baseline as the established fact. The difference between the two timings is not the quality of the analysis; it is whether the analysis happened while the number could still move.

The deeper payoff is what a standing quarterly review does to every future negotiation. An estate with a live matrix, activity per service, tiers mapped to feature use, duplicates swept, never presents Microsoft with an inflated count to price, never faces a true up surprise, and never negotiates against its own drift. The account team's audit and yours converge on the same clean number, which is precisely the outcome the account team's version of the audit exists to prevent. The renewal machinery this feeds lives in the renewal playbook, the count reset in the renewals brief, and the practice library in the Microsoft hub.

Watch the briefing · 4:02The Microsoft EA Preparation Playbook: The Work That Wins the RenewalThe usage file workstream in context: the license position, the demand forecast, and the ask list drafted before Microsoft drafts theirs.

Across 30 to 40 led reviews, the assigned to used gap was the story on every estate:

The patterns: assignment treated as usage in every internal report, the review scheduled inside the notice window where its findings could no longer move the baseline, and the estates with quarterly matrices walking into renewals the account team could not surprise.

The buyer side move is to be the first auditor of your own estate. The wider library sits in the Microsoft practice.

Why run a usage review before a Microsoft renewal?

Because the renewal sets your cost for three years, and walking in without usage evidence means accepting the vendor's view of your demand, which is rarely conservative. The review converts opinion into numbers you can defend: seats to reclaim, tiers to right size, and proof that counters an uplift ask.

When should the usage review start?

Nine to twelve months before renewal. That leaves time to reclaim seats, observe the lower baseline for a quarter, and lock the reduced count into the renewal rather than the inflated one. A review run in the notice window finds the waste too late to remove it from the baseline.

Where does Microsoft usage data come from?

Mostly from what you already have: the Microsoft 365 admin center active users reports per service, the license assignment export, and Entra ID sign in logs for dormant accounts. Native reports cover most enterprises; a SAM tool helps at large scale or across hybrid estates.

What did the reviews actually find?

Three recurring pools: inactive or unassigned seats at 5 to 15 percent of paid licenses on almost every estate, E5 users with no use of advanced security or analytics features at 20 to 30 percent of E5 seats, and duplicate add ons already inside a suite billing twice on 10 to 20 percent of tenants.

How does the reclaim logic work?

In tiers, from clearest waste to judgment calls. Tier 1: assigned but never activated, reclaim immediately. Tier 2: inactive 90 days or more, reclaim after a manager check. Tier 3: overlicensed, downgrade to the lower suite tier. Document each decision so the renewal team can defend the count.

When is an E5 user really an E3 user?

When a full quarter shows no use of advanced security, compliance, voice, or analytics features. Pay for E5 capability where it is exercised and downgrade the rest; the feature usage data, not the job title, makes the call defensible.

How often should the review run?

Quarterly as a standing matrix, not once before renewal. Each row is a license type, each column a decision input, and the output is a reclaim or downgrade action. The annual habit keeps the baseline honest, which means every future renewal starts from a defended count instead of an inflated one.