How to use this. The numbered blocks follow the twelve briefings. If your renewal is in 2027, you are already inside the preparation window, so start at block 1 and put dates against blocks 4, 6 and 9 this quarter. If the proposal is already on your desk, start at block 10 and work outward, because optimising the estate is worth more than any discount you will be offered on it.
01 · Rebuild the baseline
- Find your EA anniversary date and your notice deadline, and put both where your finance lead can see them.
- Accept that discount levels A to D are gone. Price the renewal against list, not against the level you used to hold.
- Reprice the suites at the current numbers: E3 at 39 dollars, E5 at 60, F3 at 10.
- Recalculate support on the new licence total. A 10 million dollar estate becomes about 12.5 million before you buy anything new.
- Note that Microsoft has committed to EA renewals only through 2027, so this is a decision about the vehicle as well as the price.
02 · Build the stacked increase before they quote it
- One page, four rows, built from your own invoice: discount level reset, suite increases, product level increases, support recalculated on the new total.
- Price the discount reset at your own level: about 6 percent at level B, 9 percent at C, up to 12 percent at D.
- Include the quiet products nobody reviews, where the increases run into double digits.
- Expect roughly 25 percent more for buying nothing new, and take that number to your finance lead before Microsoft presents theirs.
03 · Choose the vehicle on numbers
- Model your next three years twice, once as an EA renewal and once as MCA-E, with transition cost included.
- Value price protection honestly rather than assuming it, and check who owns support under each vehicle.
- Price CSP through a partner as the third option, particularly if your organisation lost EA eligibility.
- Keep a credible MCA-E path alive. It is worth 3 to 7 percent, and only while it is genuinely credible.
- Recognise migration pressure when it arrives dressed as convenience.
04 · Read your own estate first
- Build the three column file from your own admin centre rather than the account team's deck: purchased, assigned, actually active.
- Put an annual dollar value beside every gap, so the waste is stated in money rather than in seats.
- Search the four hiding places: leavers still licensed, duplicate tooling, over tiering, and the middle of the estate nobody reviews.
- Refresh it quarterly, owned by one named person, starting this quarter rather than at renewal.
- Treat the file as leverage, not as a confession. Decide what you show, and when, before the meeting.
05 · Decide the tier by persona
- Split your users into four populations and price each at the tier its role actually needs.
- Test E7 against its parts: about 90 dollars bought separately against 99 as a bundle.
- Put E7 only against the population that was already going to get E5 and Copilot anyway.
- Value what is inside at what you would have bought, never at list.
- Negotiate mixed tiers, step up rights, and no promotion cliff at the end of the first term.
06 · Prove Copilot before you commit
- Start the pilot now. Ninety days of usage data is leverage only if it exists before the negotiation.
- Know the real cost of a seat: about 69 dollars on E3 and 90 dollars on E5, once the suite underneath it is counted.
- Pilot two job families over ninety days against three measures and one written decision rule.
- Plan against the adoption record: roughly a third of seats go unused inside ninety days.
- Buy in tranches with a price hold and reassignment rights, rather than at full scale at renewal.
07 · Forecast the metered layer
- Map what sits on what: the suite, then the seat, then the meter underneath both.
- Put one agent into one real process and measure credits per task for a month.
- Extrapolate from that measured use case rather than from anyone's projection.
- Negotiate the four consumption terms: a rate hold, rollover of unused credits, capped overage agreed in advance, and alerting before you cross a threshold.
- If you are not ready to run agents at scale in 2027, treat that as a legitimate answer and buy accordingly.
08 · Size the Azure commit yourself
- Pull your actual Azure spend month by month for the last twelve months, and put evidenced growth on top of it.
- Defend that number. Vendor forecasts in this area have overrun actual consumption by 22 to 38 percent.
- Understand what a MACC is: a promise to spend, not a discount.
- Win the six protections. They are worth more than the headline rate.
- Sequence the trade deliberately, and never volunteer the commitment before you have priced what it buys.
- Align your infrastructure and licensing teams first. One vendor, one story.
09 · Attack Unified Support early
- Work out your cost per support case for the last twelve months and set it beside your annual support fee.
- Understand that support is priced as a percentage, so it compounds every time your licence total rises.
- Price the three options: downshift a tier, split the estate, or replace it. Third party alternatives benchmark 20 to 40 percent cheaper.
- Start a year out. Later than that and you cannot change it in time for this renewal.
- Cap it, fix it, or separate it from the licence total, but do not let it renew by default.
10 · Run the optimisation pass first
- Put an annual dollar figure against each of the four levers: tier mix, frontline sizing, duplicate tooling, and dormant assignments.
- Size the F tier population properly rather than defaulting the whole workforce to E3.
- Remember every dollar removed here counts three times: licence, support, and the base your uplift is applied to.
- Optimise before you negotiate. A discount on waste is still waste.
11 · Run the countdown
- Put the eighteen month countdown in the calendar today, with a named owner on each milestone.
- Mark the notice deadline in red. Missing it hands the vendor the renewal at their numbers.
- Work their calendar: a 30 June year end, and the quarters leading to it.
- Map the approval chain behind the account team, because the account team is not the decision.
- Prioritise terms over percentage. Terms outlast any discount you win this cycle.
- Guard the last six weeks, which is where value quietly leaks out of a good deal.
12 · The three moves, this quarter
- Move one: write down the three dates. Your anniversary, your notice deadline, and minus twelve months.
- Move two: build the estate file. Purchased, assigned, active, priced.
- Move three: model the do nothing number, and tell your CFO before Microsoft does.
- Then decide whether you want an independent second opinion, and if so bring them in early enough to change the outcome.