If you’re running a large UK enterprise on Microsoft software, you almost certainly signed a three-year Enterprise Agreement at some point between 2019 and 2023. At the time, this was the obvious thing to do. The EA was the grown-up way to buy Microsoft licensing. It gave you committed volume pricing, predictable annual true-ups, and a single line in the budget that the CFO could plan around. For organisations that knew their headcount, knew their product mix, and weren’t planning anything weird, the EA was a perfectly sensible instrument.
Almost all of those assumptions have stopped being true.
Microsoft has spent the last two years steadily restructuring how it sells licensing, and the direction of travel is unmistakable. The Enterprise Agreement is being quietly de-emphasised in favour of two newer routes: the Microsoft Customer Agreement, which is a direct digital contract between Microsoft and the customer, and the Cloud Solution Provider model, where licensing is intermediated by a partner who can offer more flexibility and bundled support. Both of these models suit Microsoft strategically, because both make it easier for customers to consume more services more elastically, and both reduce the friction of adopting new products as they launch. Copilot, in particular, has accelerated the shift. The EA was not designed for a product whose value depends on how many of your users actually pick it up.
The practical consequence for most CIOs is that the licensing model they’re sitting on no longer fits the way their organisation buys software.
The mismatch shows up in a few specific ways. Headcount volatility is the obvious one. EAs assume a relatively stable user base; the post-pandemic mid-market has had three years of either rapid hiring or rapid restructuring, often both, and the rigid commit-true-up cycle of an EA has trapped a lot of organisations in licence positions that don’t match their actual user count. Product mix is the second. The right blend of E3, E5, F3, Teams Premium, Copilot and the various security add-ons has become a multi-axis optimisation problem that nobody negotiating an EA in 2021 could have anticipated. Adoption uncertainty is the third — there’s no clean way to flex Copilot licences down inside an EA if the rollout doesn’t go to plan, which means a non-trivial number of organisations are paying for capability they aren’t using.
The CSP route, in particular, has matured quickly and is now the default recommendation for any organisation that wants flexibility without sacrificing volume pricing. The CSP model lets you scale licences monthly rather than annually, mix and match product SKUs without renegotiating a master agreement, and consolidate support and licensing through a single partner relationship. The trade-off used to be that CSP pricing was less competitive than EA pricing at large volumes; that gap has narrowed considerably as Microsoft has aligned its incentives behind the model.
The catch, as ever, is that the partner matters more than most buyers expect. CSP is a partner-mediated relationship, which means the quality of the partner determines the quality of the experience. The good ones — and there’s a small group of Microsoft CSP partners running this properly — operate with full pricing transparency, give you a portal to manage your own licences, don’t lock you into multi-year contracts at their end, and provide genuine licensing optimisation advice rather than upselling you toward the most expensive SKU. The bad ones do the opposite of all of these things, and a surprising number of mid-tier IT firms have built a quiet business out of doing the bad version while telling customers it’s industry standard. It isn’t.
For organisations approaching an EA renewal in the next twelve to eighteen months, the question worth asking is no longer “what’s the best EA renewal price we can negotiate?” It’s “is the EA still the right vehicle at all?” In a meaningful number of cases, it isn’t. A blended approach, with CSP for the licences that fluctuate, MCA-Enterprise for the stable core, and perhaps a smaller EA for specific committed product lines, produces a structure that fits how the organisation actually consumes software, rather than the structure it would have made sense to buy in 2020.
See also:Automating Website Health Checks with an SEO Audit Reporting Tool
There’s a broader point here that’s easy to miss. Microsoft licensing has stopped being a procurement exercise and become an ongoing operational discipline. The decisions made at signing matter less; the decisions made each month matter more. Organisations that haven’t internalised this are paying somewhere between five and twenty percent more than they need to, every month, for licences that don’t quite match their consumption. That’s the kind of recurring inefficiency that, compounded over the term of a three-year agreement, adds up to a number that would justify a great deal of attention.
Most CIOs will look at their licensing again when the renewal comes around. The ones doing it well are looking at it now.
