The indirect route, the New Commerce terms, where partner margin sits, and when CSP genuinely beats buying direct. Three knowledge checks along the way, and 1 clip from a senior cloud advisor.
This is a taught session, not a talking head. The instructor works through analyst grade slides, and three times the video stops on a question with four options on screen. Pause, commit to an answer, and the next slide explains which option is right and why each of the others is wrong. Once in the session the frame splits and a senior cloud advisor gives the view from inside real Oracle negotiations, and the instructor picks the clip apart when the slides return.
The full narration of this session, section by section, for reading and reference. Guest analyst clips are marked.
Welcome back, session four of forty, and the third of the three vehicles. We have done the Enterprise Agreement, three years of price certainty bought with a commitment that only counts upward. We have done the Microsoft Customer Agreement, evergreen, where nothing expires and nothing is protected either. Today, CSP, the Cloud Solution Provider programme, which is the indirect route: you buy through a partner who bills you, supports you, and prices you. CSP is sold as the flexible alternative to the EA, and here is the honest version, it genuinely can be, but only if you understand two things, the New Commerce terms and the partner margin baked into your price. Get either wrong and you end up with a vehicle that costs like flexibility and behaves like a commitment, which is the most common CSP disappointment I know. Today: the indirect model, the three terms, the margin, choosing a partner, and where CSP genuinely wins. Let's go.
Five takeaways. One, the model: what indirect buying actually means, who bills you, who supports you, and who stands between you and Microsoft, because that party is now part of your product. Two, the terms: New Commerce monthly, annual, and three year, and specifically the way an annual term removes the flexibility most buyers believe they have bought. Three, the margin: where it sits in your price, how much it runs to when nobody benchmarks, and how to measure it without a confrontation. Four, the partner: what a good one does beyond forwarding invoices, and what to write into the agreement rather than hear in the pitch. And five, the fit: the seat profiles where CSP genuinely wins, and the hybrid split that in our review data beat committing an entire estate to any single vehicle. One number to hold from the start: across the vehicle decisions we advised in 2024 and 2025, the cheapest outcome was almost never a single vehicle.
The indirect model, three consequences of putting a partner between you and Microsoft. They bill you: your invoice comes from the partner on their commercial terms, which means payment terms, currency, and how the lines are consolidated are all negotiable with a party who actively wants your business, and that is a genuine advantage the EA structure rarely offers. They support you: first line support comes from the partner rather than from paid Microsoft support, and for a mid size estate that is often both better and cheaper, though notice the dependency, it is entirely a function of which partner you picked. And they price you: the partner buys at their cost and sells at yours, so a margin sits inside every single line of your invoice. I want to be careful here because this gets moralised and it should not be. The margin is legitimate. It pays for the support, the administration, the reporting. The question is never whether it exists, it is whether it is proportionate to what you actually receive, and you cannot answer that question until you have measured it, which is the third section of today.
The New Commerce terms, and the framing I want you to take away is that the term is the product. Monthly: genuine flexibility, quantities can fall month to month, and it carries a premium over annual because you are buying optionality and optionality has a price. Annual: a lower unit price, and quantities lock for twelve months. Three year: the lowest CSP unit price with a longer hold, which is commitment approaching an EA without the enterprise wide structure. And then the fourth row, which is not a term but an error, and it is the one that shows up everywhere: signing annual because the price is better and then planning as though the subscription were monthly. Across the CSP estates we reviewed in 2024 and 2025, New Commerce annual terms locked seats that the buyer was treating as monthly in twenty five to forty percent of estates. That is not a rare edge case, that is between a quarter and a half of everyone. So say it plainly: the price difference between monthly and annual is the price of flexibility. If you take the cheaper price, you sold the flexibility, and that is a perfectly reasonable trade as long as you know you made it. First check.
First check. You moved to CSP for flexibility and took annual terms because the unit price was better. In month five you need to shed two hundred seats after a restructure. What happens? A, seats can be reduced at the next monthly billing cycle. B, the partner can reduce them since they own the billing relationship. C, the annual term locks the quantity for twelve months, so the reduction waits for the term anniversary: you bought the annual price and the annual commitment together. Or D, CSP always allows reductions, that is the point of the programme. Pause here. Which term did you actually sign, and what did that lower unit price pay for?
The answer is C, and it is the disappointment I referred to at the top, arriving on schedule. The monthly premium exists precisely because monthly flexibility is worth something, so choosing annual for the lower unit price was a deliberate trade of flexibility for price, whether or not it felt like a trade in the moment. And this appeared in twenty five to forty percent of the estates we reviewed, always, without exception, as a surprise to somebody. A and D are the monthly term applied to an annual subscription, which is the assumption itself rather than an answer. B is worth a moment because it sounds plausible: the partner owns the billing relationship, so surely they can adjust it. They administer your subscriptions inside Microsoft's programme rules; they do not have private authority to break a term. Though I will say this, a good partner warns you about exactly this before you sign rather than after, and if yours did not, that tells you something useful about the margin you are paying. The instruction from this check: match the term to the seat, population by population, and expect to run more than one term inside the same estate. That is not untidy. That is correct.
Where the margin sits, five points. It is inside the unit price: the partner buys at their cost and sells at yours, and you will never see a margin line on an invoice, you see a price with the margin already inside it. It varies widely: where buyers never benchmarked, partner margin and uplift added five to fifteen percent over direct or EA equivalents across the estates we reviewed, and that is a wide enough range that guessing is not a strategy. It is legitimate: the margin funds support, administration, and whatever services the partner genuinely provides, so the question is proportionality rather than existence. Benchmark by competition: quote the same basket with two or three partners, and the spread between them is your margin range, measured without asking anyone to disclose anything, which is why it is the polite method as well as the effective one. And then negotiate the service: once the number is visible, the conversation stops being about price and becomes about what the margin buys, support levels, reporting, licence optimisation, named contacts, written into the partner agreement rather than described in a meeting. Our guest analyst has seen what happens when none of this is done.
Guest analyst The CSP estate I use as a warning was a professional services group, about two thousand two hundred seats, who had moved off an Enterprise Agreement specifically to get flexibility. That was the stated business case, written down, approved by their board: our headcount moves, we need to be able to move with it. Two years later I was asked to review why their Microsoft costs had not fallen despite a headcount reduction of nearly three hundred people. The answer was in the terms. Every subscription in the estate was on a New Commerce annual term. Every single one. Their partner had quoted annual because it was cheaper, the buyer had accepted annual because it was cheaper, and nobody in the conversation had connected the price to the commitment. They had left an EA to buy flexibility and then bought the one CSP term that does not provide it. And when we benchmarked the price, quoting the same basket with two other partners, their margin was running about twelve percent above what the market offered for that estate size, because in four years nobody had tested it once. The fix was not complicated. We moved the volatile populations to monthly at the next term anniversary, kept the stable core on annual where the price is genuinely better, and renegotiated the partner agreement with the service written down instead of assumed. Net effect, about nine percent off the annual run rate. So my two questions for anyone on CSP, and you can answer both this week: which term is each of my subscriptions on, and when did I last benchmark my partner's price against another partner. If either answer is I do not know, that is where your money is.
They left an EA to buy flexibility and then bought the one CSP term that does not provide it, with a margin twelve percent above market because nobody had ever tested it. Two questions, answerable this week: which term is each subscription on, and when did I last benchmark the partner. Second check is that benchmark.
Check two. You quote the same twelve hundred seat basket with three CSP partners and the prices come back spread eleven percent apart. What have you learned, and what should you do? A, two partners are overcharging, take the cheapest and move on. B, you have measured the margin range without asking anyone to disclose it, so now price the service alongside it: the cheapest quote is right only if its support and administration match what the others include. C, the spread is a pricing error and should be queried with Microsoft. Or D, nothing useful, partner prices are arbitrary. Pause here. An eleven percent spread on identical software has to come from somewhere. Where?
The answer is B. Identical software at three different prices means the difference is margin and service, and a competitive quote has just measured it for you without anyone having to disclose anything, which is why I recommend this as the standard practice rather than as an aggressive move. But, and this is where A goes wrong, the cheapest number is only the best deal if the service behind it matches. A partner who absorbs your first line support, reclaims unused licences before your renewal, and produces monthly assigned versus active reporting is earning part of that spread and probably saving you more than the difference. A partner who forwards invoices is not. So the correct sequence is: benchmark the price, then specify the service in the partner agreement so that the margin has something to be proportionate to. D denies information sitting directly in front of you. And C misunderstands the model, partner pricing is the partner's own commercial decision inside Microsoft's programme rules, not a published rate that could be in error. One practical note: you do not need to run this every year. Every couple of years, on the same basket, is enough to keep the number honest.
Choosing and managing the partner, three things worth specifying in writing, because in CSP the partner is genuinely part of the product rather than a channel detail. Support that is real: named contacts, response commitments, and escalation into Microsoft when it is needed, and my favourite test question, ask what happens at two in the morning on a Sunday, and then ask for that answer in the agreement rather than in the pitch deck. Administration and reporting: monthly assigned versus active reporting, visibility of every term expiry, and proactive reclamation, and I will make a strong claim here, a partner who tells you about your unused licences before your renewal is worth materially more margin than one who does not, because they are actively reducing the bill they earn from. And portability: how subscriptions transfer if you change partner, what notice applies, what data comes with you, asked before you need it, because switching under pressure is exactly where continuity breaks and where an unhappy relationship becomes an expensive one. Session thirty six runs the full partner economics and a switching playbook. For today, the point is that the partner relationship is negotiable, reviewable, and changeable, and treating it as fixed infrastructure is how estates end up paying a premium for a mailbox.
When CSP wins, and this table is the honest version. Below roughly two thousand four hundred stable seats: strong, CSP typically undercuts the EA by five to twelve percent, because you stop paying for idle committed seats. Volatile or seasonal populations: strong, monthly terms let the count fall when the people leave, which is precisely the thing an EA cannot do. Unproven new products: strong, and this one matters right now, commitment risk is lower on CSP while adoption is unknown, which is directly relevant to Copilot in module four, where committing three years of seats to a product nobody has used yet is a genuine risk. Large stable standardised base: weak, above the break even the EA rate wins by six to fourteen percent and the flexibility is worth little to a population that does not move. And the last row is the important one, any estate with both shapes: hybrid, and the number is striking, the split returned nine to sixteen percent against a single vehicle across the decisions we advised. Which gives the finding I opened with: the cheapest outcome was almost never a single vehicle. Deciding the seat split is the actual work. Asking Microsoft for a rate is what happens afterwards.
Last check. A five thousand seat company has three thousand six hundred permanent staff and one thousand four hundred seasonal and project users who turn over constantly. Their EA renewal is due. The strongest structure: A, renew all five thousand on the EA for the best volume tier. B, move all five thousand to CSP for flexibility. C, split deliberately: the three thousand six hundred stable seats on the EA where the rate wins, and the volatile one thousand four hundred on CSP monthly terms where the count can fall, which is the pattern that returned nine to sixteen percent against a single vehicle. Or D, split randomly to hedge across both vehicles. Pause here. Two populations behave completely differently. Why would one vehicle price both of them well?
The answer is C, and the reasoning is that two populations with genuinely different behaviour should not be priced by one mechanism. The stable three thousand six hundred sit comfortably above the break even, where the EA rate wins by six to fourteen percent and the commitment costs them nothing because they are not going anywhere. The volatile fourteen hundred are exactly the seats an enterprise wide commitment punishes you for: they arrive at a true up and, as session two established, they never leave. A buys a better volume tier and then pays for fourteen hundred seats throughout their entire absence, which is the worse deal wearing the better discount. B abandons a real rate advantage on seventy two percent of the estate to solve a problem that only affects twenty eight percent of it. D is the caricature that makes the correct answer sound unserious, and I include it deliberately, because a split is only valuable when it is drawn along the line where behaviour actually changes, which requires knowing which of your seats churn. If you cannot draw that line, you cannot run this play, which is exactly why the homework is to draw it.
The CSP discipline, three habits, none of them difficult, all of them skipped in the estates that overpay. Match the term to the seat: monthly for populations that churn, annual or three year for the ones that do not, reviewed each year because behaviour changes, and understand that running one term across an entire estate guarantees one of two failures, paying for flexibility you never use or lacking it exactly where you need it. Benchmark the margin periodically: a competitive quote every couple of years on the same basket with the service specified, which keeps the number honest without a confrontation and tells you what the market rate for an estate your size actually looks like. And keep the term calendar: every subscription's term end and reduction window, with an owner, because in CSP the flexibility is genuinely real and it lives inside specific windows, so a missed window is not an inconvenience, it is another year of seats. Three habits, maybe two hours a year in total. Tom's client got nine percent off their run rate by doing all three at once after four years of doing none of them. Next session brings all three vehicles together into a single decision framework.
Session four, three sentences. One: CSP is the indirect route where a partner bills you, supports you, and prices you, which makes the partner part of the product and the margin inside your price legitimate, variable, and entirely worth measuring. Two: the New Commerce term is the product, monthly buys real flexibility at a premium and annual locks quantities for twelve months, and treating an annual term as though it were monthly is the single most common CSP disappointment, appearing in a quarter to a half of the estates we reviewed. Three: CSP wins below the break even, on volatile populations, and on unproven products like Copilot, and in estates that contain both shapes the deliberate split beat a single vehicle by nine to sixteen percent. Next week we put all three vehicles on one page and build the decision framework. See you there.
Homework, about an hour, and it produces the input for next session. One, segment the population: from HR and identity data, permanent staff, fixed term and seasonal, contractors, project users, four numbers, and rough is fine. Two, measure the churn: how many licences were added and removed in each segment over the last twelve months, because it is the churn rate rather than the headcount that decides which term fits. Three, find your CSP terms if you already buy through a partner: which subscriptions are monthly, annual, or three year, and when does each term end, and I will predict now that most estates cannot answer this on demand, which is itself the finding. Four, price the flexibility: for your most volatile segment, what would those seats have cost on an annual lock versus monthly over the last year, because that difference is what flexibility was actually worth to you, in your own numbers rather than in principle. And five, draft the split: one line, which populations belong on a committed vehicle and which on a flexible one. Bring that line to session five, where we build the full framework around it.
Five reads before next session, all free on redress compliance dot com. First, the CSP buyer side guide, which carries the New Commerce terms, the partner margin, and the review data behind this session, including the twenty five to forty percent finding on annual terms. Second, the CSP versus EA pillar for 2026, which frames the choice as a portfolio question and is the natural preparation for session five. Third, the CSP licence guide, for the programme mechanics in reference form when you need the detail. Fourth, the Microsoft CSP knowledge hub, which collects everything else including partner selection material. And fifth, Azure CSP versus EA, because the same choice behaves differently for Azure consumption than it does for seats, and module five will come back to it. That is session four. The partner is part of the product, the term is the product, and the split is usually the answer. Next week, all three vehicles, one framework. See you there.