What an alliance actually runs on

The short version

We once closed the price gap to $15,000 on a $42M bid and still lost, because I had assumed the partner's executives were behind us instead of securing their commitment. Most of what I know about alliances is a longer version of that lesson, worked out at NTT across 31 business units and then inside AWS on the global MongoDB partnership.

• An announcement is a promise made by executives. Sellers keep it only when their pay and their rules of engagement point the same way as the slide.
• Executive commitment has to be secured early, at a level that can settle a field conflict. Assumed support fails at the worst possible moment.
• A partner's pay change can split two field teams working the same account, or join them. When MongoDB moved its reps' credit onto annualized consumption, we used it to keep both field teams selling MongoDB Atlas together.
• Rules of engagement, deal registration and a record of who sourced every deal have to exist before the first conflict, because an alliance whose numbers finance cannot trace loses its budget.
• Cloud marketplaces turn a customer's committed cloud spend into the shortest buying path. The motion my team built ran 727 AWS Marketplace transactions worth $130M.
• A healthy alliance runs to one plan, reviewed every quarter, across teams the alliance leader does not own, and it needs a reason to grow that neither side had alone.

1. Alliances are announced on a slide and decided in the field.

The announcement deserves credit first, because a joint press release and a logo slide do real work: they tell customers that two companies intend to show up together, and they give sellers on both sides permission to pick up the phone. Executives need a public moment like that to put their names to something.

Trouble starts the morning after, when sellers on each side go back to quotas set before the partnership existed and to credit rules that may not recognize a joint deal, while the account conflicts nobody settled are still waiting for them. Whether the alliance produces anything depends on what those sellers are paid to do and on who settles the first fight.

Research on partner programs points the same way: PartnerStack and Wynter surveyed 100 senior revenue, marketing and partnership leaders at B2B software companies with $50M or more in revenue, and published the results in September 2025. Internal team misalignment was the top blocker to partner revenue for 37 percent of them. Smaller groups named their inability to see what partners contribute to the funnel (14 percent) and a lack of executive support (9 percent). Read from the failure side, those answers line up with what an alliance actually runs on.

Exhibit 1 · What carries the announcement

Everything else in the alliance standards I measure myself against sits on top of those columns: enablement, co-selling, the marketplace, joint planning and the data that manages the motion. None of it holds when the field is paid to compete.

2. Executive commitment has to be secured early, because assumed support fails at the worst moment.

On that bid we were the challenger, and afterward the customer's own win/loss review said our offer was the better one and that price was a non-issue. The incumbent was also one of the customer's banking clients, and when the customer asked our partner which bidder was better, the partner's executive answered that a customer that size should call the partner first. I had assumed the partner's support instead of securing it. Since then, on every deal where a partner matters, I have secured that support at the executive level, early. Secured means a named executive on the partner's side has said yes to what the partner will put into the deal. A friendly field team and a slide showing two logos together do not count.

I built the NTT cross-sell program for the Americas on the same rule, and it reported into the NTT Global Business Office in Tokyo. Within its first 90 days the leaders of every operating company had signed up to it and agreed who leads what. Operating-company leadership sat in every regional review after that. Over 22 months the program delivered more than $200M in contract value and more than $1B in pipeline across 31 business units.

At AWS the weekly whole-team call on the MongoDB account always asked where executives needed to be involved. Asking every week kept the question from waiting until a deal was already in trouble, which is when executives are least able to help.

Take a hypothetical case, in which you are leading a large bid where a partner's software sits inside your solution and the partner's field team has been encouraging. I would put the partner's executive commitment on the deal review as its own line, find the executive who can commit the partner's price and people, and secure that commitment before the final number goes in. If I could not get it, I would treat the deal as unqualified, however warm the relationship in the field felt.

3. Pay decides whether two field teams sell together or against each other.

Compensation neutrality is the alliance standard that gets the least attention in the announcement and decides the most in the field. The principle is simple: a direct seller should earn credit on a deal the partner brings, or at least lose nothing by helping, so that helping is the rational choice.

At NTT I designed the cross-sell program's cross-entity compensation model, which protected the credit of any seller whose deal booked in a sister company. Without it, 31 business units with separate legal entities and separate P&Ls would have had every reason to keep their accounts to themselves.

The test at AWS came from the partner's side: I owned the account plan for the global MongoDB partnership, and MongoDB moved its reps' credit onto annualized consumption. A change like that alters what a partner's seller wants from every account, and it lands on the other field team whether anyone planned for it or not. I read it as the opening, and worked with the AWS partner organization so both field teams kept selling MongoDB Atlas together.

It was an opening because consumption is the way a cloud provider already earns its revenue. A partner seller paid on what customers actually run wants workloads live and growing in the account, and that is the outcome the cloud provider wants too. Left alone, the same change can push two teams apart, because each keeps chasing its own credit in the same customer.

Exhibit 2 · One pay change, two possible field motions

We carried a $172M number on that account and grew it 24.6% year over year in a down market. Growth in that year came from keeping both field teams pointed the same way, starting with how MongoDB's own sellers were paid.

4. Rules of engagement and attribution settle the fight before it starts.

Every alliance meets its first conflict within weeks: two sellers claiming the same opportunity, or a partner registering a deal the direct team believed was already theirs. If the rules for those moments are written afterward, whoever escalated loudest ends up writing them.

Across 31 business units at NTT, with separate legal entities and separate P&Ls, we set who is prime, secondary and tertiary on a deal and who leads on which capability. Deal intake qualified each cross-sell opportunity and put a cross-company team on it. At AWS we ran MongoDB's deal registration process and tracked it in our reporting, and we ran deal registration with OEM partners as well.

Attribution sounds like bookkeeping, yet it is the part that keeps an alliance funded. For every opportunity on the MongoDB account we recorded which side sourced it, MongoDB or the AWS account team. In the NTT program we always recorded the originating company, even when another company ended up prime on the contract. Partner-sourced and partner-influenced revenue have to be reported separately, because the first time finance doubts a partner-sourced number, the alliance loses its budget and the field loses its reason to share.

5. The marketplace turns a customer's committed cloud spend into the shortest buying path.

Large enterprises sign multi-year commitments to spend a set amount with their cloud provider. Canalys estimated in August 2024 that enterprise customers had committed more than $360 billion to the three largest providers on a multi-year basis, and forecast that software sold through those providers' marketplaces would grow from $16 billion in 2023 to $85 billion in 2028. Tackle's 2025 State of Cloud GTM report, a survey of people who run marketplace and co-sell programs (three quarters of them in alliance and partner roles), found 74 percent reporting a significant benefit from access to customers' committed spend.

The mechanism is plain: a customer can use the commitment on cloud services directly or, where the provider's rules make a listing eligible, on software bought through the marketplace. Instead of a new vendor setup and a fresh budget, the purchase draws down money already committed, and it lands on a bill procurement already pays.

Exhibit 4 · Two ways to buy the partner's product

That is the path my team built for MongoDB at AWS, where the partner and marketplace motion ran 727 AWS Marketplace transactions worth $130M, and MongoDB named AWS its Cloud Partner of the Year for Marketplace in 2023 while my team held the account. MongoDB's own citation for the award pointed to AWS Marketplace leading in transactions for MongoDB customers, and to the MongoDB Atlas pay-as-you-go listing making Atlas easier to buy.

I would build the marketplace path early in any alliance with a cloud provider. A buyer who can spend money already committed has one fewer reason to wait, and the budget question is often the last thing standing between a buyer who agrees and a signed order.

6. One plan has to run teams the alliance leader does not own.

On paper the MongoDB account team looked simple, since I owned the account plan and led the fourteen-person account team behind it. The sell-to team reported to me directly, and the partner and service teams aligned to my plan.

Aligning them came down to rhythm: a weekly whole-team call covered what was pressing and where executives needed to step in, a biweekly call with each team covered plans and operations, and a quarterly account review measured every team against the plan. With MongoDB itself we ran joint business plans and reviewed them every quarter.

Exhibit 5 · One account plan, three teams, one partner

Authority over people you do not manage comes from the plan being the most useful document in the room. When the partner team can find its own targets inside the account plan, and the service team can see where its work lands in the number, they run to it because it helps them do their jobs.

Enablement belongs in that same rhythm, and in the NTT program it ran through the regional quarterly business reviews and through lunch-and-learns held twice a month in sales offices.

7. A healthy alliance is measured every quarter and given a reason to grow.

Alliances drift when nobody measures them between announcements, and the NTT cross-sell program was measured in Salesforce, with resources following the results. My region was Top Region in the NTT Group three years running, from 2016 to 2018. The scorecard below is how I would review any alliance each quarter.

A scorecard keeps an alliance honest, and growth needs something more: a reason to sell together that neither side had alone. In 2023 I co-led the AWS work that brought MongoDB into Amazon CodeWhisperer, AWS's AI coding assistant. MongoDB and AWS announced it on 6 November 2023, describing CodeWhisperer trained on curated MongoDB content to give developers better suggestions when building on MongoDB. Making MongoDB easier to build on with AI gave both companies a new conversation with developers.

AI now helps with the mechanics as well, because registration, attribution and joint pipeline are exactly the records an AI can read and reconcile every week, provided every figure carries its source and a person makes the judgment calls.

8. What this means for alliance and sales leaders

1. Map the money before the logo. Find where the partner already touches your customers and who gets paid when they do. That map tells you where the alliance will work first.
2. Secure the partner's executive support before the bid. Make it a line in every deal review where a partner matters, and treat assumed support as a reason the deal is not qualified.
3. Settle pay before the first joint deal. Every seller should come out whole wherever the deal books. When a partner changes how its own sellers are paid, decide quickly which way that change will push your field.
4. Write the rules of engagement before the first conflict. Decide who leads and who is prime, run deal registration through your reporting, and record which side sourced every opportunity so finance can trust the number.
5. Build the marketplace path against committed spend. Help customers buy the partner's product with money they have already committed to the cloud provider.
6. Run the alliance on one plan and a quarterly scorecard. Prove the motion on a handful of accounts, then scale it through marketplaces and systems integrators.

The $42M bid taught me the expensive version of all this: an alliance is worth what its two field teams are paid to do together, and the executives above them have to say yes before the bid goes in.

Sources

• PartnerStack and Wynter, The State of Partnerships in GTM 2026 (September 2025), a survey of 100 senior revenue, marketing and partnership leaders at B2B software companies with $50M or more in revenue: 37 percent name team misalignment as their top partnerships blocker, 14 percent lack visibility into partner impact, 9 percent lack executive support. partnerstack.com, research lab chart and press release.
• Canalys, "Hyperscaler cloud marketplace sales to hit US$85 billion by 2028", 16 August 2024: $16 billion in 2023, $85 billion forecast for 2028, more than $360 billion in multi-year customer commitments to the top three providers. canalys.com. Also reported by ChannelDrive.
• Tackle, 2025 State of Cloud GTM Report: 74 percent of respondents report a significant benefit from access to committed spend with cloud providers, which customers can use for direct cloud spend or for marketplace spend. Respondents are involved in marketplace selling, co-selling or both, and 75 percent work in cloud alliances or partner management. tackle.io, report PDF.
• MongoDB, 2023 Partner of the Year awards, naming AWS Cloud Partner of the Year for Marketplace and citing AWS Marketplace transactions and the MongoDB Atlas pay-as-you-go listing: mongodb.com blog.
• MongoDB and AWS, "MongoDB and AWS Collaborate to Optimize Amazon CodeWhisperer", 6 November 2023: mongodb.com press release.
• AWS MongoDB account figures ($172M, 24.6%, 727 transactions, $130M): AWS performance record, by reference. NTT cross-sell program figures: NTT records, by reference.