The Partner Line

One process. Four kinds of partners.

A practical guide to building and running partner programs with technology companies, hyperscalers and marketplaces, GSIs, and resellers.

The fundamentals are pretty consistent across all four. You still need to choose the right partners, build something worthwhile together, get their teams up to speed, work on deals, and measure what's happening. But the way you approach each step changes quite a bit depending on who you're working with.

And with AI agents starting to change how software gets built, sold, and delivered, there's another dimension to consider as we head into 2027.

By Andrew Mackenzie. Last updated November 2026.

Start with the big idea About me

Why partners increasingly decide who wins

A lot of enterprise technology is no longer bought through a straightforward vendor-to-customer relationship. Partners influence what gets considered, how products are implemented, where customers spend their budgets, and whether they stick around.

That makes your partner strategy a pretty important part of your overall go-to-market strategy. It shouldn't be something you get around to once the direct sales team is up and running.

A few numbers help explain why.

70%

of global IT spend in 2025 goes to, through, or with channel partners.

Canalys, 2025 1
19.6%

of enterprise software spending already goes through cloud marketplaces, based on a survey of 833 buyers.

Futurum, 2026 8
$163B

projected annual spend through hyperscaler marketplaces by 2030, up from $16B in 2023. Around 59% is expected to flow through channel partners.

Canalys / Omdia 2, 3
58%

less likely to churn when customers use a product's integrations.

Crossbeam vendor survey, 2023 4

A note on the numbers. Some of the research in this space is commissioned by cloud providers and software companies with a commercial interest in the findings. I've called that out where relevant. The figures are useful for understanding the direction of travel, but I'd always put your own customer and partner data first.

Start by understanding what the partner actually wants

One of the mistakes I see in partner programs is assuming that every partner is motivated by the same things.

They're not.

A technology company might care most about retention. A cloud provider wants to grow consumption. A GSI is looking for services revenue and opportunities to expand its practice. A reseller wants a healthy margin and a reason to keep investing in the relationship.

These differences matter more than most partner teams realise.

In my experience, partnerships tend to work when what you're offering helps the other company achieve something it already cares about. They tend to struggle when the value is mostly on your side.

If there's one principle I'd take away from this guide, it's to understand what your partner gets paid for, then build the relationship around helping them do more of it.

  1. Technology partners

    Their currency is Retention

    They want integrations that make their product more valuable and harder for customers to leave. They also want access to new customers and opportunities that help their own sales pipeline.

  2. Hyperscalers and marketplaces

    Their currency is Consumption

    Cloud providers want customers using more of their infrastructure. Your product can help them get there, particularly when a deal draws down a customer's existing cloud commitment and contributes to the account team's targets.

  3. Global systems integrators

    Their currency is Services revenue

    GSIs need opportunities to sell consulting, implementation, and ongoing services. Your product should give them a reason to build a practice around it and bring it into larger transformation projects.

  4. Resellers

    Their currency is Profit

    Resellers need to make money on the deals they bring in. Increasingly, they also want recurring revenue from managing and supporting customers after the initial sale.

    Both matter, and both need to be protected. If your direct sales team keeps taking the opportunities or the ongoing services revenue, don't expect resellers to stay motivated.

The five-step process

These are the five steps I'd use to build and run a partner program. The process stays pretty much the same regardless of partner type, but the priorities and tactics can be very different.

Choose a partner type to see how I'd approach each one. For every step, I'll cover the move I'd make, the mistake I'd watch out for, the metric I'd track, what's likely to change in 2027, and an example from a company that's put it into practice.

  1. Choose

    Be selective. The right five partners are worth more than fifty names on a spreadsheet.

    Every partnership takes time to build and maintain. Before recruiting another partner, I'd want to know why they're a good fit and what both sides stand to gain.

    Is this partner worth investing in?

    Here's a quick checklist I'd use before recommending a partner to a customer.

    0 of 6. Tick the traits this partner has.

    Use the checklist to get a better sense of where a potential partner stands. Not every partner needs to tick every box, but the more of these traits they have, the more likely they are to become a productive relationship.

  2. Build

    Start small. Get something working before you try to build a big partnership.

    Expect the first project to be rough

    Two companies can agree that there's a great opportunity to work together and still struggle to get the first project off the ground. Different approval processes, competing priorities, and unclear ownership can slow things down surprisingly quickly.

    I'd plan for that rather than assume everything will fall into place. A few things help.

    • Keep the first project small enough to complete in a few weeks.
    • Agree on one clear measure of success.
    • Put a named owner on each side who is accountable for getting it done.
    • Understand the other company's legal, security, and technical approval processes before starting development.

    A good first project might be a joint sales training session or a simple integration. The actual deliverable matters, but so does learning how the two companies work together.

  3. Enable

    Give their salespeople a reason to sell your product, and make it easy for them to do so.

    You can put together the best partner training materials in the world, but they won't make much difference if a partner's sales team can't confidently explain your product to a customer.

    The quickest way to change that is to give them something they can actually use.

    A live demo using a prospect's own data is usually far more effective than another presentation deck. It gives the seller a chance to show how the products work together in a real situation.

  4. Co-sell

    Turn the relationship into actual opportunities.

    At some point, the partnership needs to move beyond training sessions, integrations, and joint announcements. You need to see it showing up in real customer conversations and sales opportunities.

    This is where a lot of programs struggle. Both companies may agree to work together, but nobody has a clear plan for identifying deals, making introductions, or following up.

  5. Measure

    Look at what the partnership actually delivers, not how busy everyone is.

    Partner teams can easily fall into the trap of measuring activity. How many training sessions did we run? How many integrations did we launch? How many partners attended the quarterly business review?

    Those numbers have their place, but they don't tell you whether the partnership is working.

    I'd start with the outcome that matters most to the particular partner and work backwards from there.

From the field

The views below are my own. They're based on more than a decade of working on partner programs, rather than an industry consensus or the official position of any company I've worked for.

Some of these lessons took longer to learn than they should have. Hopefully, sharing them saves someone else a bit of time.

GSIs

A GSI partnership isn't real until someone's number depends on it.

It's relatively easy to sign a global agreement with a large systems integrator. Getting anyone inside that organisation to do something with it is another matter.

The partnerships I've seen work best all had something in common. There was a practice leader whose own targets were affected by the success of the relationship.

That's the person you need to find.

Before signing a big agreement, work out who owns the relevant practice, what they're trying to achieve, and how your product helps them get there. Build the business case around their priorities, not just your own. If necessary, put some money behind the first few engagements to help get things moving.

A logo on a partner page doesn't create a pipeline. Having someone internally who is genuinely motivated to build the business does.

Hyperscalers

Cloud sellers don't need another product pitch. They need to know how you help them hit their number.

I've seen plenty of software companies approach cloud providers with a standard product presentation. That's usually not what the account teams are looking for.

The people selling cloud services care about their customers' consumption, committed spend, and account targets. If your product helps them grow cloud usage or get a customer to draw down an existing commitment, you've got a much more interesting conversation.

Lead with that.

Understand how the cloud provider measures its sales teams, make marketplace private offers a normal part of how you transact, and get involved in account planning wherever possible.

The goal is simple. Become the partner who makes it easier for the cloud account team to have a good quarter.

Technology partners

Build the integration before you announce the partnership.

There's a lot of pressure in AI to announce partnerships quickly. Everyone wants to be associated with the latest model provider, platform, or AI company.

But an announcement doesn't mean customers have a reason to care.

I'd much rather see two companies release a small integration that people actually use than announce a strategic partnership with nothing meaningful behind it.

Find one workflow where the products genuinely complement each other. Build something useful, get customers using it, and pay attention to activation.

Once you've got that, co-marketing becomes much easier. You have real customers and real use cases to talk about, rather than a partnership announcement that everyone forgets a week later.

Model providers

Your model provider might build your feature. Figure out what it can't easily replace.

This is one of the more uncomfortable realities for software companies building in AI.

The model providers are moving incredibly quickly. Features that once gave an independent software vendor a meaningful advantage can suddenly appear in a model provider's next release.

I've watched AI labs become very good, very quickly, at building capabilities that other companies previously considered their own territory.

Trying to stay ahead by simply building a slightly better feature isn't a particularly sustainable strategy.

Instead, think about what you have that a model provider would struggle to replicate. It might be proprietary data, deep expertise in a particular workflow, established customer relationships, or the ability to work across multiple models without being tied to one provider.

Build your business around those advantages, and don't wait for a model provider to give you permission to grow.

There's another practical point worth considering. Model providers often reach out when they're preparing to launch something and need credible partners around it. The time between getting invited and the actual launch can be surprisingly short.

If you're building in this space, keep some engineering capacity available. Being able to respond quickly to a new model release can create opportunities that disappear if you need three months to get something built.

Channel

Your first few channel partners will teach you more than the next hundred.

When companies start building a channel program, there's often a temptation to recruit as many partners as possible. The thinking is that more partners mean more reach and, eventually, more revenue.

I'd take a different approach.

Start with a small group of partners and work closely with them. Get involved in their sales conversations, understand how they position your product, and pay attention to where deals get stuck.

You'll learn things about your product, pricing, sales process, and onboarding that you probably wouldn't discover sitting inside your own company.

Use those lessons to build the channel playbook. Once you have something that works, start recruiting more broadly.

Recruiting hundreds of partners before you've figured out how to help them sell is a good way to create a much bigger version of the same problem.

Every partner type

Your toughest partner to win over might be your own sales team.

I've seen partner programs struggle not because the external partners weren't interested, but because the company's own sales organisation wasn't particularly keen on working with them.

It's understandable. If a direct sales rep thinks bringing a partner into a deal means giving up revenue or commission, they're going to be reluctant to make that introduction.

You can build the best partner strategy in the world, but it won't get very far if your compensation plan encourages the direct team to work around it.

Fix the incentives first.

Make sure your salespeople benefit when partners help close a deal. Get a senior executive to visibly support the program, and make a point of sharing partner wins internally.

The partner team can't spend all its time trying to build external relationships while fighting its own sales organisation at the same time.

Check a partner's health

Partnerships rarely fall apart overnight. More often, things gradually go quiet. People stop responding, training attendance drops, opportunities dry up, and nobody quite knows who's supposed to do what.

By the time someone notices, the relationship may already be in trouble.

I like to do a simple health check every quarter. It doesn't need to be complicated. The idea is to take an honest look at three things, which are engagement, capability, and performance.

Rate each one as Strong, Mixed, or Weak. It's a rule of thumb rather than a scientific scoring model, but it's a useful way to spot problems before they become bigger ones.

1. Engagement

Are they actually showing up?

Look at how responsive the partner is, whether people attend enablement sessions, how active they are in your partner portal, and whether anyone senior is involved.

A partnership where nobody has the time or authority to make decisions is going to struggle, regardless of how promising it looked at the start.

2. Capability

Are they putting real resources behind the relationship?

Look for certified people, dedicated resources, a named practice or business lead, and evidence that they're investing time in learning and demonstrating your product.

A partner might be enthusiastic about working together, but enthusiasm isn't the same as having the capability to deliver.

3. Performance

Is the partnership producing results?

Look at registered pipeline, win rates, delivery quality, revenue trends, and what happens after customers go live.

I'd pay particular attention to customer expansion and adoption. Winning the initial deal is important, but a partner who helps customers get more value from your product is often much more valuable over time.

Rate all three areas

Your result and a recommended next step will appear here.

What to do with the results

  • If engagement is strong but performance is weak, the partner may need better sales support or a clearer proposition.
  • If capability is weak, you may need to invest in training, certification, or a more focused solution.
  • If engagement is weak across the board, it's worth having a direct conversation about whether the relationship is still a priority for them.
  • And if all three areas are weak, be realistic about how much more time and money you want to invest.

Run the same assessment every quarter. The trend is often more useful than any individual score.

Looking ahead to 2027

AI agents are starting to change the economics of partner businesses. They're affecting what partners sell, how they deliver services, and how they charge customers.

They're also changing how customers discover solutions in the first place.

I see five shifts that partner leaders should be thinking about, regardless of whether they work with technology companies, hyperscalers, GSIs, or resellers.

  1. Reselling licences and running projects Delivering outcomes with AI agents

    Partners are adding AI and agents to more of their service offerings. Some of the traditional work, particularly routine migration, implementation, and support tasks, will become cheaper as agents take on more of the workload.

    That doesn't necessarily mean less opportunity for partners. It means the opportunity is moving.

    There will be more demand for partners who can design, build, deploy, and manage AI agents that solve specific customer problems.

    The number worth watching

    82% of AWS partners already deliver some form of AI to customers. Omdia 2025, via AWS 5

  2. Hourly rates and fixed fees Pricing based on business outcomes

    Customers are becoming more interested in paying for results rather than simply paying for the time it takes to deliver something.

    That creates an opportunity for partners to rethink how they package and price their services.

    Instead of selling a fixed number of consulting hours, consider offers where both the software vendor and the partner have a financial incentive to achieve a measurable business outcome.

    It also changes who you need to sell to. These conversations increasingly belong with business leaders who own the outcome, not just the IT team responsible for buying the technology.

    The numbers worth watching

    Outcome-based pricing doubled from 18% to 36% of engagements, while fixed-price work fell from 49% to 29%. The same research found that 75% of line-of-business buyers consider partner relationships essential. IDC, via AWS 5

  3. Individual vendor listings Multi-vendor solutions discovered through AI

    Cloud marketplaces are becoming more interesting than simple places to buy software licences.

    We're starting to see offers that combine several software products with professional services, all packaged together and billed through a single marketplace transaction.

    That's a meaningful change for partner teams.

    A customer might describe a problem to an AI assistant and get recommended a solution involving three or four different vendors and a services partner. The opportunity is no longer just about getting your own product listed. It's about being part of the solution that gets recommended.

    Your marketplace listing also needs to do a better job of explaining what the product actually helps customers achieve. That information needs to make sense to a human buyer and be easy for AI-powered discovery tools to interpret.

    There are already some interesting examples.

    AWS Marketplace introduced multi-product solutions, more flexible service payment options, and an AI agent mode for discovery in late 2025. AWS 5

    Google Cloud has also introduced an agent finder that lets buyers search for partner-built agents using natural language. Those agents can be priced against outcomes, such as the number of support tickets resolved. Google Cloud, October 2025 12

  4. Partners selling your product Partners building their own IP on your platform

    The best partners will increasingly do more than implement and resell your software.

    They'll build their own intellectual property on top of it, such as industry-specific agents, reusable accelerators, specialised workflows, and packaged solutions that they can sell repeatedly.

    That's a much more interesting relationship than one based entirely on referral fees or resale margins.

    Your job is to give partners the tools, APIs, documentation, and support they need to build those solutions.

    If they can make more money by building on your platform, and their own IP becomes more valuable because your product is underneath it, you've created a stronger reason for them to stay.

  5. Closing the deal and moving on Driving usage after the sale

    The traditional software sales model put a lot of emphasis on getting the contract signed. Increasingly, particularly with consumption-based pricing, that's only the beginning.

    When customers pay for API calls, tokens, or compute, a signed contract doesn't automatically translate into meaningful revenue. Customers need to use the product, and ideally find more ways to use it over time.

    This is particularly important with AI. Customers are experimenting quickly, but many still need help turning pilots into something that works reliably in production.

    That creates a significant opportunity for partners who stay involved after go-live. They can help customers identify new use cases, improve adoption, and make sure the technology is delivering something useful.

    These are the partners who become difficult to replace.

    The number worth watching

    Canalys estimated that helping customers adopt and manage Google Cloud accounts for 35% of the partner services opportunity around the platform. Its research also suggested that AI projects typically take until their third year to move from pilot to production. Canalys, commissioned by Google Cloud, 2025 9

Run your partner team on agents

If you're asking your partners to sell AI and build AI-powered solutions, there's a pretty good argument for using it inside your own partner organisation first.

Partner management involves a surprising amount of repetitive work. Researching accounts, checking customer overlaps, following up on referrals, updating partner portals, answering the same product questions, and putting together quarterly reviews can easily consume most of a team's week.

A lot of that work is necessary. But it doesn't all need to be done manually.

Here's how I'd think about introducing agents into a partner organisation, where they can help, and how to get started without turning the whole thing into an expensive technology project.

Why bother?

The obvious reason is productivity. If an agent can take care of some of the research, administration, and follow-up work, your team gets more time to spend with partners.

Salesforce's 2026 State of Sales report found that sellers spend only around 40% of their time actually selling. Salespeople using agents expect research time to fall by 34% and email drafting time by 36%. Salesforce State of Sales, 2026 6

There's another reason that's easy to overlook, and that's credibility.

It's much easier to have a useful conversation with a partner about AI when you've already worked through some of the practical challenges yourself. You'll have a much better understanding of what works, where the limitations are, and where human involvement still matters.

When should you use an agent?

Not every task needs an agent. In fact, I'd be pretty cautious about trying to automate everything at once.

A task is a good candidate when four things are true.

  • It happens regularly, ideally every week.
  • The information needed to complete it already exists somewhere, such as your CRM or partner portal.
  • Someone on the team understands the task well enough to check whether the result is correct.
  • You have a reasonable way to measure how long it takes today.

I'd leave relationship-building, sensitive negotiations, and important commercial decisions with people.

There's also a good reason not to rush. Gartner predicted that more than 40% of agentic AI projects could be scrapped by 2027 because of costs, unclear business value, or inadequate risk controls. Gartner, June 2025 7

Start with a problem that's easy to understand and has a measurable payoff. Don't start with an agent just because it's technically possible to build one.

How I'd approach it

I'd take a fairly gradual approach.

  1. Map the workflow first. Pick one task and write down exactly how the team currently completes it. Don't skip this step. If you don't understand the existing process, you won't know whether the agent is improving it.
  2. Start with AI assistance. Have someone use AI to help with the work they're already doing. Compare the quality of the output and the time involved.
  3. Let an agent prepare the work. Once you're comfortable with the results, let an agent prepare drafts, research, or recommendations. A person still reviews and approves everything.
  4. Give it more responsibility gradually. Only allow an agent to take actions independently once you've established clear rules, tested the workflow, and understand what can go wrong.
  5. Measure the results. Track time saved, accuracy, and how much manual correction is still needed. If the results are good, move on to another workflow.

One thing I'd put in place from the beginning is a clear set of guardrails. Anything being sent to a partner or customer should have human approval. Agents should only access the information they need, and their actions should be logged.

Where agents can help across the five steps

There are opportunities throughout the partner lifecycle.

Choose

Research potential partners, look at account overlap reports, and turn the findings into a shortlist ranked by relevance.

Build

Prepare first drafts of joint solution briefs, integration requirements, and marketplace listings for individual partners.

Enable

Answer routine product questions, help partners find the right documentation, and generate demo scripts and sample data for specific prospects.

Co-sell

Identify account overlaps, route opportunities to the right salespeople, and keep referral information in sync with cloud partner portals.

Measure

Review partner activity and performance every week. Flag relationships that are starting to deteriorate rather than waiting until the next quarterly business review.

Crawl, walk, run

I'd approach this as an 18-month journey rather than a project that needs to be completed in one quarter.

The timings below are rough guides for a mid-sized partner team. You might move faster or slower depending on your systems, team size, and appetite for change.

Select a stage to see what it involves.

Want to compare notes?

I'm Andrew Mackenzie. For more than a decade I've built partner programs at Snyk, 360pi, and Phreesia, and advised early-stage SaaS companies on partner strategy. Together, that work covers all four partner types on this page.

If you're designing or fixing a partner program, I'm happy to brainstorm. Pick a 30-minute slot that suits you.

Book a call
  • Hyperscalers. Ran Snyk's Google Cloud relationship, including Google Cloud Marketplace and co-sell. Recognised as Google Cloud's 2025 Technology Partner of the Year for Application Development.
  • Technology partners. Launched Snyk's AI partner program, building integrations with AI companies including Anthropic, OpenAI, Cursor, and Cognition.
  • GSIs. Led Snyk's global Accenture relationship, including a co-investment to build a dedicated Snyk practice. Earlier, launched 360pi's first alliance program with partners including IBM, SAS, and Wipro.
  • Channel. Built Phreesia's initial EMR channel program.