Referral, Reseller, or Build: Which Partner Model Should You Start With?

Referral, Reseller, or Build: Which Partner Model Should You Start With?

Referral, almost always. It is the only model where a partner can send you revenue without either side changing how they operate. Reselling means your partner owns the customer relationship and the invoice, which changes your pricing, your support and your revenue recognition all at once.

The reason this gets muddled is that "partnership" describes four different commercial arrangements, and companies commit to the wrong one because a single enthusiastic prospect asked for it.

Here is how the large platforms actually structure this, and a framework for choosing.

What Are the Models, Really?

AWS publishes one of the clearest taxonomies, and it splits partners into five paths. Its Software Path is "for organizations that develop software that runs on or is integrated with AWS." Its Hardware Path covers devices. Its Services Path is "for organizations that deliver consulting, professional, managed, and value-added resale services." Its Training Path covers organisations that "sell, deliver, or incorporate AWS training." And its Distribution Path is "for organizations that recruit, onboard, and support their partners to resell and develop AWS solutions."

Notice what that structure reveals. Reselling is not a separate path in AWS's model; it sits inside services, alongside consulting and managed services. And distribution exists as its own path because managing a network of resellers is a different job from reselling.

For an early B2B SaaS company, only two of those five are realistic in year one: something like the Software Path if you integrate, and a referral arrangement that eventually grows into services.

How Does a Referral Model Actually Pay?

As a commission on revenue you collect, for a fixed period. HubSpot's Solutions Partner Program is a useful published example, because it states its terms plainly: "Partners earn 20% commission for 3 years for deals they bring to HubSpot."

It also documents a second structure for co-selling: partners can earn "20% commission for 1 year on specific types of upmarket deals that partners co-sell with us, that originated from our direct sales team." So the partner who sourced the deal earns for three years, and the partner who helped close a deal you sourced earns for one.

That asymmetry is worth copying. It prices the thing that is genuinely scarce, which is a partner bringing you a customer you would not have found, rather than a partner helping with a customer you already had.

What Do Partners Actually Get From You Besides Money?

Access to customers, and it matters more than the commission. HubSpot describes its partners as "service firms, consultancies, and agencies that help businesses attract, engage, and retain mid-market and enterprise customers," specialising in "consulting on AI, business, technology, sales, marketing, or customer service strategy," in "tech implementation that handles CRM, systems integrations, agentic builds, or IT services," and in "hands-on services in marketing, sales, customer service, or AI transformation."

Those firms are not primarily motivated by your 20%. They are motivated by billable implementation work, by being findable in your directory, and by co-selling access to your pipeline. The commission is a tiebreaker.

HubSpot also charges for entry, listing a Solutions Partner Package starting at $400 per month. That is a design choice worth understanding: a paid programme filters for partners with a real commercial intent, and it funds the programme rather than making it a cost centre.

When Is Reselling Worth the Complexity?

When your buyer cannot or will not purchase from you directly. That is the honest test, and it is usually about procurement rather than product. A government body with an approved supplier list. A large enterprise that will only transact through an existing vendor. A market where you have no legal entity.

In those cases reselling is not a growth tactic, it is a route to revenue you otherwise cannot access. That is a good reason. Wanting more pipeline is not, because a reseller adds a layer between you and the customer at exactly the stage where you most need to learn from them.

The costs are real and they arrive together: you lose direct pricing control, you inherit a support chain where the customer calls someone else first, and your renewal depends on a relationship you do not own. Before you agree to any of that, make sure you have a pricing structure that can survive a discount tier, which is the groundwork we set out in pricing and packaging tiers.

What Kills Partner Programmes Early?

Three things, in our experience of watching them. The first is no deal registration. If two partners and your own sales team can all claim the same customer, you will have a fight, and you will lose at least one of the three.

The second is unclear boundaries on services. If partners implement your product and you also implement your product, you are competing with the people you asked to sell for you. Decide which work is yours and publish it.

The third, and the most common, is launching with twenty partners instead of three. Twenty partners is a support load with no revenue attached. Three good partners who each closed something is a programme. We made the same argument about channel sequencing in partner marketing at an early SaaS company.

How Do You Choose the Right First Partners?

Pick partners who already have your customer and cannot serve them without something like you. That overlap is the whole thing. A consultancy whose clients repeatedly ask for a capability you provide is a good partner. A consultancy that merely admires your product is a contact.

Our practical filter is three questions. Do they already have a relationship with your ideal customer profile? Do they make money from work adjacent to your product, so that recommending you increases their revenue rather than replacing it? And have they closed anything for anyone else, so you know they can sell?

The third question eliminates most candidates and saves the most time. Plenty of firms sign partner agreements and never sell anything, because selling someone else's software is genuinely hard and nobody in the firm owns it.

What Should the Agreement Actually Say?

Five things, in plain language. What triggers a commission and on what revenue. For how long, with the AWS and HubSpot programmes showing that a fixed multi-year term is normal rather than perpetual. How deals get registered and how conflicts get resolved. What each side may say publicly, including whether the partner may claim certification. And how either side exits.

Write the exit clause first. A partner programme with no clean way out accumulates inactive partners who still appear in your directory, still claim your badge, and occasionally still sell your product badly.

Keep the whole document short enough that a partner's founder will actually read it. A twelve page agreement gets forwarded to a lawyer and stalls for six weeks.

How Do You Know If It Is Working?

One metric, chosen before launch: partner-sourced revenue, meaning deals where a partner brought you a customer you did not have. Not partner-influenced. Not registered deals. Not partner count.

Partner count is the vanity metric that kills these programmes, because it is the easiest number to grow and the least connected to outcome. A programme with 40 partners and two that sell is a programme with two partners and 38 support tickets.

Give it a fair window. A referral partner needs to encounter a suitable client, which depends on their deal flow, not yours. Two quarters is a reasonable minimum before you conclude anything, and that patience is the same discipline we argued for in choosing a first go-to-market channel.

Should You Build a Marketplace or Directory?

Only once you have partners worth listing. A directory with four entries advertises that nobody uses your product. A directory with forty is a genuine distribution asset, because prospects search it and partners fight for placement in it.

The sequence matters. Recruit partners first, get several to close real deals, then build the directory as a reward for the ones who did. Building it first, in the hope that it attracts partners, inverts the incentive and produces a page of logos belonging to firms with no revenue from you.

When you do build it, treat it as a real part of the site rather than an afterthought page. Partner directories rank well, they get used by buyers as a shortlist, and they are one of the few pages where your partners will actively link to you. That makes it a genuine channel rather than a formality, in the same way the service-plus-software combinations we described in packaging services with software become a channel rather than a line item.

What Would We Do in the First Ninety Days?

Write a one-page referral agreement with a fixed commission and term. Approach five firms who already serve your ideal customer and earn money from adjacent work. Aim to sign three, and expect one to produce anything. Track only partner-sourced revenue. Review at ninety days and again at one hundred and eighty, and decide whether to invest more or stop.

That is deliberately unambitious, because partner programmes fail from over-commitment far more often than from under-commitment. The expensive version, with tiers, portals, certification and a directory, is a real product that needs a real owner. Very few companies should build it before a simple referral arrangement has proven that partners can sell their product at all.

If you want the site, directory and partner-facing content built so a programme like this can actually run, that is the kind of thing we do. You can reach us at phoenix.studio.