How do you build a partnership that drives growth?
Most early-stage partnerships don't drive growth — they delay it. Paul Graham's observation that partnerships 'especially don't work as a way to get growth started' isn't a reason to avoid them forever, but it is a reason to be ruthlessly honest about timing, incentives, and what you're actually getting. The partnerships that do work share a common structure: they amplify something already moving, not substitute for traction you haven't earned yet.
Why Most Early Partnerships Fail
The most common version of the bad partnership story goes like this: a startup signs an MOU or a 'strategic agreement' with a large company, spends three to six months navigating legal review and internal champions, and emerges with nothing but a PDF and a depleted runway. The failure mode isn't bad luck — it's structural. Big companies move slowly because slowness protects them from risk. Startups move fast because speed is their only competitive edge. These incentives are fundamentally opposed.
Paul Graham's point about inexperienced founders is worth taking seriously: the belief that a partnership with a big company will be the 'big break' is a specific cognitive trap, not just a rookie mistake. It's attractive because it feels like leverage — one deal that replaces dozens of direct sales. But leverage without a fulcrum doesn't work. If your product hasn't yet proven it can retain and delight users on its own terms, a distribution partnership just exposes that weakness at scale, faster.
The honest test is this: if the partnership disappeared tomorrow, would you still grow? If the answer is no, the partnership isn't driving your growth — it's propping up a product that hasn't found its own legs. That's a product problem, not a partnerships problem. Fix the product first.
What Makes a Partnership Worth Pursuing
A growth-driving partnership has three properties: it reaches users you genuinely cannot reach efficiently on your own, it requires your partner to do something that's already in their interest, and it's structured so that both sides feel pain if it doesn't perform. The third point is almost always missing. Partnerships that survive are ones where your partner's champion has staked something — budget, a quarterly number, a public commitment — on the outcome.
The timing question matters as much as the structure. Partnerships that work tend to come after you've already figured out what makes users stick. Facebook's expansion strategy — starting at Harvard, then specific colleges, then universities broadly — is a useful model not because it was a partnership strategy per se, but because it shows the value of proving deep resonance in a narrow market before expanding. A partnership should be a mechanism for replicating that proven resonance into adjacent territory, not a bet that your unproven product will resonate somewhere new.
The best early partnerships are often informal and asymmetric. A single large customer who gets unusually hands-on service — almost consultant-level attention — teaches you more about your product's real value than any formal agreement. Graham describes this as acting like a consultant for one user, using their specific needs as the mold for your product. When that relationship produces a reference case or a referral, it functions like a partnership without the legal overhead.
How to Structure a Partnership So It Actually Performs
Structure determines behavior. A partnership that isn't tied to a measurable outcome will drift toward the lowest-effort interpretation for both sides — usually a logo on a website and occasional co-marketing emails. If you're the smaller party, you're almost always the one who loses in that scenario.
Before signing anything, define the specific user behavior you're trying to change (signups, activations, revenue, retention) and agree on a time-bound test. Six to eight weeks is usually enough to see a directional signal. Build in a checkpoint that gives both parties permission to walk away with no reputational cost if the numbers aren't there. This feels aggressive to put on paper, but partners who are genuinely aligned on outcomes will agree to it; partners who resist it are signaling that they want the association without the accountability.
On the execution side, assign a single owner on each side with decision-making authority. The partnership will move at the speed of the least-empowered person involved. If your counterpart needs three approvals to send a co-branded email, budget two extra weeks for every deliverable and decide whether that's a dealbreaker before you start. Your time is your most constrained resource at the early stage, and a partnership that eats 15 hours a week of founder attention for modest returns is destroying value even if the press release looks good.
Growing from Traction You Already Have
The partnerships that produce outsized returns for startups are almost always pulled by existing momentum rather than pushed as a growth strategy. When Stripe began gaining traction, other companies wanted to integrate because the product was already working. The traction created the inbound; the partnerships then accelerated what was already moving. Patrick Collison's description of the shift from pushing a boulder to riding a train captures the difference: a good partnership feels like releasing the brakes on a vehicle with its own engine.
This means the prerequisite for a growth-driving partnership is a cohort of genuinely happy users who will evangelize unprompted. Garry Tan's point about founders imitating the flaws of big companies applies directly here: small startups sometimes pursue formal partnerships to appear more established, when the more effective move is to go deep with individual users in a way that no large company can replicate. That depth creates the word-of-mouth that makes a later partnership valuable — because your partner is attaching their name to something their audience already wants.
The practical implication: before pitching a partnership, ask whether your existing users would describe your product to a stranger without prompting. If not, the partnership is premature. If yes, you have something worth distributing, and a partner now gets real value from being associated with you — which shifts the negotiating dynamic in your favor.
“Partnerships too usually don't work. They especially don't work as a way to get growth started.”
— Paul Graham, source
The one thing to do
Before pursuing any partnership, confirm your existing users would recommend your product unprompted — that's the only condition under which a partner can actually accelerate your growth.
Frequently asked questions
When is the right time to pursue a formal partnership?
After you have a cohort of retained, happy users who can serve as social proof for the partner. A partnership before that point is asking a third party to validate something you haven't validated yourself — and they usually won't.
How do you get a large company to take a small startup seriously as a partner?
Find the internal champion whose specific quarterly goal your product can advance, and frame everything in terms of their number, not your product. The partnership becomes real when it's tied to something they're already accountable for.
What's the difference between a channel partnership and a distribution trap?
A channel partnership extends reach you've already proven in a narrower market. A distribution trap is when you use a partner's audience to test whether your product works at all — you get one shot, and failure poisons the relationship permanently.
How much founder time should a partnership take?
If a partnership requires more than a few hours a week of your personal attention after the first 30 days of setup, it's either understaffed on your end or the partner isn't genuinely engaged. Build an operational owner into the structure from day one.
Sources
- Do Things that Don't Scale — Paul Graham
- Billionaires Build — Paul Graham
- How to Raise Money — Paul Graham