How a partnership grows: five stages from discovery to scale

In the first post in this series, I explained how I think about partnerships: start with your customer’s ecosystem, make sure every partner gets a real win and agree on who owns which lane.

That’s the strategy, and in part 2 I covered how to choose which partners to pursue first. This post is about what happens next: the day-to-day work of turning those partners into partnerships that last.

At Thumbtack, I broke partnership development into five stages: discover, experiment, land, expand and grow. Every partner you’re talking to sits in one of them, and each stage asks something different of you. Knowing which stage a partner is in tells you what to do next and, just as important, what not to do yet.

Discover: start with your customer, not the partner

Discovery doesn’t start with a list of companies you’d like to work with. It starts with your customer and the jobs to be done.

Research those jobs, then group them into categories. Build real expertise in each category before you pitch anyone. This step isn’t optional: you need to understand how the work actually happens and who’s already doing it. How to build that expertise deserves a post of its own. Then pick the companies that best represent each category and start talking to a handful of them.

Those first conversations are where partnerships and sales part ways. You’re not there to close. You’re there to find out what the partner needs and what’s blocking them from working with you today. Write down what you’d have to build to unblock them. That gap list becomes your product input for everything that follows.

For us, the customer is the warranty company, and its jobs go beyond the claim itself. There’s the work around every claim: taking the first notice of loss, deciding what a job needs before a truck rolls, authorizing the repair, getting the right part and paying the pro. And there’s the work of running the operation: evaluating new vendors, implementing them and managing risk. Other companies already touch every one of those jobs.

Experiment: test it before anyone builds anything

Before anyone signs anything, build the relationship and test whether the partnership is real.

The best tests are small, and they’re run by operations rather than engineering. AI hasn’t changed that. The only difference is that prototypes and high-fidelity mockups are now much faster to make. Do the work with people and process first, so you learn whether the partnership creates value before you spend engineering resources on it. If a pilot can’t show a win for both sides, it costs you a few weeks, not a quarter of the engineering budget.

Here are a couple of examples from my firsthand experience:

Real estate partnerships had been essential to Setter’s growth. After Thumbtack acquired Setter, those partnerships became much harder to build, so I started looking at MLS marketplaces. Instead of starting with the biggest one, I went to a smaller one that was eager to experiment. We started with the simplest test we could run: emailing people who had recently bought or sold a home and pointing them to a painted-door page, a page for a service that doesn’t exist yet, to see who would raise their hand.

A page like that takes minutes to set up. Email campaigns rarely produce exciting results, but this one gave us a way to start working together. More experiments followed. In the end, customer salience and partner affinity, the two filters from part 2, weren’t high enough. We let it go before either side had spent any engineering time.

At Sears, the result went the other way. As CPO, I had engineers available to build a video diagnostic experience. Instead, I set up a free 60-day pilot with a video support provider. I messaged customers who had recently scheduled a repair with a link to book a video call, and put a few of our techs on a round robin to take those calls. Two weeks in, the results were clear enough that we moved into contract negotiations and integration. I had the data before I spent the engineering dollars.

This is also the stage to decide how you’ll build for the partnership. You can test, learn and build as you go. Or you can bet on a category and build in parallel with the sales cycle, so the product is ready when the deal closes. Both can work, but they carry different risks. The right choice depends on how confident you are in the category, what you can afford to build, and what else you’d have to say no to this quarter or this year to make room for it. Whichever you choose, your product and partnerships teams need to agree on it.

Land: one partner, done well

In the land stage, the goal is narrow on purpose: get one partner in a category live and make it work well.

Start by proving the core set of features it takes to unlock that partnership. That doesn’t mean every feature the category might eventually want, only the ones this partner needs to say yes. Product and partnerships have to agree on what that minimum version is, and on what it takes to earn the right to play.

Once it’s live, grow adoption together. That’s go-to-market work: shared messaging, introductions, events and marketing that gets your customers to actually turn the integration on.

When we sit down with a partner to plan that work, we try to leave with five things agreed:

• How we describe ourselves together

• A roadmap for the next few months

• Who owns which lane

• A plan for the customers we share

• An operating rhythm: how often we meet and what we review

Expand: go deeper with the partners you have

Once a partnership is working, the next job is to go deeper.

That means building tighter integrations and putting the partnership in front of more of your customers. It means tracking what each partner actually drives, so you know where to invest your time. And it means continuing to build the features that make the partnership harder to replace, because those features are your moat.

Grow: turn it into a playbook

Growth is where a partnership becomes a channel.

Take what worked with your first partner in a category and turn it into a playbook: the pitch, the features, the go-to-market plan and the operating rhythm. Then repeat it with the other partners in that category. The playbook is what lets the second partner move faster than the first, and it’s why landing one partner well matters so much.

After that, look beyond the category, because what you built for partners in one category can sometimes be offered in another.

What moves a partner to the next stage

A partner moves forward when the partnership keeps paying off for them. In part 1, I covered the three things every partner needs from you: net new customers, better retention or incremental revenue. At every stage, check that you’re still delivering at least one.

It works in both directions, too. When your customers use a partner more, that partner does better, and they become more likely to bring you to their own customers.

When a partner hesitates, the objection tends to fall into one of three buckets:

• Reliability. Will this work, every time, for their customers?

• Efficiency. Will it make their operation faster or cheaper, or will it add work?

• Experience. Will their customers have a better experience, or a worse one?

Those are the cornerstones of what any partner needs. Have an answer for all three before you ask a partner to move forward.

Putting it together

That’s the series: how I think about partnerships, how to choose where to start and how a partnership grows from there.

If you’re a supplier, technology company, professional services firm, service provider or data provider working with warranty companies, this is the path we’d walk together: start with the warranty company’s work, test something small, agree on lanes and how we describe ourselves, then go deeper. If that sounds like a fit, let’s talk.