9 Lessons Learned Launching Integrations With 750+ Apps

After working with hundreds of SaaS APIs, we’ve noticed plenty of best practices–and a handful of anti-patterns–of a technical nature. It’s rarely the status codes or authentication mechanisms that make the difference with these integrations. There’s a human element that is much more important, as we’ve noticed expanding Zapier from just a handful of apps five years ago to over 750 today, helping 1.5 million users build processes that automate time-consuming tasks and boost productivity.

This post is based on a talk I gave at the Small Business Web Summit

So, let's rewind the clock a bit. When Zapier was younger we were approached by a big company, one you'd definitely recognize, to build an integration with one of their big products. We got excited and jumped right into building even though the API was a bit strange and not super well documented. I think we all know where this story ends. We turned the integration on, the partner didn't help promote it at all, and eventually only 4 people out of 400 that tried the integration successfully set it up. That was over 100 engineering hours wasted. That sucked.

Contrast that with our integration with Typeform, where both of our teams talked upfront about what a great integration would look like. Ultimately, we were able to build this cool experience where Typeform users can set up Zaps right within Typeform. This integration helps Zapier get dozens of new users every day, while also helping Typeform reduce churn because users with Zapier integrations enabled churn at half the rate of Typeform's other users.

So how can we do more integrations like the Zapier Typeform integration? I’ve identified nine lessons learned.

1. Start Small

Everyone tends to want to build this magical integration that solves everything at once, something that is simple to use and makes billing and login perfect for users. But this almost always adds complications which delay launch and rarely improve adoption. Sometimes, "magical" integrations even hinder adoption because it's too hard to manage the complexity.

Instead, start small. Once you have something small in place you can better listen to user feedback and improve on the areas your users actually care about rather than the problems you invented in your head.

2. Define use cases

Integration is a loaded term. What does the user experience actually look like for an integration? Does it mean a widget in the UI? Does it mean pushing data automatically between two apps? Does it mean two way sync where you maintain state? It is really important to define what you mean by integration with both your users and your partners before you set about building one.

Defining use cases will help you stay focused even when a big company emails and gets you excited about partnering. If they start talking about features that don't match what your customers want, you should rightfully know to be skeptical.

So talk to your users. Understand the use cases they want solved with an integration. That way you'll be able to better design your API and integrations to match up with what users actually want.

3. Define success

What do you actually want out of this integration? Are you trying to attract a new audience to your product? Are you trying to make your existing customers happier and do a better job at retaining them? This matters and affects the type of integrations and partnerships you should seek.

People	Action	Conversion Rate
100,000	Email	10%
10,000	Click to Landing Page	30%
3,000	Convert to Trail	20%
600	Click to Paid	20%
120	New Customers	N/A

If your goal is to gain new users, understand the mechanics of how that will happen. For example, consider a potential partner that says they'll email their 100,000 customers about your new integration but do nothing else to promote the integration. If only 10% of those customers open the email, you're down to 10,000 customers. Then if only 30% click through to the landing page you're down to 3,000 customers. Then if only 20% convert to a trial you're down to 600 users. Then if only 20% convert to paid you are down to 120 new customers from the partnership.

Depending on how large your company is and how much your customers pay for your product, this could either be a great deal or a horrible deal. But if you don't define this up front you won't know.

4. Quality Over Quantity

You can't fake a good integration. Quality > quantity. Make sure everyone agrees to standards.

Don't chase logos. Everyone wants that big list of logos to show off how they have tons of integrations with the best companies. But if you do that, you'll likely bring on half-baked integrations that don't solve a market need.

Instead, set a quality bar or a style guide internally that defines what an acceptable bar is for integrations you support. If you measure NPS, you can track the NPS for customers who use the integration against those that don't.

A style guide helps you make fast decisions that maintain quality, too. If standards are already agreed on, you won't have to rehash the same discussion for every potential integration. Instead can confidently say "yes" or "no" to opportunities based on your rubric and not on other fuzzy metrics like "hot startup" or "great logo".

5. Bake In Discovery

The most common mistake I see our partners make is to start building an integration with no plan in place for how users will find this integration. There's nothing more depressing to see a couple talented engineers spend a few weeks to a month working on an integration, only to have no one ever use it because two partners didn't think about how users would discover the integration.

It's far better to talk upfront about what leads to a successful integration. In our experience, things like press and blog posts are nice but don't drive long term engagement with an integration. Much better is a high profile integrations directory that's marketed to all new users.

Include integrations in your onboarding education material. Include them directly in key spots in your app. And include integrations in your help docs where users are directly seeking out answers to questions. That's what will make your integration a success with users.

6. Commit to Support

Understand the cost and commit to supporting and maintaining the integration before working on it. If you're going to build integrations, be sure you can commit to supporting the users that will use the integration. There's an implicit promise when you launch this to your users that you'll support it and not break it.

So make sure you train your sales and support staff on the integration. Make sure your integration partner can commit to the same, too. Too often I see partners provide an API, while never wanting to help their customers out with issues that arise from their API breaking in unexpected ways. If your core product breaks, you would help your users out. If your API breaks, that affects your core users too and you should commit to helping them.

One of the most impactful decisions we made early on was that all our engineers would do support shifts. As a result, all our integrations get great support and our engineers go above and beyond to build tooling that helps us support customers because we can better see how APIs are working and breaking.

7. Prioritize Your Options

You only have so much ammo. You can't build for everyone.

Action	Option 1	Option 2
Directory	Yes	No
In app	Yes	No
Help docs	No	No
Customer email	25,000	300,000
Dev time	1 week	4 weeks
Blog post	No	Yes
User Type	Enterprise	SMB

All of us are resource constrained. We all wish we had more engineers on our staff. And integrations aren't the only thing you can spend time on; you often have features that you can work on internally that will affect more customers.

So make sure to prioritize all the integrations you can build. The book Traction by Gabriel Weinberg has a great exercise they recommend for marketers testing new channels that I find works really well for figuring out which integrations you should spend time on. When you go out and assess potential partnerships, ask your partners what they can commit to and make educated guesses when you can't get firm answers. Then you can figure out which ones will be successful and which ones won't.

Remember: It's rare to have any given direct integration get more than a few percent of users using it. So you want to make sure you are getting the most bang for your buck when you spend time on an integration.

8. Transition Into a Platform

As you grow, you'll get approached more and more about building integrations. If you find demand is high enough you can sometimes offload this work to your platform. This usually starts with a public API that partners can adopt in a self-serve fashion, which helps you scale out integrations faster than you can build on your own.

But remember: If you do have this API, you need to commit to supporting apps built on the API. Make sure you are transparent with users about what you can support. You don't want your users to get burnt by low quality apps; that will affect your users' perception of your company.

For example, we noticed more partners wanting to build experiences like the Typeform example. So we built out the Partner API that allows people to more easily showcase the best Zaps for their Apps in strategic places in their app. This has allowed partners like Unbounce and MeisterTask to take advantage of better integrations in their apps.

9. Win-win for Partners

As you grow, you'll get pulled more into "strategic partnerships"—which I've found is often code for asking you to do a boatload of speculative work that may or may not work out.

Make sure to lookout for yourself in your integrations, and don't get pressured into doing something that doesn't support your business.

At some point you'll grow and have more leverage over partners, too. So remember not to pressure your partners to do things that are bad for their business but might be better for you. For example, don't ask partners to make price cuts on their product that aren't profitable for them. The best partnerships are ones where both of you work together to solve a problem for customers and both of you get paid.

---

The Zapier Developer Platform allows anyone to connect their app to Zapier for free. If you make your app public to all 1.5 million Zapier users, we offer co-marketing opportunities to get the most of the new integration. Our partner kit has the full story to give you a head start creating successful integrations.