Zapier: Solving the Cold Start Problem and Building a Flywheel
Funnels are out; flywheels are in. Why? Better yet, what does a good flywheel look like? Let's look at Zapier. Plus, noodle on what they could build next.
👋 Happy Sunday, friends!
It feels somewhat strange to write about Zapier given their recent layoffs. I was fortunate to work at Zapier for a year, and everyone I worked with was incredibly talented, smart, and kind. My heart goes out to everyone impacted.
If you’re hiring, I’d encourage you to reach out to alumni@zapier.com. So many stellar folks are available now.
Cheers,
Jeremey
Thanks for reading another edition of Strategy Meets World – a weekly deep dive into building products that serve customers and the business. If you’re new, check out the intro post and subscribe to get future issues in your inbox.
📬 Either way, email me at jeremeylduvall@gmail.com and tell me what you’re building. I’d love to hear it.
What Is a Flywheel? Let’s Use Zapier as a Case Study
If you wanted to move information between two apps - say your email and Google Sheets - back in 2011, you had to do it all manually. Meaning, a ton of inefficient copying and pasting.
Enter Zapier. The automation tool for your business.
Zapier connects any one tool (Gmail) to over 2,000 others (Airtable, Buffer, etc). No more spending time copying and pasting data from this form over here. Zapier makes it all just work.
Today, we’re digging into the Zapier flywheel. In doing that, we’ll answer two specific questions about Zapier’s growth:
How did Zapier solve the cold start problem that plagues two-sided marketplaces?
What are flywheels? Why are they important? What does Zapier’s flywheel look like?
Finally, we’ll end by putting on our CPO hat and brainstorming what Zapier could build next to further their flywheel. Let’s dive in.
Solving the Cold Start
When getting off the ground, Zapier faced a challenge that plagues all two-sided marketplaces - the cold start problem.
Zapier has two distinct customers - developers that build integrations on Zapier and the Zapier customers that build “Zaps” to automate tasks. The a classic chicken or the egg. You need users to convince developers to build integrations on your platform. Yet, you also need integrations to convince users to sign up. AKA, a cold start.
Here, we get a sneak peek at Zapier’s first flywheel.
For their initial integrations, the co-founders perused forum threads for popular apps looking for integration requests. For example, a customer in the Mailchimp forums might request an integration with Gmail. For Mailchimp, that likely falls far down the priority list. For Zapier, this is exactly the kind of use case they were looking for. This gave the founders validation for what they were building.
Then, they got to work. Since the co-founders all had web development experience themselves, they built the first few integrations themselves - all targeted towards popular apps at the time, often validated through actual use cases.
Once the integration was live, they’d follow-up directly in the forum. Now, they had targeted inbound leads coming to the Zapier website with a very specific problem. That’s a great lead generator, but they didn’t stop there.
Once the integration was live, they’d build a public page on the Zapier site that touted how to connect Mailchimp with other apps. This page was picked up by Google. Now, anyone that searched for “Mailchimp integration” was likely to stumble upon Zapier. Eventually, they’d reach out to brands and co-promote the integration.
Each integration then pulled in more leads for the product and attracted more developers for new integrations. The makings of a flywheel…
Stepping Back and Breaking Down the Flywheel Concept
I first bumped into the term “flywheel” when reading Working Backwards, a deep dive into the world of Amazon. Paraphrasing their definition:
[A flywheel is] a model of how a set of controllable input metrics drives a single key output metric. Flywheels are closed-loop systems, as you inject energy into any one element, or all of them, the flywheel spins faster.
Three distinct elements jump out:
Controllable input metrics - You have control over all aspects of a flywheel. Through directed effort, you can move them one way or another.
Single key output metric - Flywheels have to be grounded in the business model. Otherwise, they won’t have much of an impact.
Closed-loop system - Each step leads directly to the other and propels the flywheel to spin faster and faster, effectively building a competitive advantage for the business. Loops are self-reinforcing.
We can see these distinct pieces when we look at Amazon’s flywheel. Here’s the version Jeff Bezos allegedly wrote on a napkin.
Starting at the top, you can read this as:
An unrivaled selection of items
Feeds into a great customer experience while shopping
Which generates more traffic through repeat visits, word of mouth referrals, higher search rankings, more inbound links, etc
This traffic brings more sellers looking for sales
These sellers then increase the selection of items…
And the loop cycles again and again, all powering the single, core metric in the middle - growth.
But, What About Funnels?
Previously, when asked about how to grow a business, leaders were more likely to draw a top-down funnel than a circle. “Our product grows by attracting leads, getting them to try our product, converting them, and then retaining them.” The infamous AARRR funnel, which we’ve talked about previously. Customers move from the top down.
In this model, a company grows by improving one segment of the funnel - acquiring more leads, converting more to paid, decreasing churn, etc.
Now, the answer to “how does your product grow?” is more likely to be circular (at least for the most successful companies).
Let’s look at Pinterest’s growth loop, for example. In this example, users sign up for Pinterest. They’re served relevant content that encourages them to add new or re-pin existing content. This then feeds into Pinterest’s search engine ranking, boosting certain pages higher in results. Prospective users find the content in search results and sign up for Pinterest…the cycle goes on.
Why are loops so powerful as opposed to flywheels? The Reforge team offers three reasons, and I’ll add a fourth.
Funnels create silos; Loops create synergy.
It’s easy to view yourself as a member of the acquisition team that only works on top-of-funnel growth or the revenue team that works on pricing and packaging. But, this quickly create silos where teams don’t talk with one another or share information. Or, if we go off and develop our strategies individually, there’s a big chance we’ll conflict.
Funnels can create perverse incentives.
If we’re trying to optimize the funnel, I can optimize my metric and tank yours. The simplest example is someone on acquisition bringing in lower quality leads that will never convert to paying accounts. Acquisition goes up; revenue doesn’t change.
Funnels are only top-down.
Generally, funnels operate in one direction. You pour more in the top to get more out of the bottom. By contrast, loops compound. Since they’re a closed system, pouring energy into one step subsequently powers the next.
Funnels are business-centric; Loops are customer-centric.
When optimizing the funnel, it’s all about acquiring more leads, increasing conversion, decreasing churn, etc. All important elements for the business but not relevant for the customer.
By contrast, we can orient loops around the core transaction - the primary way users get value from your product or service.
Elements of a Successful Flywheel
Once you start to think in terms of flywheels over funnels, you’ll start to see them everywhere. Here are five pieces I try to keep in mind when breaking down a product or service.
What’s the single action that drives value for your customers? Is it publishing posts, completing tasks, swiping their card, etc. When you boil most businesses down, there’s a single action that represents value.
What powers your business model? This should form the basis for the entire loop. For example, if your pricing model is based on seats (e.g. Notion) versus usage (e.g. Zapier), this should be a consideration.
How do your customers discover your product? At the end of the day, it’s about growth. Is your business driven by top-down, enterprise-y sales (e.g. Oracle) or a bottom-up cycle (e.g. Zoom)?
How can we make each rotation easier than the previous one? We should optimize the flywheel to gain momentum over time. This requires us to optimize each step and ensure data flows smoothly from one piece to the next.
Finally, avoid the temptation to represent your business with a ton of loops. It’s usually better to focus on a single one. From the Reforge team:
You'll be tempted to draw a ton of loops for your product, but what that typically means is that you just have a ton of low-powered loops that aren't sustainable. The fastest growing products are typically powered by 1- 2 major loops that transition over time.
How Has Zapier’s Flywheel Evolved?
Above, we saw how Zapier solved the cold start problem and built their first growth loop. This flywheel still exists today. If you search for “Google sheets automation,” you’ll still find a sponsored search result for Zapier at the top. As expected though, the flywheel has matured quite a bit.
Let’s revisit our points above to set the stage.
Customers derive value through completed Zaps. Think of a Zap as a specific workflow. For example, “every time someone fills out a Google Form, alert me in Slack.” Each time that runs successfully, I derive some value from Zapier as a product.
Zapier’s business model is driven by two things:
Revenue is primarily based on usage. If you look at their pricing page, you’ll see a fundamental driver is number of completed tasks (setting aside the Enterprise plans). A single Zap can have multiple tasks. For example, maybe I want my Google Form to alert me in Slack and then email the customer that completed the form. That’s two tasks (alert me, email them). More tasks = a higher plan = more revenue for Zapier.
Zapier is a two-sided marketplace. Customers use Zapier and developers build integrations. More customers encourage more developers. Better app selection brings in more customers.
Customers discover Zapier through 1) colleagues sharing workflows 2) organic search (“connect XYZ with ABC”) and 3) co-marketing with other apps and brands.
Our primitives can help direct our efforts. Each time I build a Zap,
Let’s plug it all in.
Customers create new Zaps to automate a core piece of their work.
Those Zaps run successfully and generate tasks.
This generates an endless combination of recipes and workflows that can be shared directly by users or bundled by Zapier and marketed publicly.
This flurry of activity draws more developers to the platform to build integrations.
This increases the supply of available apps, which draws in more perspective users.
More apps and potential integrations encourages users to create more Zaps…
This fits our definition of a classic growth loop or flywheel - as you inject more energy into one step, it makes the whole thing spin faster. I’m also gaining tiny bits of data across each rotation I can plug back into the flywheel to make it spin faster (better targeting, faster onboarding, etc).
What Could Zapier Build Next?
Now we have a new lens for evaluating new projects. “How does this help our flywheel spin faster?” With that framing in mind, let’s put our CPO hat on and dream up what Zapier could build next.
“Next Step” Smart Suggestions
One of the hardest aspects of creating a new zap is figuring out where to start. If you have a specific use case in mind, it’s relatively straightforward once you learn the tool. If you’re exploring though, it’s tougher to wrap your head around what’s possible.
This directly feeds into the “customers create new Zaps” segment of our loop. Without this step, the whole loop dies.
Zapier has put some significant effort in here allowing customers to share workflows with one another and generating starter templates for specific job functions.
Here’s an idea though - once I select the first step of my Zap (aka “trigger step”), generate smart suggestions on what to do next based on 1) what you know about me 2) what other users have done. No doubt Zapier has a ton of valuable insight into what successful Zaps look like. Use that data to guide me.
There’s no reason why this block to select an app is the same every time you start a Zap. I probably don’t want to connect Gmail to GitHub. Don’t even suggest it.
Community and Context-Driven Discovery
A huge part of Zapier’s flywheel is how to use the core primitive - Zaps customers build - to attract more users and encourage usage.
Zapier already does this in several ways through their blog and Zap sharing, but I think there’s opportunity to lean more into this area. Two examples with different directions.
ChatGPT allows you to share your conversations for anyone to view.
Notion allows users to generate (and even sell) Notion templates to other users.
Both are effective ways to get new users started and allow your power users to show off.
For Zapier, this look like a public version of your Zapier profile where you can effectively “expose” all of your Zaps for public consumption (and search indexing!) instead of sharing them individually one-by-one. I’ve seen many tweets where Zapier customers share stories of running their entire advertising budget off of Zapier; I’ve often wished I could peek at how they have it setup.
If you found something valuable in this edition, it would mean the world to me if you shared it with a friend.
Better yet, reach out and tell me what you enjoyed. I’m always here at jeremeylduvall@gmail.com.
See you next week,