PLG to PLS: A Case Study
A deep-dive on how Patreon's creator acquisition motion has changed over time
Hi there, it’s Adam. 🤗 Welcome to my weekly newsletter. I started this newsletter to provide a no-bullshit, guided approach to solving some of the hardest problems for people and companies. That includes Growth, Product, company building and parenting while working. Subscribe and never miss an issue. I’ve got a podcast on fatherhood and startups - check out Startup Dad here. Questions? Ask them here.
Patreon is one of the most impressive creator economy companies of the last decade. Currently valued at several billion dollars, with more than $1.5 billion in annual recurring GMV and at least $150M in annual recurring revenue it has been an impressive company since the start in 2013. I joined ~3 years after it was founded and helped lead it to 50+ months of consecutive MoM growth in GMV and revenue.
I’ve written a bit about how rebranding the company helped with our growth here and here. I’ve also written about the Growth strategy which has quickly become one of my most popular posts of all time and one of the most bookmarked Reforge artifacts. My friends and professional colleagues have written a lot about PLG and PLS as well, people like Leah Tharin, Elena Verna, Kyle Poyar, and Ben Williams.
Today’s newsletter is about how Patreon’s creator acquisition engine evolved over time: from high velocity sales outreach, to product-led growth, to a combination of product-led-sales and traditional, enterprise sales.
Let’s dive in!
The Three Phases of Patreon’s Acquisition Engine
For as long as Patreon has been around this has been the core of its growth engine: creator signs up and launches on Patreon, creator shares their membership with their fans to get patrons, creator makes money, other creators see this and sign up thus starting the cycle over again.
But something has to get this flywheel started and get it spinning. In the very early days that was Jack’s direct outreach through his own creator network. He famously got rejected hundreds of times in his outreach to other creators. He refined his pitch over and over and over again (side note: this is why founders need to get out and sell). And eventually, after paying himself a $1 salary and making all his money off of Patreon, other creators started to take notice.
But that wasn’t really the first phase of our acquisition because Jack sending emails out to his creator network doesn’t scale very well.
Phase 1: Lighthouse Creators + High Velocity Outreach
Our earliest scaled creator acquisition came from a combination of a category-based lighthouse creator strategy and high velocity outreach.
We had two small sales teams of approximately 3-7 people organized first by creator potential (or what we thought their potential was) and divided into two groups: big, recognizable household names and recognizable creators who weren’t necessarily household names (yet). More on that second group in a second.
In the “big, recognizable” bucket you had creators like Snoop Dogg and Youtubers like PewDiePie. They had managers, teams, and ran their creative empire as a business with them as the face. It turned out that these creators weren’t all that great for Patreon so we didn’t keep focusing there, but the motion looked a lot like enterprise sales given the layers of people you had to pitch to get to the decision maker.
Within the second group “recognizable, but not household names” we had creators who were well known in their category. People like Phil DeFranco, Amanda Palmer, Kinda Funny, podcasters Chapo Traphouse, and web comic Zach Weinersmith.
These were still, at a category level, what we called “lighthouse creators.” The term lighthouse customer, or early adopter, comes from a 1962 book called Diffusion of Innovation. In our case these were recognizable names amongst creators within that same category. Getting them on the platform could pave the way for other creators in that same category to launch as their mere presence provided validation and trust for the platform.
So this is where we leaned in early on. We had a handful of people identifying leads from Social Blade based on the creator’s data and analytics. We would then craft highly customized, individual, one-on-one emails or DMs to those creators. High velocity was a bit of a misnomer as this process was very methodical and slow. But it was still faster than trying to work with the Snoop Doggs of the world and their layers of management so by comparison it was high(er) velocity.
This strategy worked well and others have talked about using it. One of the reasons it worked so well was leveraging the lighthouse creators.
Here’s an example in the web comics category:
Unless you’re The Oatmeal it’s hard to monetize an online comic. However, they do have highly devoted fans and they produce regular, serialized content. These factors mean that webcomics are uniquely well suited for Patreon. But none of them would launch because they wanted to see someone else doing it successfully.
Enter Zach Weinersmith, creator of the Saturday Morning Breakfast Cereal comic. After a painstaking process we were eventually able to convince him to join the platform and once we did that it was open season. We immediately got in touch with hundreds of other webcomics who had previously said “no” and leveraged Zach’s notoriety to help launch many of them.
This was why the lighthouse + high velocity strategy worked so well to get the flywheel going. But Phase 2 is where it really accelerated.
Phase 2: PLG, Earnings Amount + Self-serve Onboarding
For several years we maintained the lighthouse + high velocity strategy, but it gradually contributed less to our overall acquisition. It was still important to have recognizable names launch on the platform but more of them were appearing as “walk up” creators and our initial sales outreach was no longer the first time a creator had heard about Patreon.
We were entering the second phase of our acquisition engine—the PLG phase. As referenced above in the core growth loop Patreon had always had the foundations of a strong product-led growth motion. User generated content and word of mouth coupled with self-service onboarding and a “free-to-get-started” pricing model. But 2016-2017 is when we really started to pay attention to all of the elements of this growth model.
We knew that the creator page, the first page that a potential patron saw, was also a “landing page” for selling Patreon to prospective creators. And our theory was that the more we fine-tuned that page and increased traffic to it the more powerful it would be. If we were successful we would end up with hundreds of thousands of individual, user created landing pages advertising what you could offer and how much you could make on Patreon. Spinning our core growth loop even faster than Phase 1 could ever have.
In the first few years I was at Patreon we did the following to lean into product-led growth. This is a non-exhaustive list:
Analyzed every element of the best performing creator pages and then put a team on onboarding who ran dozens upon dozens of experiments on this experience.
Overhauled our SEO strategy to remove a prevalence of spam in the index, cross-link between creator pages, and improve all of the elements of a creator’s page.
Optimized our purchase flow for fans to drive more money to creators.
Iterated on creator page sharing.
Built out dashboards and alerts to detect fast-growing, newly-launched creators.
Built canonical pages that organized recognizable creators by category.
Built automated creator-to-creator recommendations (the successful precursor to Substack’s recommendations feature).
Engineered the world’s most shareable email.
And of course, re-branded the company.
As successful as this phase was, helping us scale to hundreds of millions of dollars of GMV annually, we still knew that some creators had questions and would benefit from a personal, guided approach. We also had our own nagging question: could human intervention help successful creators be even more successful?
Phase 3: PLG and PLS
In early 2019 we decided to try to confidently answer the question above. We had record numbers of creators launching on the platform and earning never-before-seen incomes. Still, we felt like we were missing some high potential creators who got stuck in onboarding, launched unsuccessfully, or made suboptimal choices during onboarding.
Of course we continued to tune our onboarding experience to help nudge them in the right direction.
One such example is the $1 tier. This pricing tier was a relic of the “tip jar” era of Patreon but still hugely popular because creators borrowed strategies from other creators when launching. The $1 tier was problematic for several reasons: it was a bad deal financially for creators given transaction fees and membership fees, it devalued their work by setting an artificially low price, and it had worse retention than slightly higher-priced tiers.
We weren’t doing much to discourage it though. We had a grayed-out $1 suggested amount in the tier pricing field when you created a new tier. So we changed that and defaulted it to $5. You can probably guess what happened? The presence of the $1 tier plummeted while other values (especially $5) skyrocketed. Sometimes the smallest changes lead to meaningful outcomes.
Back to high potential creators. We wanted to learn more about whether human intervention would help well qualified creators do better. But how do you do that when you’re signing up thousands of creators per day?
We did two things simultaneously: ran a randomized, controlled experiment to trigger human outreach and attempted to productize the collection of variables that would signal whether we should reach out to a creator.
For the first experiment, we took every creator who signed up and randomly bucketed them into one of two groups:
Group A - no treatment
Group B - human intervention
Over an extended period of time we evaluated the performance of these two groups, including controlling for additional variables like whether someone responded to our intervention or not. When the results came back they showed us that when comparing two, equivalent creators from each group, the one that received human intervention saw a 25% increase in their earnings at the end of the first two months. This was a significant finding for us because, as I’ve mentioned in prior newsletters, a creator’s lifetime value was based on a formula that included their first months of earnings and the velocity at which they got to certain earnings thresholds.
Home run right? Not exactly.
It wasn’t a 100% success because we also engaged with a lot of creators who had no chance of being successful. In order to minimize engagement with those creators we needed to identify them and in order to do that we needed to qualify them.
We had spent years trying to build models that predicted a creator’s earnings potential using hundreds of different variables. Success was determined by both measurable and unmeasurable variables.
Measurable variables were those that signaled engagement with your content. This could be follower count but more predictive were variables like “last 5 video views,” or “number of comments or likes on a post” or “podcast downloads.” You could have a smaller audience that would hang on your every word (watch all your videos, comment, etc.) and that would translate to higher earnings potential than a larger, less engaged audience.
The reason that “last 5” mattered so much was because a lot of Youtube creators built very large followings (subscribers) during the early days of Youtube and they then outgrew their audience. Or, more likely, their early audience outgrew them. Looking at the engagement on recent videos would tell you how popular a creator was today.
At first we tried to get this data quantitatively by asking creators to provide links to their social platforms. That didn’t work because creators would paste other people’s links. Or they’d tell us they were PewDiePie and link to the PewDiePie account but they definitely weren’t him.
Eventually, we moved to a “Social Connect” construct. By getting creators to connect their main social handles we could verify who they were and that they were, in fact, who they said they were.
We built this into the onboarding experience and used it as a way to segment high potential creators.
Unmeasurable variables were just that, unmeasurable. We knew they existed—intrinsic motivation, desire to monetize, looking into the camera, etc.—but we couldn’t verify or detect them with data in the same way we would with connected data sources. As an aside, for a hack day project we had a person see if they could accurately predict the success of a creator and then write down all of the variables and cues they were evaluating subconsciously. That’s where we identified variables like “looking directly into the camera” as a success factor.
So we had to rely on quantitative (measurable) variables.
Here’s what we did:
Through a series of experiments we added more and more of the social variables identified above and tried to optimize for creators connecting those accounts. When a creator connected an account we grabbed several data points and pulled them into our record for that creator. Several of the data points were ones I mentioned above. In real time we then evaluated those variables against a set of predetermined thresholds for success and if a creator had a data point at or above that threshold we would surface those as a “lead” for our inside sales team and kick off an automated workflow to trigger outreach and connection with that creator. The goal was to get them into a conversation with our team so that the “human intervention” bucket was filled with qualified creators.
On the back of these two sets of experiments we were able to bridge from a purely self-service onboarding to a hybrid, or PLS, model.
Leveraging PLS to Drive More PLG
One of the beautiful things about moving from purely PLG to a PLS motion is that you can use your sales efforts as a form of customer discovery and feed that back into your PLG engine to make it better. And this is exactly what we did.
In conversations with creators we identified as high potential we learned about some of their barriers around foundational elements like “what do I offer my fans” or “how do I build momentum for a launch.” We were able to then take the advice that we gave creators via human intervention and build those into the onboarding and setup experience for every creator.
Examples of this were tier and benefit creation and ideas like the $5 tier, pre-defined benefits, and special offers to help with launch momentum. All of these are still in the product today.
Wrapping it all up
At Patreon, after our cold start phase, we went through three distinct motions for acquisition and onboarding from ~2015 to 2020.
Phase 1: A high velocity strategy powered by leveraging lighthouse creators.
Phase 2: A PLG phase with a relentless focus on onboarding and fueling the WoM, content curation and sharing of a creator’s page
Phase 3: Bridging PLG -> PLS -> PLG (lather, rinse, repeat)
Each phase unlocked another step function change in growth for the business and was made possible by some of the work we did in between (like rebranding, for example).
Today, Patreon is primarily a PLG motion which is sort of what I anticipated given their size and scale. The product is still primarily for the “mid-tier” or “torso” creator (head / torso / tail) and the core flywheel still exists and works well for that segment.
Given their ongoing investment in community and shopping that could be opening the door for a more enterprise-level creator who is less interested in monetizing via membership than in managing a community. Time will tell if they’ll need to bridge again to PLS or an enterprise sales team.
Thanks for reading!