Pinecone’s journey from seed to 10,000 sign-ups per day
VP of Marketing Greg Kogan on early go-to-market strategy and PLG
👋 Hi, it’s Kyle from OpenView. Welcome to another free edition of Growth Unhinged, my newsletter that explores the unexpected behind the fastest-growing startups. This week I’m featuring Pinecone, the picks-and-shovels star of the AI boom.
New to Growth Unhinged? Here’s what you missed in July:
I’ve reached that stage of my life when I no longer say “I knew them when…” about bands. But I do, from time to time, get to say it about fast-growing startups.
Pinecone, the vector database startup, recently announced a new $100 million investment at a $750 million valuation. Demand for their products is truly off-the-charts, notching over 10,000 sign ups per day (although they might point out that sign-ups are a lousy measure of growth!).
I met the brilliant crew at Pinecone in October 2021 as they were just launching their self-serve version and building an early PLG motion. Greg Kogan was heading up those efforts and had a rare sense of clarity about how to “do PLG” when selling an enterprise-grade product. He joined Pinecone as employee #10-ish after nearly eight years of consulting startups.
Greg and I reconnected to reflect on Pinecone’s epic growth journey. Keep reading for his insights into how to set up go-to-market for an early stage startup and how to adapt PLG to sell high-ACV deals.
Taking a bet on Pinecone as VP of Marketing and employee #10-ish
Before joining Pinecone, Greg spent eight years consulting enterprise software startups on marketing. He actually began as a consultant for the business, helping Edo Liberty (Founder & CEO) and team on a three-month messaging and positioning project.
At the time Edo had a product vision based on his observations at Amazon and Yahoo. The business was extremely early with one pilot and zero paying customers. Edo knew it could be useful, but didn’t know what to call it, how to describe its benefits, or exactly who it was for.
In the process of working on messaging and positioning, Greg convinced himself that the product had legs. He saw the market opportunity and he had a clear sense about how Pinecone could win the market. His hypothesis was that Pinecone should lean into self-service and target engineers rather than data scientists who were the default persona at the time for AI-oriented products.
“As soon as I joined, we were able to make some strong strategic decisions early on and they turned out to be right. One was to target engineers and not data scientists. The other was to commit to PLG. The third was to take a developer-focused marketing approach.” - Greg Kogan
From there Greg went to work to educate the market, helping engineers discover, explore, and build with vector search. They had decided to create a category and their first opportunity to do so was by announcing the seed funding with Wing. This fundraise announcement forced Pinecone to refine the messaging. “Thankfully, the story and messaging stuck,” Greg recalled.
For the rest of the year, Greg focused on refining the messaging further and continuing to look for signals of what was working and what wasn’t. He recognized the need to segment engineers by their familiarity with machine learning (ML), which he described as follows:
Aspiring: they’re interested in ML, but don’t have a use case
Average: they’re committed to bringing AI/ML into their company for some use case, but don’t know how to do it
Advanced: they’re actively in market for a vector search or vector database solution
Pinecone went after the advanced group first as potential users, while also targeting the “average” group with educational materials that would help them graduate into the “advanced” group. Even still, it took the team a while to realize that while the vector database concept got folks curious, it still left people with too much thinking to do about how to use it today and for what purpose. “We experimented with different approaches in our messaging and found what works best is to focus on the semantic search use case, enabled by the vector database,” Greg told me.
His success metric early on was around post-signup activations and demo requests, a signal that folks were responding to Pinecone’s message. He would also regularly feed insights from the market and developer community to the product team to inform the product roadmap.
Setting up a PLG motion for developers
Where did PLG fit in the priority list, I asked?
#1 was the response. This was actually a pretty big reversal for Greg, who initially was skeptical about the value of PLG when you’d ultimately need to sell to the buyer/executive. Wouldn’t PLG just take more effort and time to see a payoff?
After getting clear on messaging, Greg took a deep dive into the market. He talked to users/prospects, read everything, and watched what other successful PLG companies were doing. He also developed an initial marketing strategy (who’s our audience, what do they need, what do we say to them, and how do we reach them?) and set up the right tooling.
That upfront work drove extreme clarity of execution. Greg knew that engineers discover new products through learning and from their peers in communities like Hacker News. So Pinecone set out to build the best learning center for vector search and vector databases, and consistently share new articles in those communities.
“Instead of hiring content marketers or demand gen marketers to run ads, our first hires were former engineers who wanted to do developer advocacy. They started creating a bunch of educational materials around adjacent topics to Pinecone.” - Greg Kogan
At the time, people weren’t searching for Pinecone in Google, but they were searching for certain topics around vector search. The developer advocates would write about those topics and then push it to Hacker News. They’d engage with people on Twitter and elsewhere to get them to read about the topic. At the same time, the team did a lot of user interviews, diving deep into the niche communities and the people who were active in them. They fed this information back to product and engineering who were able to use it to make the product more intuitive and easy to use for those same developers.
Pinecone’s initial PLG tooling was very simple. They used HubSpot to capture and send emails. They had Google Analytics on the website. In terms of KPIs, early on Greg decided that the measurements would be adoption and revenue. He looked at signups, activations, upgrades (paid customers), and active usage. Revenue was driven by usage and so was an outcome of people adopting Pinecone.
There were other foundational things that Greg’s team set up, including Pinecone’s documentation – not a traditional marketing capability. He knew that self-service and usability would be hugely important, and Pinecone’s docs had to be really good otherwise all of the other marketing wouldn’t have an impact.
“In general, it’s getting really involved in the community and seeing what’s sticking and what’s not. We were paying close attention to what confuses a new user and what makes sense to them.” - Greg Kogan
Scaling PLG for an enterprise-grade product
Pinecone scaled demand from zero to 10,000 sign-ups per day in less than three years. What contributed to this blitzscaling?
As a starting point, a few things that didn’t work include:
Sponsoring large events
Lead generation
Sales outreach
Content on introductory or general machine learning topics
In other words, the traditional B2B marketing playbook didn’t land for Pinecone’s product. What was successful was scaling what had already been working – developer advocacy and enabling self-service – and hiring up the team accordingly.
Specific things that inflected growth included:
Consistent quality content (i.e. technical content)
Competitors got behind Pinecone’s messaging, creating pull for the category
Enabling self-serve upgrades
Launching and opening up Pinecone’s free plan
Key partner relationships and collaborations (e.g. OpenAI)
The wave of market generated by ChatGPT and the GenerativeAI boom (it doesn’t hurt to be in the right place at the right time!)
Greg reflected that marketing in a PLG company is about drawing users to the product and helping them succeed. This is different than enterprise marketing, which requires drawing buyers to your value proposition and getting them to commit.
Pinecone makes a great case study for applying PLG efforts to ultimately win large ACV opportunities. Companies like HubSpot, Shopify, and Zapier, for instance, went from signing up for Pinecone’s free plan to production deployment in a matter of days – which wouldn’t have happened if they couldn’t self-serve.
Greg believes that developers at enterprises have similar evaluation criteria as their peers. “What’s different is the bar,” Greg reflected. “If you’re a developer at a Fortune 500 company and you want to introduce something to your team, the bar for the product’s credibility, reliability, scalability, integration list, performance… it’s just way higher.”
The product plays a big role here. Regardless of what’s on the website, the user will want to sign up and kick the tires. “If they get stuck or encounter a lot of friction in the first 15 minutes, you’ve lost them. The product has to stand on its own,” he emphasized.
Greg offered four tips for aspiring enterprise-PLG products:
Social proof. “In order to get them into the product, social proof is huge. If people hear that a lot of other developers and companies are using it, they’re just coming on the website to find the sign-up button at that point.”
Documentation. “The docs are super important, too. If the website were just the homepage and docs, conversion wouldn’t be that different than today. A lot of developers just jump straight to the docs.”
Pricing page. “People want to know if it’s open source, SaaS, etc. How will pricing scale? Is it usage-based, hardware, seat-based, something else?”
Community. “Outside of the website, building a community and getting people to talk about your product are hugely important. That’s specifically why we have a free plan. Part of the free strategy was to get people to even use Pinecone for personal or side projects with the idea that they want to share these side projects and get the word out.”
Closing advice for aspiring PLG leaders
I wrapped up by asking for Greg’s thoughts about unconventional advice for building and scaling PLG products. He didn’t hold back.
No new categories. “Although we created a new category with Pinecone, I still caution others that it’s harder and more time-consuming than they think. Instead, consider disrupting an established category or jumping onto an emerging category.”
Hire your users. “You need marketers who deeply know your users, or better yet, is part of your target audience.”
Signups shmainups. “Sign-ups are unreliable. They’re too far removed from business goals.”
The first pancakes always suck. “Everyone who joins my team is required to push something to production within three days. Some people freeze up upon hearing this on their first day. ‘The first pancake always sucks,’ I say. ‘So let’s just get it out of the way.’”
Try the soup. “Try your product, watch others try it, and talk to them constantly.”
Users dream, too. “They’re not just looking for features. They want to accomplish their own ambitions.”
Don’t confuse PLG with open-source or B2C. “You can pursue PLG and still win huge customers.”
PS — Follow Greg for more insights like this, or better yet: work with him!
Before you go…
There was no Growth Unhinged last week, but I did collab with
of to finally answer the question: “what is good free-to-paid conversion?”2023 marks our 7th (!) annual SaaS benchmarks survey. Please consider participating in the survey in order to contribute to the best dataset in SaaS. As always, I’ll share the TL;DR in this newsletter (it’ll be ready in October).
PLG tech: swipe left or right? I unpacked the tech you need to turn your product into a growth engine.
🙏 Thank you for reading. Tune in next week for a more personal post. If you’re finding this newsletter valuable, share it with a friend or consider subscribing (it’s free).
Thanks Kyle! This is an awesome writeup. I absolutely love how Greg took a customer-first approach to understanding Pinecone's opportunity and discovered the PLG playbook for developers.
This is such a good lesson for any marketing leader. As a principle, throw away your playbook and understand your customers first!