The PLG Playbook
PLG as the new inbound
The fastest-growing companies right now, Cursor, Perplexity, and most of the AI tools gaining traction, are all running product-led motions. But having a free plan is not a growth strategy. The question is what you do with the signal it generates.
Jay Desai, Growth Lead at Navattic, and Dima Khlopkov, GTM Engineer at Gumloop, both run hybrid motions where product usage data feeds directly into outbound. They walked through the actual plays they have running in Unify today, from signup qualification to closed-lost reactivation, and showed how a small team can build a system that catches the accounts self-serve will not convert on its own.
Start with one or two signals, not twenty
The instinct is to instrument everything. Both panelists pushed back on that.
Jay's approach at Navattic starts with identifying the one or two key actions you want users to take in the product, then building plays around those milestones. For Navattic, that means tracking when someone hits an upgrade alert five or more times, or when a free user publishes and republishes their demo, because those behaviors correlate with conversion at 10% or higher. They validate cohorts in PostHog first, and only build a Unify play once the data confirms the signal is real.
Dima's take was even more direct: people overthink signals. The biggest unlock is often the simplest one, outreaching to qualified signups immediately after they complete onboarding. At Gumloop, they capture onboarding data in PostHog, pass it to Unify, and use it to route prospects based on role, tools connected, and company size. If someone finishes onboarding, they are in the best possible state to have a conversation.
Tiering decides who gets a human
Not every signal deserves the same response. Both teams use tiering to match the motion to the account.
Dima walked through Gumloop's three-tier system in detail. Tier 1, the largest companies with the most complex sales cycles, gets fully manual outreach with personal context that AI cannot surface. Tier 2 is semi-automated, with automated emails but manual LinkedIn steps. Tier 3 is fully automated end to end. The split is based on company size and the complexity of the deal, and it means reps spend time only where it matters.
Jay described a similar philosophy. At Navattic, some plays just flag accounts to the rep's Slack channel so they can decide whether to engage. Others spin up automated sequences. The key is validating the signal with sales first. Jay runs Slack notifications before building sequences, letting the sales team tell him which signals are actually worth automating.
Name-dropping as a personalization strategy
One of the most tactical moments in the session was Dima's walkthrough of Gumloop's name-drop play. When a qualified signup does not reply to the initial outreach after seven days, the play automatically prospects 10 more people at the same company, a level above or on the same team, and enrolls them in a sequence that references the original signup by name.
The messaging stays short and role-specific. Marketing teams get a different angle than GTM teams. The personalization is not AI-generated fluff scraped from the internet. It is internal context, someone at your company just tried Gumloop, that the recipient cannot get from any other email in their inbox.
Dima's framing: timing beats messaging at every level. When someone just tried your product, they are in the best state to engage. Combine that with a name-drop, and response rates stay in the 10-12% range even at scale.
Closed-lost is not closed forever
Jay shared a play that flags closed-lost accounts that return to the product. If someone from a previously closed-lost company signs back in or creates a new account, and there has been no sales activity in the last 90 days, the play fires a notification to the sales team.
This one is still in validation mode. Jay is running it as Slack notifications rather than automated sequences, intentionally, to get feedback from reps on whether the signal is high-quality before scaling it. That pattern, notify first, automate later, came up multiple times as a best practice for testing new product signals.
Signal-based outreach and the end of pure cold
The session closed with both speakers pointing in the same direction. Signal-based outbound, where you reach people at the moment they are engaging with your product or content, fundamentally outperforms cold outreach.
Dima shared that Gumloop is now combining product signals with LinkedIn engagement signals. If a prospect viewed the founder's profile or engaged with company content, that is the trigger. They are manufacturing more of those signals intentionally through employee-generated and executive-generated content, and attaching plays to each one. The result is outreach that feels relevant, not because of AI personalization, but because the timing is right.
Jay's focus going forward is experimentation. More plays in Unify as Navattic identifies new high-converting cohorts, and more product experiments to get users into those cohorts faster. Not everyone who signs up for a free plan will raise their hand. The system needs to catch the ones who will not.
Meet the speakers' companies
About the speakers
Gumloop is an AI automation platform that lets teams build and deploy workflow automations without writing code. Backed by Benchmark in their recent Series B, Gumloop is used by teams at DoorDash, Ramp, and Gusto. Want to try it? Use code UNIFY for a free month of Gumloop Pro.
Navattic powers interactive product demos and tours that let prospects experience your product before they ever talk to sales. Companies like Ramp and Unify use Navattic to drive product-qualified leads straight from their website.
