Serverless Gateways

90% of new organizations failed to activate after signup. Not a minor usability gap. A structural growth problem. To get started, users had to provision a dataplane, connect it via CLI, create a service, and define a route. Each step assumed architecture knowledge most users didn't have yet. Each step required jumping between Konnect, documentation, and a local terminal. 60 to 70% dropped off at the first wall they hit.

95% of ARR came through Sales because the product couldn't close itself. I designed Serverless Gateway to change that. Remove infrastructure as a prerequisite. Provision a ready gateway at signup. Put users in front of value before complexity could stop them.

Kong Konnect saw around 1,000 new organizations per week. 90% of them failed to activate after signup. 60 to 70% dropped off at a single step: Connect a Dataplane.

To get past that step, users had to provision infrastructure, connect a local dataplane via CLI, then return to the console to create a service and define a route. Each step assumed knowledge they didn't have yet and required context-switching across three different environments. They were being asked to set up before they could evaluate.

Backed by funnel data, support tickets, and user interviews, four core blockers emerged:

•Send API requests right after signup. Users expected to evaluate quickly. Setup was a blocker.

•Get guidance, not architecture lessons. They needed help configuring services and routes, not understanding control planes.

•Explore safely without setup. Users wanted to try Kong in a secure environment without managing infrastructure.

•Know what comes next. They needed a clear path to scale when ready.

I want to try Kong, but I don't want to spend hours setting it up just to see what it does.

Kong customer, user interview

Serverless Gateway eliminated infrastructure as a prerequisite. When a new organization signed up, a gateway was provisioned automatically in under 30 seconds. No CLI. No dataplane. No documentation required before first use.

The new flow went from seven steps spread across three environments to four steps entirely inside Konnect. Users could go from signup to sending real API traffic in under five minutes.

Serverless Gateway was never meant to be the final destination. The design included a clear upgrade path for users who were ready to move to self-managed infrastructure at production scale.

Upgrade guidance surfaced at usage limits. Configuration handoff to hybrid gateways was guided step by step. Serverless became a launchpad, not a ceiling.

Serverless Gateway launched July 2024.

Activation turned out to be a systems problem, not a UI problem. Fixing the entry point helped, but every step I touched surfaced dependencies on things upstream or downstream I hadn't yet designed. The whole journey from signup to first value had to be coherent before any single piece of it could land.

The insight that stuck with me: developers don't evaluate products by reading about them. They need to touch something real. Every minute of setup before that first moment of use is a risk to the relationship. Not a preference. A risk.

What surprised me most was how much of the friction was conceptual, not technical. Users weren't blocked because the UI was hard. They were blocked because they didn't understand what a gateway was or why they needed one. The architecture diagrams and progressive disclosure patterns weren't polish. They were load-bearing.

This project became a reference point internally for how Kong could approach PLG. Not a playbook I handed off. A pattern other teams could build on.