Founder Fridays: Shots on goal matter with Tim Dalrymple, Roadway and Josh Kim, Notion

Roadway is positioning attribution as the control center for growth marketing—now with an AI layer built to help teams decide what’s working, what’s failing, and what to do next. The pitch is straightforward: growth models go stale quickly when teams can’t measure the right things in the right way, and most growth marketers spend too much time wrestling with data rather than living inside it. Roadway’s “first AI platform for growth marketing teams,” launched alongside its attribution foundation, aims to turn messy, cross-channel performance data into usable, full-funnel attribution and optimization workflows.

The origin story traces back to the recurring attribution problem Tim Dalrymple saw while building growth models at Notion and Web Flow. Those acquisition-loop models—designed with clear assumptions—tend to lose accuracy after a couple of weeks when measurement doesn’t match the model’s structure. The breakthrough came from combining attribution with AI’s ability to interpret data in more structured ways and, crucially, to give teams context that helps them “live in the numbers.” That emphasis on operational fluency—building intuition and nuance internally—also underpins the argument that hiring and training matters more than outsourcing to agencies.

Validation didn’t arrive cleanly at first. Early onboarding delivered value that felt closer to another BI tool than a true attribution system, because the initial product work focused on analytics “table stakes” while the core attribution model and channel-deduplication capabilities were still being built. Over time, the platform evolved from trend-level chart understanding to chart-level analysis, and then toward full-funnel goals, guardrails, and self-serve construction—so growth marketers can build and iterate without waiting on bespoke analysis.

Dalrymple also frames today’s AI marketing landscape as less about replacement and more about applying enduring growth principles to new channels. Panic often comes from FOMO and from trying tools as if they will “replace” established workflows. Instead, AI search should be treated as part of SEO’s evolving ecosystem: consideration may shift into chat interfaces, but the underlying ranking signals and content behaviors remain. Teams that experiment in the “messy middle” without abandoning core principles—especially around distribution loops and evidence—are more likely to win.

A recurring theme is that the barrier to building is falling, but the barrier to quality is rising. AI can speed execution, yet it can also waste time when outputs are wrong—leading to back-and-forth and rework. Roadway’s approach leans on structured data and carefully scoped context rather than brute-force long-context reasoning. The system uses sub-agents with specialized context and passes the right data to each step, with the goal that AI stays “invisible” inside the workflow instead of becoming a separate feature.

Inside Roadway, velocity comes less from tools and more from decision mechanics: project owners drive direction, seek input but avoid consensus, and use a “disagree and commit” culture. That speed is supported by organizational breathing room to test, roll back, and try again. The advice to founders is similarly blunt—nothing is easy, action creates momentum, and conviction beats endless searching for what might resonate. Finally, Dalrymple says Roadway would have started with narrower ICP focus earlier and prioritized fast data connectors sooner, after learning where demand actually clustered.