I led growth UX for GoodData's repositioning as GoodData.ai, owning the acquisition funnel, a clearer two-path journey across AI-Driven BI and Agentic Analytics, and partnering with the enterprise-product UX designers to carry that story into the trial.
Overview
I led growth UX for GoodData's repositioning as GoodData.ai. I owned the acquisition funnel end to end and partnered with the enterprise-product UX designers on the trial, so the agentic promise held from first touch to first insight.
The job: introduce Agentic Analytics without disconnecting it from the BI platform buyers already trusted, then make sure that story carried into the trial so interest actually converted.
+290%
Contact us conversions, pricing-page H1 test
2
Buyer paths, from a sprawling product menu
6
Surfaces redesigned in a phased rollout
The problem
GoodData was becoming GoodData.ai, but the funnel still had to protect the analytics buyer journey people came for. The site showed too many capabilities side by side, choice without hierarchy, and the agentic promise risked falling apart the moment a buyer entered the trial.
The goal spanned the whole journey: rebuild acquisition around two clear paths, AI-Driven BI and Agentic Analytics, then make the trial deliver on that story so activation and conversion held up, not just clicks.
"Selling agentic AI is a translation problem. Buyers don't want a chatbot, they want trust that an autonomous workflow won't embarrass them in a board meeting."
What I owned
Product Marketing owned positioning. I owned the acquisition funnel UX end to end, and partnered with the enterprise-product UX designers to extend that story into the trial.
Information architecture (owned)
Restructured acquisition around AI-Driven BI and Agentic Analytics, validated with card sorting and buyer research.
Conversion & pricing (owned)
Aligned Request a demo and Contact us flows and clarified the pricing narrative, the +290% H1 test lived here.
Trial activation (partnered)
Worked with product on the in-trial first-run so users reached a first insight fast, the strongest predictor of conversion.
Growth loops (partnered)
Contributed to contextual upgrade moments, feature gating and limits, that turn active trials into paid plans.
From a portfolio menu to two offerings
The old menu put BI, Analytics Lake, Analytics as Code, AI Assistant, and Embedded Analytics at the same level, with a Product Overview on top adding another decision layer.
We rebuilt around two primary offerings and carried that structure through navigation and homepage, so users met one consistent mental model: the analytics they knew, and the new strategic AI.
Tradeoff
Some grouping decisions had to be made before every visual detail was final. We shipped once the structure tested well and kept refining consistency after launch.
Validated with card sorting & research
We started with internal card sorting to align teams, then tested externally with buyer personas. The research separated product strategy from user comprehension.
AI Assistant moved under Agentic Analytics to match product direction; Embedded Analytics stayed under AI-Driven BI where buyers expected it; Security & Compliance became a global trust layer, not a product item.
Keep "analytics" explicit
Buyers still came for analytics. Pushed too far into AI terminology, the core value, and especially the pricing story, got harder to understand.
So we kept analytics explicit where intent was strongest. On the pricing page, an H1 test that added "analytics" lifted Contact us conversions by 290%. We also restructured pricing into three buyer-facing categories: AI / Agentic Analytics, Business Intelligence, and Analytics Lake.
"Agentic AI was compelling, but analytics remained the anchor for trust, intent, and category recognition."
Separate the category page from the conversion page
Explaining a new category and driving one specific action are different jobs. The Agentic Analytics page became the category pillar, teaching the offering and its capabilities.
The Agent Builder page became the conversion surface, focused on how to build and deploy governed AI agents with control.
Tradeoff
More surfaces to maintain, but a cleaner funnel: understand the category first, then move to a focused capability page when closer to action.
From sign-up to first insight
Acquisition only pays off if the trial activates. The biggest predictor of conversion was whether a user reached a real insight in their first session, so the trial had to remove the blank-prompt problem.
Inside the workspace, the AI assistant sits beside live dashboards and offers concrete starting points, search a dashboard, create a visualization, answer a business question, turning curiosity into a first win in minutes, not days.
"The marketing screenshot and the real screenshot should be the same screenshot."
Growth loops built into the trial
The trial wasn't just a demo, it was a conversion surface. We designed the upgrade moment to show up where users felt the ceiling, not as a generic paywall.
Feature gating, early-access unlocks, and query limits created natural, contextual reasons to upgrade: the more value a user pulled from agentic features, the clearer the case for the paid plan. Upgrade and Contact sales stayed one click away throughout.
Tradeoff
Gating risks frustrating trial users if it's too aggressive. We gated advanced agentic features while keeping the core insight loop fully usable, so the upgrade felt earned, not forced.
Rolled out to reduce journey risk
Changing everything at once risked broken paths and inconsistent messaging, so we shipped in stages: navigation, homepage, Agent Builder, Agentic Analytics, pricing, then supporting pages.
Navigation came first because it defined the architecture, if users couldn't grasp the structure, nothing downstream would land.
Tradeoff
Some surfaces were temporarily more aligned than others. We accepted that to validate the structure and ship the highest-impact changes first.
