Iris Unification was a platform initiative to redesign how Growth teams at Coinbase plan, build, and launch marketing campaigns inside Iris—the internal multi-channel marketing tool for targeting, messaging, presentation, incentives, and experimentation.

‍

When we started, Iris wasn’t one product experience. It was a patchwork of systems: more than seven flows owned by different teams, spread across siloed surfaces like Notifications Systems, Presentation Portal, Rewards Platform, Iterable, Contentful, Experimentation, and others. The interface mirrored backend architecture one-to-one, so campaign managers often faced complex and arbitrary technical jargon. Setup was not self-serve and teams relied heavily on Growth Foundations and a small set of platform experts to create, deploy, and manage campaigns.

‍

Discovery surfaced a clear definition problem: there was no single source of truth for a campaign. To run a typical experimental campaign—target users, deliver an incentive, send notifications, and surface upsells—a CRM manager had to navigate through six different flows across three different platforms and manually verify everything was wired correctly. Launches routinely took four to five weeks, with managers needing to coordinate across systems, teams, and specialists. Assets were created and approved in isolation, with little confidence that segments, notifications, variables, or incentives were actually connected.

‍

Our north star was one guided campaign flow where every decision builds on the last through progressive disclosure, so creation feels continuous and intuitive rather than a scavenger hunt across tools.

The fragmentation showed up differently depending on where you were in the workflow, which is why the opportunity had to be systemic rather than a series of one-off fixes.

‍

Variables were created manually and often required referencing out-dated or hard to parse documentation, with no validation and no reuse across notifications or assets. Hydration required complicated SQL or fragile placeholder mapping, so small mistakes compounded across channels. Targeting lived in a SQL-heavy world as well that wasn’t truly self-serve and often required critical partnership with data science and platform experts. Email—the highest-volume channel, accounting for roughly half of notification sends—used a clunky builder that made creation painful, and produced inconsistent output when variables or HTML were involved.

‍

Governance amplified the operational cost. Approvals were siloed across journeys, notifications, targeting, rewards, and upsells, each with its own individually built status system. The status systems didn’t reflect mental models, so teams couldn’t trust what they were seeing when trying to move a campaign forward.

‍

These pain points led to a single design opportunity: collapse the stack into one campaign container with a requirements-driven flow, shared objects, and guardrails that make the right path the easy path—without removing power for advanced users who still need SQL, experimentation, or complex channel setups.

Iris Unification was a platform initiative to redesign how Growth teams at Coinbase plan, build, and launch marketing campaigns inside Iris—the internal multi-channel marketing tool for targeting, messaging, presentation, incentives, and experimentation.

‍

When we started, Iris wasn’t one product experience. It was a patchwork of systems: more than seven flows owned by different teams, spread across siloed surfaces like Notifications Systems, Presentation Portal, Rewards Platform, Iterable, Contentful, Experimentation, and others. The interface mirrored backend architecture one-to-one, so campaign managers often faced complex and arbitrary technical jargon. Setup was not self-serve and teams relied heavily on Growth Foundations and a small set of platform experts to create, deploy, and manage campaigns.

‍

Discovery surfaced a clear definition problem: there was no single source of truth for a campaign. To run a typical experimental campaign—target users, deliver an incentive, send notifications, and surface upsells—a CRM manager had to navigate through six different flows across three different platforms and manually verify everything was wired correctly. Launches routinely took four to five weeks, with managers needing to coordinate across systems, teams, and specialists. Assets were created and approved in isolation, with little confidence that segments, notifications, variables, or incentives were actually connected.

‍

Our north star was one guided campaign flow where every decision builds on the last through progressive disclosure, so creation feels continuous and intuitive rather than a scavenger hunt across tools.

The fragmentation showed up differently depending on where you were in the workflow, which is why the opportunity had to be systemic rather than a series of one-off fixes.

‍

Variables were created manually and often required referencing out-dated or hard to parse documentation, with no validation and no reuse across notifications or assets. Hydration required complicated SQL or fragile placeholder mapping, so small mistakes compounded across channels. Targeting lived in a SQL-heavy world as well that wasn’t truly self-serve and often required critical partnership with data science and platform experts. Email—the highest-volume channel, accounting for roughly half of notification sends—used a clunky builder that made creation painful, and produced inconsistent output when variables or HTML were involved.

‍

Governance amplified the operational cost. Approvals were siloed across journeys, notifications, targeting, rewards, and upsells, each with its own individually built status system. The status systems didn’t reflect mental models, so teams couldn’t trust what they were seeing when trying to move a campaign forward.

‍

These pain points led to a single design opportunity: collapse the stack into one campaign container with a requirements-driven flow, shared objects, and guardrails that make the right path the easy path—without removing power for advanced users who still need SQL, experimentation, or complex channel setups.

We framed the solution in two layers—architecture first, then experience—so engineering alignment and user-facing design could move together.

‍

Architecturally, the legacy model treated targeting, delivery, notifications, upsells, incentives, and experiments as separate hops with separate approvals. The future model centers around one Unified Campaign Builder as the container: one place where all campaign details, audience segmentation, incentives, notifications, presentation, and experimentation could connect intentionally, with modular support for push, in-app, email, upsell, and SMS channels.

‍

On the experience side, we led with requirements rather than empty forms. Builders declare what the campaign should include at the top—what channels to utilize and whether or not to include an incentive—and the flow generates only the steps required. Core metadata, timezone and evergreen support, unified targeting, image management, and experimentation were included in every campaign by default. We managed complexity through progressive disclosure. SQL targeting remains available, but the default path uses events, attributes, and plain-language summaries so more people can participate without requiring technical fluency or legacy knowledge. Variables hydrate on the backend, work across assets, and stay organized in a side sheet so the main UI stays calm. For channels where quality mattered most, we redesigned email as a modern drag-and-drop experience and brought upsells into the flow through direct integration with the presentation layer, starting with announcement cards, menu promos, and overlays.

‍

In order to support one Unified Campaign Builder we also built one unified dashboard experience where builders could manage every part of their campaign—post creation—all in one place through a simplified tab system.

‍

Apart from the end-to-end unified flow itself, dashboards were created for managing every part of the campaign all in one place through a simplified tab system. Where

‍

The guiding experience principle was continuity: campaign creation should read as one conversation where each decision sets up the next, not a tour of disconnected admin screens.