All work

Growth & access

Designing Microsoft's Copilot Growth Engine

Turning a patchwork of upgrade prompts into one contextual, intent-aware system for how millions of people discover, access, and adopt premium Copilot across Microsoft's products.

Role
Product Designer · Growth
Surfaces
M365 · Teams · Edge · Mobile · Copilot
Timeline
2025-present
Focus
Growth · Access

Overview

As Copilot spread across Microsoft 365, Teams, Edge, and mobile, the upgrade prompts fell apart. They were inconsistent and rarely matched what someone was trying to do. I helped build one system for how millions of people find and get access to premium Copilot, whether they buy it, start a trial, request a license, or get assigned one by an admin. The goal was to make monetization part of the product instead of a layer bolted on top.

The challenge

Premium grew up one entry point at a time, scattered across the surfaces. Inconsistent patterns and unclear value made upgrading confusing, and hard to scale.

Research found the root cause. People couldn't tell what premium offered, or why it mattered.

My role

Because the work spanned every surface, it was as much about alignment as design. I partnered across product, engineering, growth, research, content, and accessibility to pull scattered experiences into one system.

Reframing the problem

It would have been easy to design better upgrade screens. But isolated prompts were what fragmented the experience in the first place. The real question wasn't the screen, it was the system around it.

“How might we surface premium value at the exact moment a user needs it?”

That reframe moved us from generic paywalls to contextual value moments, with premium as the answer to what someone was already trying to do.

What research told us

Four patterns kept showing up, and each one shaped the system.

Users wanted clarity before commitment

People wanted to see what they'd gain before committing. A clear comparison beat a simple prompt.

Context beat promotion

Contextual moments far outperformed generic entry points. Growth data named them the single strongest conversion driver.

Enterprise users needed a justification, not a button

Many enterprise users couldn't just click “upgrade.” They needed enough to justify a license request to a manager.

Friction lived everywhere

SKU choices, approval workflows, and too many options drove dropoff, often after intent was already there.

Three principles that guided the work

A system of upgrade patterns and paths

Instead of one-size-fits-all prompts, I designed the upgrade experience as a system: three reusable patterns, each matched to a different reason someone might upgrade, plus one acquisition framework that connected the paths to convert. Every surface got a consistent, intent-aware vocabulary instead of bespoke screens.

1 · Persistent upsells

Always-available entry points, used when awareness was the main gap.

Persistent upsell in the left navigation: the plan name and Upgrade button, highlighted. A persistent Upgrade call-to-action in the Copilot Cowork experience.
Persistent upsells: always-available entry points. The plan/Upgrade card in the navigation (highlighted) and an Upgrade call-to-action in Copilot Cowork.

2 · Contextual upsells

Triggered by what the user was actively trying to do.

Rather than sell a subscription, we surfaced the exact capability they reached for.

Teams Intelligent Recap prompting a premium upgrade to unlock the feature.
Teams Intelligent Recap
The Agent Store, where premium agents drive a contextual upgrade moment.
Agent Store
Image generation surfacing an upgrade at the moment a usage limit is reached.
Image generation
Contextual upsells: premium surfaced at the moment of intent, the exact capability a user was already reaching for.

3 · Constraint-based upsells

Triggered at a limit.

The hard part was tone. These moments had to feel informative and fair, never punitive.

Copilot on the web showing a calm inline notice: 'You've reached the file upload limit. Upgrade to continue now, or come back tomorrow,' with an Upgrade button.
File limits
Word with the Copilot pane explaining it is experiencing high demand and prioritizing Premium customers, offering to try again later or upgrade for priority access.
Copilot throttling
Word with the Copilot pane showing an 'You're out of credits' card, explaining Premium offers the highest AI usage limits, with an Upgrade button.
Out of credits
Constraint-based upsells: informative, non-punitive moments at a limit. A file cap, throttling under high demand, and credit depletion.

4 · One acquisition framework, many paths

The same thinking applied to how people got access. The system had to support several entitlement models at once:

Rather than design each in isolation, I treated them as connected pathways in one framework, so a user or admin could move between them without hitting a wall.

The Copilot upgrade dialog for a commercial user: Microsoft 365 Copilot (Premium) with a Request license action and a Buy now action, alongside the apps it works across and a list of benefits. The Copilot plan comparison: Microsoft 365 Copilot Chat (Basic, current plan) next to Premium with a Start trial action, noting the organization can approve continuation after a 30-day trial.
One acquisition framework, many paths: request a license or buy now (left), or start a low-friction trial with org approval (right). Connected, not isolated.

Impact

44%
Click through rate
85k+
Monthly license assignments
5+
Surfaces unified

More than any single number, the work changed how the team thought, from designing paywalls to designing value-discovery experiences.

What I learned

The biggest lesson: monetization works best as part of the product, not a layer on top. People accept an upgrade tied to a goal they're already chasing, and they resist a generic pitch that interrupts them. Designing for that difference, across every surface, turned disconnected prompts into a growth engine.