Teaching Copilot to sell without selling out

The mandate

When Copilot became a product people trusted for answers, it also had to become a business. Our job was to bring sponsored content to our AI Native experiences in a way that would create value for users and build revenue.

The stakes were (and still are) huge. With something as deeply personable as an LLM-driven chat experience, poorly handled sponsored content could break trust with the user and harm the user experience, both critical elements of a new product and new interaction paradigms.

Key decisions

Decision 01: Principles before pixels

Options: Start exploring ad formats right away, or agree first on what monetization in an assistant must never do.
Trade-off: Weeks before the first concept reached executives.
The call: We co-wrote monetization principles and brand guidance, and every concept had to answer to them.
Why: In an AI assistant, one bad ad experience costs trust in every answer.

A few key principles:

Copilot works for you. Period. Ads never influence your conversation with Copilot. What Copilot suggests is based on what’s helpful to you, not an advertiser, not Microsoft.
It’s an ad, you’ll know it’s an ad. If something is sponsored, it will be clearly labeled. You will always know.
Your conversation stays yours. What you say to Copilot won’t be made public or shared with advertisers. The context Copilot needs to personalize your experience, including advertising, is built to be safe and follow our privacy standards.

Decision 02: Advertisers in the room early

Options: Validate with users only, or bring advertisers into prototype research too.
Trade-off: More voices, and time away from direct design work.
The call: Prototype walkthroughs with agencies and brands, plus client-specific prototypes to test real business value.
Why: A format users like but advertisers won’t buy doesn’t ship.

Key advertisers included two global automobile manufacturers, a global cosmetics company and a global marketing agency.

Decision 03: Carry ideas across surfaces, and accept pivots

The call: Concepts born in Copilot moved into Bing. Some paused or changed direction as the platform evolved.
Why: We designed for change instead of against it, so work paid off even when plans moved.

As we decided to converge the consumer and M365 Copilots, a change in philosophy and a re-prioritization of engineering followed. Extending our work and learnings to new surfaces ensured we kept momentum.

What I built

An AI Native sponsored content charter spanning Copilot, Bing and MSN.
A rhythm of executive reviews (Microsoft AI’s CEO and LT) where designers presented their own work. We delivered fresh concepts and UXR at a blistering weekly cadence.
A joint workshop and sprint with Microsoft Commerce (formerly Microsoft Shopping) to co-create monetization experiences: one holistic experience across organic and sponsored content.
Staffed, coached and directly contributed design alongside a team of five additional designers, two partner design teams, two UXR teams, and multiple product and engineering teams.

Publicly reported program impact

Microsoft received AdExchanger’s 2025 AI Innovators award.
73% higher click-through rate for conversational ads.¹
16% higher conversion rate.¹
33% shorter customer journeys.¹

What I’d do differently

Product–market fit, execution and competing priorities contributed to the setback and cancellation of two core innovative experiences. If I could go back, I’d (1) help us adjust scope, (2) re-size and commission sufficient engineering, and (3) get marketing moving earlier on locking in white-glove participants.
I’d more assertively validate engineering quality, set timelines and escalate earlier.

Notes

1.Kevin Klein and Paul Longo, “73% Higher CTRs: Why Advertisers Need to Pay Attention to Conversational AI,” Microsoft Advertising Blog, August 6, 2025, https://about.ads.microsoft.com/en/blog/post/august-2025/73-higher-ctrs-why-advertisers-need-to-pay-attention-to-conversational-ai.