I help product organizations…

  • navigate ambiguity

  • repair flawed data

  • drive strategic alignment

…through rigorous user research.

For 10 years I’ve been researching products and services to define ecosystem metrics, embed foundational frameworks, and bridge the gap between digital products and users’ real lives at global scales.

I’ve worked with…

What I do…

Lorem ipsum dolor sit amet.

Case Study: Metric Integrity on Microsoft Teams

From 2021 to 2025, I served as the research lead for Microsoft Teams meeting experiences, a product ecosystem serving hundreds of millions of daily active users. I covered foundational, discovery, and evaluative research across the high-traffic join funnel, screensharing, breakout rooms, and premium product tiers, as well as program managing a global customer-panel.

While my day-to-day involved all of the above (and more), this case study deep-dives into a specific, initiative where data integrity, cross-functional deadlock, and product strategy collided: overhauling the “Meeting Join Success” OKR.

The Challenge: Leadership was operating on a product metric that miscategorized and mischaracterized normal user behavior as failures, leading to a strategic proposal to remove critical pre-join features.

My Role: Research lead partnering directly with Data Science, Product Management, and Engineering

The Outcome: Blocked a high-risk UX deletion, re-framed the core telemetry metrics, and established new behavioral bucketing systems which fundamentally changed how our team measured “success”.

The Friction: Telemetry vs. Human Behavior

The team was focused on a single OKR which had caught the attention of our leadership: reducing drop-off in the join flow. Data analysis showed a screen where drop-off was most likely.

Because the organization was looking purely at this quantitative drop-off, the prevailing logic was, “delete that screen, shorten the funnel, fix the metric”.

But this is a classic misinterpretation of usage data, assuming a causal relationship. I stepped in to argue that this data told us what was happening, but not why.

Triangulating the “why” with a mixed-methods approach and collaboration

I leveraged previous qualitative research I’d conducted, along with auditing the data pipeline with our Data Science team to build the case that the approach of, “Just delete that screen. They can’t drop-off there if it isn’t there.” was: likely flawed, high-risk, and was full of miscoding things as “failures” that weren’t. I then conducted targeted research which revealed both that:

  1. These drop-offs weren’t a problem for users nor the business

  2. That this step in the flow was high-value to users and removing it would actually cause the very UX and satisfaction issue the business was trying to avoid.

It wouldn’t stop the drop-off; it’d just move it to another step. Because this drop-off behavior wasn’t a problem for anyone other than our metric; we were chasing a number that wasn’t reflecting a real user problem/need and wouldn’t achive our business goals.

Creating a metric that maps to our user’s reality

Ultimately, we re-designed our internal OKR based on this work from a rather crude model to a more nuanced approach; which still allowed us to track potential issues while also accepting that this natural user behavior wasn’t “bad”. Such as the two following buckets:

  1. Corrected success: users who return and join successfully within a threshold (previously coded as failures)

  2. Exit failures: Users who enter but then leave shortly thereafter or experience a settings issues (previously coded as successes)

A new discovery around the social and emotional needs during meeting join

Not only did this work re-define our organizational thinking about our usage data, but my research also uncovered new opportunities for features to improve meeting join confidence and create delight. I proposed several pro-active information features based on this work which went into roadmap planning. Some of which are now live in Teams Meetings.