Microsoft Corporation

Program Management for Search Products, 2013 - 16

I worked on two surfaces of Microsoft's search products: Bing and Cortana/Search on Windows 10. As Program Manager I studied usage patterns of our products, analyzed the market landscape, helped translate insights back into product designs, wrote technical specifications and ran online experiments.

Visual System for Bing

For every search term, a myriad of technical services provide results from different parts of the web index, and in different formats. These are then laid out in real time.

The Visual System controls the styling of each element on the page by setting out the building blocks of typography, spacing and interactive elements. In addition to coordinating the rollout of a Visual System refresh to all of our customers, I helped optimize many parts of the Visual System. The scale of the operation (in terms of technology and usage) allows the measurement of effects from minute changes to the presentation of the content.

I presented the results of our approach and experiments to the Bing Experiences division, hoping to identify not just isolated results but patterns that identify fundamental user behaviour – insights that can then flow back into product designs. The desire here is also to preserve the sanctity of a design philosophy as it undergoes significant scrutiny, optimization & change.

The renderings below show a selection of experiments that I helped define and analyze. The center column shows the base design, from top to bottom: search button, web results, knowledge graph results. The left column shows selected variations of color and spacing, the right column shows selected variations of typography.

Responsive Design for Bing

Search results adapt to screen size – selecting, repositioning or reformatting content. I helped define these decisions, operating within technological architecture defined by the stringent needs of download and rendering speed.

Search on Windows 10

Windows 10 includes a renewed effort to improve and grow the capabilities of search in the operating system. Alongside proactive content from a digital personal assistant, it provides a single surface to search for content across the web, your own computer, and a number of cloud services.

My work touched various parts of this experience, a few notable pieces are highlighted here.

The first defined how the search experience scales across devices in layout & interaction. I also ran an effort to bring together Bing's mobile search pages, which previously bifurcated across operating systems. All decisions on responsive logic are taken to ensure that content-production continues to work smoothly for teams comprising hundreds of individuals.

My second major body of work was to crystallize the interaction framework of the 'My Stuff' page – a detailed search operation across the user's machine – specifying how to display content as it comes in from various sources, how to order it and provide the ability to sift through it. After the release of Windows 10 I also took on work to help improve the relevance and ranking of the user's content.

While on the My Stuff team, I led an effort to simplify search by merging My Stuff with the initial 'query formulation' pane – as interface and through underlying technology. The resultant experience provides greater quality over a smaller surface area.

Data Analysis & Online Experimentation

My main role was to put storied onto terabytes of data generated every day, about how Search was used. A number of these efforts then translate into direct product initiatives: such as adding type-based filters, and recent files. Through this effort I also defined our approach to measurement and A/B testing on Windows: defining metrics and defining analytical lenses.

Much of our analysis work starts with well-understood, commonly used lenses. Through these we search for interesting patterns that we then dive into further by looking at samples. Hypotheses are then formed and verified with further analysis, and useful lenses operationalized to continue monitoring our findings over time.

Cortana Companion Pane

I directed initial user testing of ideas on how Cortana can provide insight specific to a particular app. The results of these tests and our design iterations later led into what is now Cortana insights inside the Edge browser.