Productivity skills for Cortana

Helping you save time with messaging & calendar skills throughout M365

Rapidly changing tools & expectations are impacting how people work

Despite having powerful productivity software, people still find themselves wasting time by performing the same simple tasks over and over again.

I wish I had an extra hand so I can focus on what matters.

Sometimes I'm not sure what's actually important.

It takes so much time just to do simple things at work.

What if you had a workplace assistant that did the small stuff for you?

My role

From 2018-2020 I was leading a team of product designers responsible for defining and realizing a vision for core voice skills across Microsoft.

Our team worked with partners and stakeholders across the globe and of every discipline.

What I did

A healthy blend of people management & IC work
Product vision & strategy
Defining design principles
Sprint facilitation
Resourcing & R.O.B
Research coordination
Executive presentations
Fostering relationships between partner teams
Documenting patterns and best practices

It's a pretty big team.

How it all started

In 2018, Microsoft kicked off efforts to realize an opportunity to unify its core product offering into a single subscription-based model dubbed "M365".

The first phase of this effort was market research to gauge customer needs around the world. In December 2018 I flew to Berlin to observe concept value testing with focus groups in order to determine unmet user needs within our current set of products.

One of the highest ranking opportunities was the presence of voice assistance within our users' pre-existing workflows.

We know our customers want help...

The key to this effort was to define the landscape and understand how we might offer differentiated value to our customers. What can set Cortana apart from other digital assistants?

At work, people have expressed they spend the most time in 3 core areas, and they really want help.

Calendar

To-Dos

...but only if it's on their terms.

The more complex the work, the more a digital assistant becomes appealing. At the same time, users still want control; they just want an assistant to make their lives more efficient.

In discovery, we found big opportunities for improvement in broad search, deep search, and complex tasks.

"The quality of the solution depends on how well we understand the problem."

Combining our knowledge

Following our week in discovery, we came together and kicked off a week-long design sprint to share what we learned and align on outcomes. Acting as the facilitator, I coordinated the sprint across multiple time zones with participants of all disciplines.

Managing differing perspectives and opinions in any project is challenging. Relying on what we learned through data and research to drive decisions was key to a successful sprint.

A key outcome of our sprint was the alignment around 3 user pain points that we all felt were the highest priority:

"Capturing tasks and calendar events is time consuming."

"I waste time searching for what's important."

"I struggle to estimate and manage my time."

Our user promises

From our pain points we articulated three user promises to frame our scenarios around.

Cortana helps me manage my calendar

Cortana helps me save time by finding what's important

Cortana helps me prepare for what's next

Taking a principled approach

As we began this process, we defined a set of principles to guide us. These have been iterated on as we've learned along the way, but they still serve as our north star.

Voice first

Voice must always serve as the reference design; visuals must scale to a voice-drive experience.

Tailor to context

Understand where people are, what they're doing, and their level of trust in us.

Less is more

Our value should always be faster and more lightweight than the alternative.

Help people act with confidence

Insights and information we provide must be actionable, offered in the right place at the right time.

Don't try to do it all

Voice does simple things well. It's not good at complex interaction. Know when to refer people elsewhere.

User journeys & storyboarding

Our team split up the 3 promise areas and went really wide on a number of scenarios to explore across all of our supported endpoints.

Cortana, join my

next meeting.

Cortana, what does my morning look like?

Cortana, schedule a meeting with Bob for today.

Gaining support

After our two week sprint, we packaged up our vision and presented to our senior leadership team. The end result was the investment in 30+ use cases related to calendar, meetings, email, and messaging.

Deep dive: Calendar creation

One thing we're always striving for with our skills is getting the user from A to B in as few steps as possible. This goal is almost always our greatest challenge; the kinds of skills we're trying to enable are traditionally very complex tasks.

Think about how you might go about scheduling a meeting in Outlook today: You have to navigate to a new tab, find the right button to click, search for a person's name, add a title, determine a mutual meeting time, location, yada yada yada. Not a simple thing to voicify.

The Outlook mobile calendar creation screen

Exploring simple solutions

Chances are, if you're using voice to schedule a meeting, it's because you're situationally handicapped in some way and will need to trust your assistant to make some decisions for you. How can we strike the right balance between providing enough context to feel informed while at the same time make the right choices for you?

One example is how to help users choose a mutually available time to meet. In Outlook, this step uses a complex visual interface where users have to traverse multiple calendars. What if we could suggest several times for you? What is the perfect balance of context and simplicity to help the user confidently make an informed decision?

4 designs for suggesting meeting times which we tested with users

Learning from our users

After reviewing user feedback, we ended up creating a new iteration. We wanted to provide a short list of quick options to choose from that provided enough information without bogging users down in a heavy calendar management tool.

In this iteration, we provide 3 choices to the user, with a subtitle that informs the user if they have items before or after that time slot. If the user wants more options, we provide them a breadcrumb to the Outlook calendar tool. We also avoid suggesting times at the very beginning or very end of a work day, such as 8AM or 5PM. I hate 8AM meetings.