Directing the adoption of a new process for building great Google products
The charter was established: Assistant was to be the next frontier for Google’s product experiences moving forward. The team grew rapidly to support this major initiative, and because of it, the design team had assumed a great deal of growing pains. And to add to the complexity, there was a newly established decision to find ways to merge the Assistant product experience more deeply with Google’s prized product experience: Search. The team was struggling to define the future of the product experience across all surfaces (phones, tablets, TVs, smart speakers, smart displays, smart headphones, automobiles, etc), and it was clear that a structural re-think could help them find a break-through.
The process started by assessing the current organizational structure, which included over a dozen teams made up of hundreds of designers.
The next step was unique to the process, but an important one: to examine the product through the lens of the user. Empathizing with the user is something the design teams always do, but not in this context. The purpose of doing so in this context was to understand how the team’s structure might inadvertently influence the product experience. But more importantly, it helps expose opportunities to improve the product. And for Assistant, there were a few.
The problem with most virtual assistants at the time, not just Google’s, was that they had under-delivered on their promise.
The technology’s limitations don’t have to be handcuffs that stunt the product’s ability to become transformative. Those limitations can be used as convenient constraints for finding creative ways to build a useful product. With this perspective and freedom, new business goals emerge. And with that come opportunities to establish a team around those goals.
User’s expectations with assistive technology
Accuracy
Assistant knows me well: Family and friends, personalization, learn by shadowing me
Efficacy
I can communicate with Assistant like a human
Utility
Assistant is useful, and helps save me time and effort
Versatility
Assistant is adaptable, amenable and adjustable
Correcting when wrong
Starting with the goals a user has for Google Assistant
The problem with most virtual assistants at the time, not just Google’s, was that they had under-delivered on their promise. And over time, that had affected users’ perception of the utility and value of virtual assistants . What was once heralded as the groundbreaking new way to interface with technology became just a quick way to set timers and reminders, or to control your media. It became unclear what one could ask of the virtual assistant to do, or not to do. And this reality can’t be ignored because understanding more deeply what users demand or expect from Google Assistant in its current state was an important step towards delivering a valuable experience, especially with those shortcomings in mind.
So what would they come to Google and Google Assistant for?
Purchase/Reserve:
I’m looking to order or reserve something more easily and expeditiously; and I know using Assistant can help because it can access my personal details and knows my preferences.
Consume Media:
I want to pick something to listen to, or watch, in a less cumbersome way, and I expect technology should be able to help me. I also would like intelligently personalized recommendations when I don't know exactly what I’m in the mood for, and it should be able to do this because it knows my tastes.
Research or Learn:
I need a quick answer to help make a decision, or I’m researching something new, or I’m acquiring more knowledge from a topic that interests me.
Coordinate and Communicate:
I want to keep in touch and socialize regularly with friends and family. I want to do this as easily as if it were being done in-person, and I expect technology should be able to help me achieve this.
Productivity:
I’m great at multi-tasking in the real-world, so when I use technology, I expect it to help me do this like I can in the real-world, or even better.
Control other technology:
I know technology is empowering us to have greater control of the digital things in our lives and I expect that it should 1. make it easy, 2. be personalizable, and 3. be accessible in a human-understandable way.
Let’s look at the Purchase/Reserve user goal, for example. To serve that user properly, it’s helpful to consider every part of the journey that they encounter as they pursue that goal. Which, in this case, includes:
Team A: Discovery
Search/Browse • Promotions • Growth
Team B: Purchase
Shopping cart • Reserve • Checkout
Team C: Post-purchase
Product reviews • Returns • Redeem • Modify
Each one of the above are important parts of the overall journey, and failure at any given point can mean disaster for a product or service. It’s critical to deeply understand each of these points such that users have a better chance at achieving their overall goal, each step of the way. Understanding this well is what can sway users away from your competitors, and build long-term business relationships with your customers. Because of this, it’s best to have teams structured to focus on each one of these points of the journey so they sympathize with users, and understand all of the minutiae in ways that allow them to optimize each step of the user’s journey. And the goal in the end is that the teams focus on getting users to successfully complete those various journeys they’ll experience with the product.
For Assistant, my goal was to take all of the organization’s divisions and re-organize them functionally according to a user-journey model. This was informed by 1. the goals that users expected of the Assistant product, and 2. the goals I believed the technology could deliver on. The result was a simplified and leaner organization made of 7 functional teams, down from about 33, with a few horizontal groups that collaborated across all verticals.
Simpliflied key user journeys for Google Assistant
Goal 1: Delegate
Resolve home-life’s administrative tasks
Team A: Submit & approve to-dos
Team B: Review, Modify, Collaborate on progress
Team C: Troubleshooting
Goal 2: Train
Improve & personalize my assistant
Team A: Teach
Team B: Review & repair learnings
By prioritizing Assistant's various 'roles' and structuring the team around user goals or journeys, the product can be more focused. This focus implicitly limits the jobs Assistant takes on, which allows users to easily remember or categorize its abilities, aids discovery, and sets more appropriate product experience expectations. In turn, the output can deliver consistent high-quality experiences.
The goal for AI in this case is to alleviate burden from the mundane, reduce unnecessary stress from repeat multi-tasking, and to give us back the time we would otherwise spend on the tasks that don’t fulfill us.
If implemented properly, users should feel in full control of what the AI is doing while feeling confident that it can operate autonomously on their behalf.
