Making complex data easier to understand
The Google Display Network is a set of websites across the web that advertisers can show their ads on. With more than half a dozen different targeting methods to choose from, and large amounts of data displayed across 7 different tabs, some Display Network advertisers found it hard to understand their campaigns’ performance. To separate important insights from noise, they needed to have the expertise to sort, filter, and slice all that tabular data in the right ways. Here’s what the old UI looked like:
We started by showing advertisers early wireframes of summarizations we thought they would find useful. We wanted to answer questions like: do advertisers care more about where their ads are showing, or the types of people the ads are reaching? Do they want to know what display ad sizes they’re missing, or do they want to see what devices their audience use?
After talking to some advertisers and hearing their feedback, we realized they had even simpler questions, like: Which of my targeting dimensions are getting the most clicks, and what age group is responding best to my ads? So we narrowed the field and focused on creating visual summarizations of existing performance data associated with advertisers’ targeting dimensions.
For the UI, we decided to use cards since they provided the flexibility needed for a dashboard. But cards were never used in AdWords before, and with so many UXers working on different AdWords features, we were always mindful of not doing things that would break consistency. So as with any new UI pattern, we documented general UI guidelines in our centralized spec site. These “living” specs were updated.
It took multiple rounds of user testing and many iterations for the Display summary project to evolve. Along the way I worked with the data visualization engineering team, defining new features for them to implement, such as the ability to show pie chart labels using whiskers. The final result was a simple, single-page summary that showed only the most important insights and allowed advertisers to dive deeper into the other tabs if required. This UI also provided inspiration for the later Material redesign of AdWords.
Making campaign creation easier
Creating a new ad campaign in AdWords used to involve filling out a long, complex form full of technical terms. In order to configure a successful campaign, advertisers had to know how to map their business goals into our features and terminology. This wasn’t easy for everyone. For example, this was the form for only the first step of the campaign creation process:
I was asked to turn this into a more human experience, and we decided to put users’ goals front and center — literally. The idea was to start by asking advertisers what their goals were, and then highlight only the relevant features.
Our two main challenges were: 1) defining a common set of goals that accurately portrayed advertiser needs and 2) creating the best possible user experience for goal selection. This was a big effort that involved a multitude of stakeholders across product areas within and beyond AdWords, who all had to agree on a commonly defined taxonomy and overall user experience.
While my PM counterpart focused on the first challenge, I started exploring many different UIs for goal selection and rapidly testing and iterating on them. Here are some of the early explorations:
We found out through user testing that advertisers preferred to see all the available goals at once. But presenting all the options without making the UI feel overwhelming was a challenge. We also had technical challenges to deal with. For example, within our four categories of goals, advertisers were only allowed to select goals from a single category. Furthermore, there were some options that were mutually-exclusive, even within the same category. Using subtle animations and other UI hints, we were able to convey which options were selectable at once.
Today, at the start of campaign creation, we ask advertisers to tell us about their goals in simple terms that everyone understands. Based on their selection, we tailor the rest of the process to highlight only the features and options that are most relevant.
Contributing to the Material Design specs
Campaign creation is a multi-step process, and we needed to indicate progress to users. To do this, we researched and designed our own stepper component. And since Google’s open culture encourages everyone to share designs and contribute to Material specs, we added it as a component to the Material Design library, helping other designers in and out of Google.
Making ads more beautiful (and effective)
Advertisers who advertise on the Display Network typically use visual banner ads. While big advertisers have the means to design and develop custom creatives, smaller advertisers sometimes use display ad templates provided by AdWords. The problem was that these templates were old and looked dated.
I really wanted to improve them, even though nobody had asked me to. This is one of the things I love the most about Google. If you see something that can be improved, you can make it better and you’re encouraged to lead.
I identified the team who had created the templates and reached out to them to tell them I was interested in improving them. They were glad to hear it. I updated 12 of the old templates and created several new ones.
In addition to the templates, the ad formats team were also working on a project that involved automatically converting text ads into rich, display ads. This was done by finding the advertisers’ logo, extracting relevant images from the advertisers’ other display ads, and then combining them with the messaging from their text ad. It was an amazing piece of technology.
There was a lot of engineering magic going on behind the scenes to extract these logos and images, but stitching them all together programmatically in a way that resulted in a beautiful creative was a challenge.
I volunteered to help. First, I specified in detail a framework for selecting the right colors, based on the existing colors of the logo in a way that would preserve the advertisers’ brand. Then, I created an extensive layout library based on the extracted image sizes and aspect ratios.
This helped developers know exactly how to lay out the creative in every possible situation and for all the major ad sizes. You can read more about richer text ads on the AdSense blog. With the performance of these new ads, this project also proved that good design can positively impact the performance of the ads and help advertisers get better results.
Helping advertisers learn about their audience
Most advertisers think they know their audience pretty well. A ski shop for example would naturally want to show their ads to skiers. But what if the people who visit their store also happen to travel more frequently than the general population? Knowing this information can help the advertiser reach more potential customers.
That’s what we did with Audience Insights—a tool that showed advertisers general information about their audience that they might have otherwise never guessed. The biggest UI challenge was distilling complex topics such as audience indices, which measure how one population compares to the general population. To make things easier to understand, I decided to use natural language and visualizations wherever possible. The result was a highly interactive tool that allowed advertisers to learn about their audience from different dimensions:
Let’s be honest; if you’re a UXer looking from the outside, working on ad products may not seem that sexy. Yet once you start working in ads you realize your incredible power to have impact as a designer. There are few areas that are as rich, complex, and multi-disciplinary as ads, and where there’s complexity, there’s opportunity for design to shine. I was fortunate to work with so many brilliant and interesting people during my time in AdWords. If you ever get a chance to work in ads, take it!