Design for a Strange New World: (Re)designing Measurement
My favorite part of researching for Facebook's ad tech products is learning about our clients: advertising agencies and large advertisers. I never worked in advertising before coming to Facebook, and my research had largely been with consumer communication products. Studying our clients was like visiting a foreign country where I didn't speak the language and didn't understand the local customs. I observed some familiar behaviors and tools, but I also observed many things that made little sense at first.
It seemed like every client or advertiser I talked to had a strange job title. Examples: Ad Ops Trafficker (is that legal?), Media Technology Specialist (do they work with AV equipment?), Digital Activation Director (no idea). I had to look up words that sounded familiar but didn't mean anything to me like placements, channels, attribution model, and endemic media. And I didn't understand why everyone said they valued careful work, but then held very different notions of what that meant.
It dawned on me that I really was studying a different, foreign culture. If I wanted to help our clients do their jobs better, I would need to learn more about their goals and motivations and the ways they actually worked, and not assume they behaved like any other office worker.
Weight of the world
Much of what we've learned about advertising agencies and their motivations was applied in the redesign of Atlas, an ad measurement platform that was relaunched last year to enable users to leverage the power of Facebook's identity-based targeting. In the last few years leading up to the relaunch, the classic version of Atlas hadn't received key design updates. Simple tasks sometimes required many clicks, and pages often took too long to load. As noted in a previous Elegant Tools post, in the work context, efficiency of completing tasks is highly valued, and certain aspects of Atlas frustrated users:
“This is hands-down the most time-consuming process and the least efficient thing that I do in my life. I am confident with that statement. There is no doubt in my mind that is true. People have wasted years of their lives doing this.”
- Client of classic Atlas, on a specific feature of the tool
When I was researching social communication tools for consumers, study participants would often say something was “cool” or “confusing,” or that they “liked” or “didn't like” a widget. Now and then they “loved” a product and shared an emotional story about how it connected them with loved ones.
But working on business design was the first time I encountered participants who spoke with passion and in detail about the tools with which they worked. Unlike consumers, professionals are often constrained in the tools they can use for work. Using these tools regularly, they have a lot of opportunity to ruminate on everything they love and hate about these products. The smallest details can become major delighters or burdens with repeated use, day in and out.
We knew that even the smallest design improvements would make a big difference. But we didn't just want a more streamlined product; we wanted to make our clients' workflows more efficient and effective.
Understanding how clients work
When you are redesigning a complex product from the ground up, there are so many opportunities for improvement: it’s temping to dive right in to making simple, obvious design changes. We realized, though, that we first needed to understand why our clients were trying to use Atlas. We wanted to create a breakthrough redesign that dramatically improved our clients' effectiveness at their job and help them focus on the most important tasks. To do that, we had to understand what they were trying to accomplish with their business and in their jobs and look beyond the steps they were taking to conform to the tool. One of our first Atlas product designers says that without understanding client motivations, “I might have made something super-efficient, but it might not have added value to their work.” We needed to understand the context that our clients operated in. Another product designer added that, “We looked outside the tool [Atlas], and how clients used spreadsheets, so that we didn’t make something isolated” from their existing workflows.
This approach yielded three key insights.
We support teams of users working toward business goals.
One of the most fundamental insights we had about Atlas was that it needed to support collaborative workflows. Advertising is a highly relationship-driven environment with complex motivations, and a solo user wouldn’t have a purpose for a tool like Atlas. Within an agency, each user contributes work that helps their client achieve a business objective, such as boosting clothing sales on an e-commerce site. A user might be part of a media planning team that decides on the right mix of publications to advertise in. That team then passes the media plan on to a trafficking team to implement in the tool. Then, as the ad campaign runs its course, analysts may step in and work with the media planning team to evaluate the success of the media plan at driving purchases of clothes. Understanding this dynamic — and facilitating easier, less redundant handoffs between teams — was important for a great redesign.
Team members have their own goals, too.
At the same time, individual users have their own objectives in how they support the advertiser. A media planner’s top priority might be getting the most value from her media budget, while a trafficker wants to make sure he has implemented the campaign exactly to spec, with no mistakes, and an analyst wants to be able to tell a clear story about how the campaign performed. To support the advertiser’s overarching business objective, Atlas’ product UI had to help all of these individuals meet their own goals.
Ideals aren’t practices.
When you interview a client about an ad campaign, they often tell you a neat and tidy story about how many people work together to launch it. But when you probe a little deeper, you hear where things break down. One of the most powerful kinds of user research is distinguishing between the cultural ideals and cultural practices of our clients.
A cultural ideal is what we aspire to. A cultural practice is the messy reality. One common cultural ideal is that every team member is fully aware of what all the others are doing to further the campaign, and why. The cultural practice may be that each team member sees only her own slice of the work. Another cultural ideal may be that the trafficker has all the materials he needs before he sits down to set up a campaign in Atlas. The cultural practice is more likely to be that he has to chase down files and information from various colleagues, doing his campaign setup work piecemeal. To support efficient use of Atlas, we needed to design something flexible for the messy reality our users operate in, not a strict workflow for an imaginary best-case scenario.
Designing for real users
We wanted to make sure that everything in the new Atlas served the real people who were using it. Often in complex tools, we see too many features and functions exposed all at once to serve “experts,” but that approach often makes the product unusable, even to these intended pros. When we started our design investigations, we studied our various user roles. One of the most important was the trafficker, who does the bulk of the setup work for a new campaign in Atlas and therefore spends the most time in the tool. For traffickers, every wasted second adds up. Some time-wasters, such as slow-loading pages, were technical and could be addressed through code improvements. Others could be visually redesigned, such as unnecessarily complex forms.
Improving form design
Here’s an example of the form for creating a media buy in the old version of Atlas. (This would be used, for example, to place an order with CNN.com to show an ad for a clothing sale in the banner ad spaces on all articles about fashion and style.) There were many, many fields that could be filled out. The form was overwhelming.
Simplifying this form would also help another problem. Our advertising agency users love and live in Microsoft Excel. They did as much of their media plan setup as they could in an Excel spreadsheet and imported it into classic Atlas. While importing from Excel was a boon for large campaigns, it was an extra step for small campaigns. We felt client productivity would improve with better data-entry forms. A user experience researcher who studied agency and large-advertiser clients for the redesign of Atlas described the problem: “People loved the import tool [in Atlas]. How can you make people who live in Excel actually start loving Atlas itself?” Redesigning the forms would encourage clients to use the Atlas interface directly.
In the redesign of Atlas, we simplified the media buy form by splitting it into two pages and including only the most essential fields. We divided the form into one for the details of the placement (such as the ad dimensions to be supported and any special tracking codes that should be appended) and one to describe cost packages (the negotiated price). That let our trafficker approach the work in chunks. For example, if he knows the ad sizes, but not the prices, he can still enter one piece of information and pick up later when he has the remaining information. This design fit his reality, in which he might not have all the information he wants when he starts working.
Making forms more efficient helped our trafficker user roles be more effective at their jobs. The redesign of Atlas also considered other ways to simplify and reduce work.
Attacking redundancy
We wanted to serve users by introducing efficiencies to the number of objects they worked with in Atlas. Traffickers in particular value accurate work. They will always spend the time to double-check their work. We knew that no workflow we put in would make them let go of that instinct, so reducing the number of objects to check would have a lot of impact.
When I started on Atlas, a designer colleague asked for some input on the “assignment page for placements.” I remember thinking, what is a placement, and why would you want to assign something to it? Since those early days, I’ve learned that ads, creatives, and placements are examples of objects in Atlas. Display ads on the web are a combination of a creative image (such as a JPG of a red dress with the word SALE) and a clickthrough URL pointing to the e-commerce site or other relevant landing page. Ads must be assigned to the placements or the ad spaces purchased on publishers’ sites, such as CNN.com or nytimes.com, where they are expected to appear.
Creative image files are especially time-consuming to work with. Traffickers spend a lot of their work time QAing creative files to make sure they are the right sizes to fit ad standards, are good resolution, and actually match the description in the campaign brief (picture of a dress and not a cat).
In the classic version of Atlas, traffickers had to upload creative images in all required sizes, and then make separate ads for each size creative required, even though, for many campaigns, the creative images were essentially the same except for size.
The new version of Atlas lets users bundle together creative images for ads that are conceptually similar but in different sizes or file formats (SWF, JPG, etc.). We call these objects “creative concepts” because they pin together related creative files of a single concept, such as all images of a red dress with the word SALE. The result is that the trafficker works with one creative concept instead of multiple, individual creative files. It simplifies the creation of an ad — he can now create one ad of multiple sizes instead of multiple ads. It reduces as well the number of ads that need to be assigned to placements.
Editing the campaign is also more efficient. If the advertiser wants to try images of blue dresses instead of red, the trafficker would need only to swap out one creative concept for another. And the creative concept can be reused for multiple campaigns, saving on the QA work.
The trafficker’s colleagues also benefit. A media planner can check in one place to see if all the sizes of the red dress ad have been accounted for and uploaded. The analyst can more effectively run reports on how well the ads featuring the red dress performed, regardless of the ad size. And finally, other traffickers can benefit by reusing the creative concepts for similar ad campaigns.
This particular change has been more challenging for clients to adopt, in part because they don’t have prior experience in which creative files are pinned together into a single concept. Although all clients have recognized the efficiencies of grouping creative files, some may not truly understand why we want them to do that. It’s not intuitively clear what they are expected to group together into a creative concept, so it may be easier to simply ignore it and stick with the old method. We feel a creative concept serves a useful purpose and makes campaign work more efficient, and we need to make sure the clients see the value, too. Great product design for businesses doesn’t just stop at the UI, it continues with education and client support.
Still listening and learning
The Atlas redesign we debuted last year was far from complete; there were and still are more features we want to add. At the same time, we are taking feedback from our customers who are using the new Atlas to make improvements to the parts of the design that didn’t land as well as we had expected. I look at this as one of the most interesting challenges in building enterprise software. We still have a lot to learn about how teams of people interact in and around Atlas to get their work done. We study as many clients as we can, and we identify patterns in behavior, but the truth is there are so many ways for teams to work together, we could never capture every single possibility in our research. We design based on the information we have, and we iterate as we get new information.
My first few months working on ad tech were a steep learning curve. Then at some point, something clicked, and things began making sense. I understood what our clients were saying, what they were trying to do, and then what we, the design team, could do to make their lives better. And I realized how much our clients were stuck with our design choices, good and bad. Working on tools in a complex domain like advertising gives all researchers and designers the opportunity to have fresh eyes and see tough new problems. It gives us the chance to stretch our minds and sharpen our skills. This challenging work has helped me be much more empathetic to users, a more vocal advocate for their needs, and a better user experience professional.
Thank you to Radhika Bhalla, Lucy Davis, David Hayward, Tom Hobbs, Courtney Kaplan, Tina Santiago Keenan, Frank Marquardt, Ed Salvana, and Margaret Stewart for input, edits, and feedback on drafts of this post.
