Automating Ad-Tech Practices
Accenture/Adaptly
Summary
Ben is a media operator. His job is to run advertising across social platforms, and he sometimes has to wake up in the middle of the night to turn ads on and off. Additionally, when figuring out his budget during shorter, custom ad flights, he uses an external spreadsheet to figure out his spend. I designed two features which automated these tasks so that media operators like Ben could free up their time for more skilled tasks. As the UX designer, I conducted user research, created prototypes, collaborated with product and the developers, and led the design of automation features from concept to delivery.
The Challenge
Our users were logging into the platform and manually updating the status of ad flights. They were using external spreadsheets to calculate spend for time periods within larger ad flights. The business was wasting money on these inefficiencies. The company was losing hours of productivity.
The Process
I organized a workshop with six of our media operators during which they ranked Ad Scheduling as their #1 most desired feature—more than twice the score of the next highest-ranking feature.
In addition, after attending a great Jared Spool workshop suggesting it, I instituted weekly shadowing sessions with our media operators to observe their use of the tool. I saw that our users were constantly leaving our tool to use external spreadsheets with custom formulas to manage their budget scheduling.
We were able to quickly test the scheduling feature and then run the pacing feature through 5 rounds of conceptual testing.
Automation: Ad Scheduling
Our media operators ran hundreds of campaigns, with thousands of ad sets. The ads needed to be turned on or off in the middle of the night, and sometimes in time zones different from those of our users.
Ad Scheduling gave our users the ability to define, ahead of time, the flights of their ads. They could sleep through the night, not work on the weekends, and be secure that their campaigns would run as planned, across multiple platforms, knowing that should anything go wrong, they would be alerted.
The majority of the design and development challenges lay in under-the-hood features: how our tool interfaced with the native platforms, and in the extensive logic around alerting.
Adoption
Platforms like Facebook offer native ad scheduling features, but many of our users actually preferred our solution, both because it’s more intuitive, and because it gave scheduling control down to the minute. And several of the other platforms don’t offer automation features, so this has been a game-changer.
“Within one hour, Ad Scheduling had already become a tremendous help with our scheduling and QA turnarounds. My account requires a significant amount of day-to-day post scheduling. Using Ad Scheduling now gives me back about 45 minutes per day. Unexpectedly, I'm using it as a QA tool deployments as well. It has been completely bug free and my team cannot imagine not using it in the future.”
- Media Operator
The complexity of this feature provided an exhilarating challenge, and I enjoyed working closely with my dev team to think through the many possible scenarios.
Automation: Pacing Segments
Defining the Problem
One main benefit which our tool offers our media operators is to provide guidance on how they should be spending their budgets, based on past performance and spend, and on time left in a campaign’s flight. However, a number of our accounts have different sub-flights which spend individual amounts (think Black Friday), and this made it impossible for our tool to know how to guide our users.
The top two sections are the main ad flight. The bottom section is the Subflight, or “heavy-up” period.
The Solution
To allow users to increase spending during specific sub-periods within a campaign, we initially conceptualized adding a sub-section to the existing flight. However, automatic spend allocation proved challenging. To address this, we iterated on the design, breaking down the flight into contiguous segments where users could manually input spending. This approach provided the necessary data for pacing guidance. In our final iteration, we further enhanced the feature by enabling users to copy and paste budget, date, and note information into external spreadsheets.
Results
In the first 2 months, 7,244 scheduled actions were taken, across 24 accounts on Facebook, Twitter, and Snapchat—4,914 were scheduled between 10pm and 6am. The feature saved each user approximately 6 hours per week, and saved the company $150K per year. Ben could sleep at night.
The budget segments tool was adopted by most of our teams and was well-received. It was a thrill to observe users making use of the feature during our weekly shadowing sessions, to watch sessions on FullStory, and to follow analytics in HEAP. We are seeing the full use of its functionality, and also some unintended benefits for QA.
As always in UX, it's illuminating to see surprising uses that users have come up with for the Pacing Segments tool, and also to see where I could have better communicated the feature's capabilities. In response, I started creating tutorials for new features, using a tool called Chameleon, which walks users step-by-step through every aspect of the new workflows.