DEATH STAR
As an extension of Essence's services, the Client Solutions team built marketing ops tools in collaboration with Google Media Lab.
Death Star, a "campaign best practices monitoring and alerting" was a legacy tool and the flagship of the relationship.
But despite Death Star's monitoring, and despite its alerts, the number of best practices issues refused to abate.
*Note: I don't have as many product visuals from this time, but Death Star wasn't pretty anyway.
PROJECT INFO
Essence was Google’s digital media agency of record and managed the plurality of Google's digital ad expenditure.
Death Star was important, because as one Google Marketing Leader put it:
"One bad impression could cost the brand."
GOALS
Our sole remit from Essence leadership was to make Google Media Lab happy.
We figured, and Media Lab agreed, that driving down campaign violations would do that.
TEAM
Hans C. - Product Manager
Me - UX Designer
Brian T. - Back-End Engineer
I handled user research and design, while assisting Hans, who managed everything else. He taught me to do things cheaply, incrementally, and with intent.
Company:
Essence Global
Services:
UX Design, Client Services/Consulting
Team:
Me – UX Designer
Hans C. – Product Manager
Brian T. – BackendEngineer
Hans and me at my wedding.
Still my boy.
DISCOVERY
Talking to other agencies was a political thing above my position and pay grade, so Essence was my only source of information.
My co-workers sat in the same open floor plan as I did.
I’d watch them work, ask them questions, and observe how they came to know and resolve their best practice violations.
GMAIL AS SISYPHUS'S BOULDER
They mostly didn’t see or think about best practice issues.
They received 100s of emails a day – Death Star’s came once a week and looked like spam.
ESSENCE BEHAVED
But when asked, all my fellow “Essentials” cared very much about adhering to best practices.
And it was true. The vast majority of Essence's campaigns had one issue, and typically, it was minor.
They had a checklist, and people managers at Essence, whose performance was based in-part on their Death Star compliance would allocate periods to clear Death Star "debt."
Death Star
A seldom-interacted interface.
MISBEHAVING RIVALS
In contrast, other agencies’ campaigns had 2 or more issues and would remain delinquent until the campaigns ended.
Sometimes that took years.
A GIANT .CSV
I also heard that the massive, unfiltered .csv that Death Star made it difficult to find one’s own best practice violations.
But I figured that much.
Questions on questions on questions.
TWEAKING SETTINGS
We had a few parameters to tweak to bring violations down.
The first was tweaking notification frequency.
But increasing it didn’t accomplish much.
Next it was changing the notification subject line.
But people still didn’t notice.
Tweaking the violation thresholds (?)
Dicey. And without data to suggest no harm would come to the brand as a result, this would be artificial.
HANS & KRIS - THE MECHANICAL TURKS
Death Star was a few years old by the time Hans and I came to manage it and it was a bit musty under the hood, too, hence the massive .csv output.
So, every week, we’d alter the Death Star .csv by hand so that it was more legible for Essentials and agencies, while Brian cleaned up the back-end.
The important take away, is that these best practices didn’t matter to anyone as much as they mattered to Media Lab.
What is the autonomy of a useful alert?
DEATH STAR
"Let's just do it for them."
A Story told through KPIs.
AUTO REMEDIATION & CAMPAIGN PAUSING
Death Star monitors campaigns and fixes violations for busy campaign managers.
Any harmful campaigns Death Star couldn't fix, Media Lab could shut off.
HOW WE GOT THERE
CAMPAIGN NOMENCLATURE
The trouble with auto-remediation is in understanding what is “correct” in context.
Fortunately, when naming a campaign, managers followed a template that encodes info about the campaign’s settings.
Leveraging the “Campaign Nomenclature” best practice allowed Death Star to identify discrepancies between campaign settings and campaign manager intent.
KEYWORD LISTS
A high priority violation-type involved bidding on keywords, trademarked terms, and website placements that could cause negative consumer sentiment or legal issues for Google.
These “negative lists” could be directly appended onto the campaign, which would prevent ads from appearing against vulgar or controversial keywords and content.
Card sorts helped identify most severe best practice violations.
RESTRAINT
To understand this next point, it helps to know where I was at the time.
Essence was my first “real” job.
It took me about 2 and 1/2 years from the time I decided to be a UX designer to finally getting interviewed to be one.
Initially, Essence rejected me, as they thought me too junior.
But for some reason, the other person ultimately didn’t workout, and Essence emailed me two weeks later to see if I was still available.
After a final interview with the department head, I received an offer a week later.
I was so happy, I cried.
The thing was though was that this was during the peak of theDribbble/Behance days.
All the cool designers had portfolios featuring slick mobile UIs and horizontal (that’s right) scrolling.
I wanted to be one of them so badly, but a Death Star can only shine so bright…
So, when it came time to create our pause feature, we needed a way to turn off campaigns in bulk.
To this point, most of the changes I'd made to Death Star had been edits to existing features: that's most of the work when you're handed a legacy system.
Finally, I thought, I'd have an opportunity to have at least something to show.
But despite how badly I wanted to "design" something, nobody wanted to use Death Star.
I knew that.
Hypothetical "Campaign Manager" Interface - Paper Prototype
Wasn't meant to be.
Marketers do use a lot of Microsoft Excel, though.
Other than their ad platform of choice, they spend their entire day in Excel.
Rather than take them out of their flow, I instead gave them an “Upload .csv” button, and a formatting template and that was enough.
CONCLUSION
Death Star would achieve improve overall compliance by magnitudes in the thousands, and between auto-remediation and campaign pausing, Hans, Brian,and I kind of automated away our jobs.
We got a few smaller projects in the meantime to keep us occupied.
THE NEXT PHASE
Death Star eventually caught the attention of Essence’s parent company, Group M.
They in turn, in partnership with Alphabet, invested 5 million $USD to extend Death Star’s capabilities to other platforms and clients.
Hans and I would spend the next year working out the product and UX strategy for project Starkiller, before leaving to join Triplelift.