Who they are
WGT is a free online golf game played by millions of people around the world. It's built as an authentic golf experience rather than an arcade take on the sport, with over 20 real courses, head-to-head multiplayer, and licensed clubs and balls from Titleist, TaylorMade, Ping, Callaway and more, and runs across PC, Mac, iOS, Android and Amazon devices.
WGT runs user acquisition across multiple ad networks, with an in-house creative team producing its finished work.
A creative process that worked, with data spread across every network
WGT's creative engine runs on two inputs, and both still do today. They track which themes are gaining traction across the market and build their own versions to test. And they keep iterating on proven top performers to extend how long those keep working. It's a disciplined process, and Segwise hasn't replaced any of it.
The constraint was never the process. It was where the data lived. Performance numbers sat in the MMP. The live creative sat inside each ad network. Nothing joined the two automatically, so connecting a result back to the asset that produced it meant doing that join by hand, one network at a time. And because each view had to be built rather than reshaped, looking at the same data by concept, by network or by market usually meant starting a new data set.
That's a structural problem rather than a WGT one. Any team running creative across several networks alongside an MMP hits the same wall: the numbers and the assets never live in the same place. The cost is that good insights arrive later than they could.
“To identify top-performing assets, we had to pull data from our MMP and then manually cross-reference it against the live creative sitting in each individual ad network. There were no shortcuts.”
What changed with Segwise
Segwise connects to the ad networks and the MMP and does that join automatically, so creative performance shows up at the concept level and at the granular per-network level in one view, with nothing to rebuild each time a new question comes up. Before committing, the team pressure-tested how accurately the reporting held up across their particular mix of networks. It held up.
Creative generation slots in as another fast way to test concepts, upstream of production. As existing assets fatigue, it keeps a steady flow of new directions moving into rotation. When one shows stronger relative performance, WGT's in-house creative team picks up that signal and builds the fully produced version. The finished ads are still made by their team. Generation is the early filter that tells them which directions are worth the production time in the first place.
The early concept-to-test cycle is about 5x faster
Identifying a concept and getting it into a live, testable state used to take two to three weeks. Now it takes two to three days. That's the early cycle specifically, not full creative production, which still runs through the in-house team on its own timeline. The gain is that live performance data comes back sooner and sharpens the next round of internal creative direction.
They're testing more assets than before, and picking better ones, with less guesswork about which concept is likely to perform. ROAS remains WGT's north star for evaluating creative, and fatigue detection now sits alongside it as an early signal, one more input into rotation calls the team was already making well.