July 14, 2018
The Trade Desk recently released a platform-wide overhaul dubbed “The Next Wave”. The overall theme of the update was a leveraging AI to take advantage of historical data, both from each advertiser and across the entire platform. I absolutely love the ideas and direction TTD is heading. They’ve embraced automation and machine learning. They’ve fixed campaign logic that was previously confusing. They’ve enabled advertisers to apply advertiser-specific learnings before spending a single dollar. That being said, the execution was definitely not flawless and some features are opaque and poorly documented. I’ve been using the new features for two weeks — enough to get a taste for how they function but not enough time to truly understand what they’re capable of.
Koa is powerful AI that improves advertisers’ decisioning and accelerates campaign performance. Koa’s robust and transparent forecasting engine is built on The Trade Desk’s valuable data set — including nearly nine million queries every second — to help buyers extend audience reach and spend more efficiently.
Of the three new features, Koa is the most disappointing. While only two weeks in, it hasn’t lived up to the marketing hype. TTD previously had a feature called auto-optimizations that accomplished the same thing. It would optimize certain variables (site, vendor, ad format, etc) to your given goal (cpa, viewability, etc) while still allowing your campaign to deliver in full. Koa’s branding made it appear as though auto-optimizations would be massively improved. So far this doesn’t seem to be the case. Koa is nowhere near as active as I expected, nor does it optimize enough variables. In two weeks Koa has optimized site lists 3 times, reducing bids on ~20 sites. It also made some ad format adjustments. While nice to have Koa optimize in the background, it doesn’t appear to optimize often enough or aggressive enough.
There is a distinct lack of documentation on how Koa functions. I’d like to learn how, when, and why Koa optimizes. Why does it only touch my ad group every few days? Why can’t it adjust my bids? Why can’t it move budget between ad groups? Why can’t it look at historical advertiser or platform wide data to make quicker decisions? While AI and automation are undoubtedly the future, Koa seems timid and ineffective today.
The Trade Desk Planner is a data-driven media planning tool that delivers audience insights and informs ad strategies across channels and devices.
I was lucky enough to get a preview of the planner a few months in advance and was excited to see it get released. The planner simply analyzes data from a pixel you’ve placed on your site and builds a media plan around that data. This means your campaign starts out with strong learnings before even spending a dollar. For example, a normal campaign would start out with relatively equal spend between mobile and desktop. But a planner campaign has pre-built optimizations for device types and operating systems that match previous site visitors, so the advertising will target valuable users right out of the gate. While sales and account managers can get a nice overview of demographics, devices, media mix, etc the true value is reaped by campaign managers. The planner can build a media plan and then with one click convert into a fully built out campaign.
I am testing one of these pre-built campaigns but I always like to custom build everything from the ground up. But in the future I will likely still create a plan and generate a campaign because the tool builds huge bid optimization lists. The campaign I generated had ~100 bid modifiers down to the county level and even built a whitelist ad group of extremely relevant sites. My old hand-built campaign had only 30 geo bid optimizations at the state level, and a 3 site whitelist that took thousands of dollars of spend to identify the sites.
This is the most valuable feature I’ve ever seen in any marketing platform. Planner can provide value to everyone in our organization. While not implemented yet, I could see sales using planner to find avails, account managers using it to identify pre-sale insights, campaign managers to streamline campaign builds, and analysts for reporting. Plans can also be exported to csv or powerpoint for quickly grabbing collateral and visualizations.
Megagon is an intuitive new user interface that proactively surfaces tailored insights and offers Koa Recommendations to help advertisers make real-time optimization decisions. Megagon helps buyers save time and advertising budget without sacrificing transparency and control.
The name Megagon sucks, no doubt about it. I’m going to call it ‘mega’ for short. I’m also going to reserve most of my review of mega until I’ve spent more time in it. I’m only running a couple campaigns with the mega interface right now. Initially it feels slightly slower — adding something like viewability takes the same amount of clicks but there is longer animation and load times, so it feels less efficient. The tile layout looks cleaner and simpler, but hides data. Previously you could see which devices you were bidding on without clicking into that specific area, but now you have to click into a tile to see what it contains. Its no longer possible to easily grok an ad group. You are forced to click into multiple tiles before understanding what the ad group is doing.
Design changes aside, the new logic makes much more sense. Previously, setting up a specific-purpose ad group (like CTV only, or mobile in-app only) was not straightforward and error prone. With mega the logic is updated so you can effectively say “Only bid on CTV”. Before mega the logic was closer to “Bid at 1x on CTV, 0x on mobile, tablet, and pc. Also don’t bid if you don’t know what the device is”.
While still a first impression, I’m very pleased with the next wave update. Koa doesn’t seem to do much, but planner will be incredibly useful, and mega will reduce potential mistakes. I’m happy with the direction The Trade Desk is headed and think they’re making smart, strategic moves. The launch wasn’t quite flawless (planner was down for almost a week) and I would’ve done some things differently, but I think the team at TTD should be very proud of what they’ve accomplished.