Crocs chooses data+Insight…omnicom makes learning faster…butler/till 6x experimentation.

This weeks ThursdAI is all about the importance of having a good lab. AI clearly (as you can see from this week’s big three stories) lets brands and agencies dramatically expand the amount of testing and experimentation they can do. But, more tests can be more noise for the analytics teams. And consumers are going to get weary quick of lots of brands trying lots of stuff. This week, we focus on the idea that in order to run experiments, you have to have a good lab.

Crocs awards business to Instrument, who showed creative blended with data, tech and automation

Evidently, the day after the pitch, Crocs was impressed enough by Instrument to hand them the business. They shone through not just with creative, or tech or automation, but showing how all three blended together into a single, integrated practice. The ability to take insights and directly feed them back into creative and production, let alone media buying, represents a real shift away from the one-direction model that agencies have been stuck in for generations.

Omnicom makes learning cheap and fast with Ideally

Ideally is a platform that promises to bring consumer insights directly into strategy and creative, and Omnicom seems to like it. They have run 60 tests across multiple categories, claiming to get research results back overnight, at a 1/10th the cost of traditional methods. This is the sort of ease and scale that will make testing a normal part of the agency process. We’re going from a few, expensive tests generating data that’s hard to utilize to a culture of testing always. But (again) with rules and methods and a governance model, we run this risk of over-testing everything (including consumer’s patience).

Butler/Till finds the guardrails of agency media experiments

This one is a kicker. Butler/Till has been running agentic tests for months now. They’ve reported good results, but they are also discovering that there is no single strategy for all clients. Each one has its unique needs and market mandates. They have begun to focus on creating guardrails, and that is the direction smart marketers are heading: humans need to set the parameters for AI systems to get smarter and better, in the right direction. That’s the essence of building a good lab.

Mover of the Week: Agencies with governance.

AI can do lots of things. But the smartest agencies, like the ones listed here, are not just turning things over to AI and celebrating the increase in margin. They are handing over tasks to take advantage of the smart scale, but keeping the human-in-the-loop, clearly directing toward the right goals. They are defining what an AI advertising agency is.

Thursdai index: this week’s numbers.

Experimentation Capability: 8.6

Learning Quality: 8.1

Laboratory Reliability: 7.9

Market Traction: 8.7

The Thursdai Index and report is the work of 61 Hawk founder Gary Stein with assistance from an OpenAI research agent. The agent sources, sorts and scores news stories. Everything you read was written by Gary, a real human being.