Excellent analysis! The shift from last-click delusion to true incrementality feels like a fundamental change we should have embracd ages ago. It makes me wonder, what if we could fully automtate the geo-lifting or budget holdout methods with advanced AI, constantly fine-tuning for optimal spend across regions, rather than just A/B tests?
The math is easy to automate, but the context isn't. An AI can shift budget around, but it won't know if a competitor just flooded Nairobi with OOH or if a payment gateway went down in Lagos. You’d end up with a machine perfectly optimising for data that’s been skewed by the real world. We’ll get better tools for the heavy lifting, but you still need a human to look at the numbers and see if they actually make sense 😊
The nuance is: it’s not just “turn ads off”, it’s where cannibalization hides. Branded search and retargeting love taking credit for conversions that were already inevitable 😅
The sneaky mechanism is trust latency. Someone sees three touches over a week, then the last click gets crowned “the winner”… even though the decision formed days earlier.
One move that saves a lot of budget pain: keep prospecting on, then selectively hold out brand/retargeting in a clean pocket. If total conversions barely move but CAC improves, you just found free efficiency ✅👌
It’s the classic attribution trap. Paying a commission to Google for people who were already standing at the door. The 'trust latency' bit is also why so many teams accidentally kill their pipeline. They cut the top-of-funnel and then wonder why conversions crater two weeks later. You’re spot on about the CAC move, too. If you can hold out retargeting and the needle doesn't move, you've just stopped subsidising sales that were going to happen anyway
Excellent analysis! The shift from last-click delusion to true incrementality feels like a fundamental change we should have embracd ages ago. It makes me wonder, what if we could fully automtate the geo-lifting or budget holdout methods with advanced AI, constantly fine-tuning for optimal spend across regions, rather than just A/B tests?
The math is easy to automate, but the context isn't. An AI can shift budget around, but it won't know if a competitor just flooded Nairobi with OOH or if a payment gateway went down in Lagos. You’d end up with a machine perfectly optimising for data that’s been skewed by the real world. We’ll get better tools for the heavy lifting, but you still need a human to look at the numbers and see if they actually make sense 😊
Big yes ✅😄 Incrementality is the truth serum.
The nuance is: it’s not just “turn ads off”, it’s where cannibalization hides. Branded search and retargeting love taking credit for conversions that were already inevitable 😅
The sneaky mechanism is trust latency. Someone sees three touches over a week, then the last click gets crowned “the winner”… even though the decision formed days earlier.
One move that saves a lot of budget pain: keep prospecting on, then selectively hold out brand/retargeting in a clean pocket. If total conversions barely move but CAC improves, you just found free efficiency ✅👌
It’s the classic attribution trap. Paying a commission to Google for people who were already standing at the door. The 'trust latency' bit is also why so many teams accidentally kill their pipeline. They cut the top-of-funnel and then wonder why conversions crater two weeks later. You’re spot on about the CAC move, too. If you can hold out retargeting and the needle doesn't move, you've just stopped subsidising sales that were going to happen anyway