|
|
|
|
|
by ruhdgjns
1701 days ago
|
|
It goes hand in hand. First, if you can’t measure conversions and tie them back to advertising, then how can you know if it’s effective?
Second, if you can’t measure effectiveness how to A/B test your approach? Finally, without labeled data how you train ML models? The whole house of cards stood upon attribution. |
|
I don't think Coca Cola needed attribution to know their ads were effective in 1900, maybe we should ask them?
> Second, if you can’t measure effectiveness how to A/B test your approach?
You can still use either time, geographical or UA-based AB testing without requiring attribution
> Finally, without labeled data how you train ML models?
Are you saying the final objective of tracking is actually to feed AGI as most data about our world as possible? That's an interesting point of view, I never thought about ML modeling as an end-goal.