[esafak]

You could seek permission from users to let you use their data to build a model to provide guidance.

[bohlenlabs]

Build a model? In the sense of machine learning or in another sense? Cool idea, what could that look like?

[esafak]

Yes, an ML model. If you built the machinery to connect marketing actions to business outcomes, as the OP complains about, you could recommend what actions to take. It's a straightforward machine learning task (supervised learning).

[bohlenlabs]

Cool thing. Guess I shouldn't reinvent the wheel here. Is there an open source example that I could learn from?

[esafak]

I don't think you will get far without knowing some ML but you could look at https://github.com/google/lightweight_mmm

[bohlenlabs]

Ah, there is food for thought and learning, here. Thanks for sharing, I will try to understand what's going on there.