A funnel is a question, not a report

A funnel is worth building when you have a specific suspicion. You think people are dropping out between the basket and the delivery step, and you want to know whether you are right.

It stops being worth much when it is built because the tool offers funnels. Those ones have seven steps, get looked at twice, and sit at the bottom of a menu for three years.

The useful discipline is to write the question down before you build anything. “Are we losing people at delivery options, and is it worse on mobile?” That question has an obvious three-step funnel behind it and an obvious thing to do with the answer.

If you cannot write the sentence, you do not want a funnel yet. You want to go and watch somebody use the thing.