The usual way to judge a data subscription is to switch it off and see what breaks. Do that for each feed in turn, rank the feeds by the damage, and cancel the ones at the bottom. It sounds rigorous. When feeds overlap, it gives the wrong answer.
A toy example
Three feeds inform a set of decisions. Feeds A and B carry much the same information; feed C carries something neither of the others has. Score the decisions at 100 with all three in place, then drop one feed at a time.
| Feeds kept | Decision score | Loss from dropping |
|---|---|---|
| A, B and C | 100 | – |
| B and C (drop A) | 98 | 2 |
| A and C (drop B) | 98 | 2 |
| A and B (drop C) | 80 | 20 |
| C only (drop A and B) | 60 | 40 |
One at a time, A and B each look nearly worthless: two points apiece. Cancel both on that evidence and the score falls by forty, twice what C was worth. Each feed was covering for the other, so neither could show its value alone.
What to do instead
Weigh feeds together, so that what two of them share is counted once and credited fairly, rather than credited to neither. Then decide renewals on that. The cheaper of two overlapping feeds is usually the one to keep.
Numbers here are illustrative. This is the question Tare answers on your own feeds.