Collaborations that compound

Most collabs do not compound. They register as a spike on both sides' analytics for 48 hours, then settle back to baseline, net nothing. The reason is almost always the same: the two audiences overlap too much, and the collab reshuffles existing viewers without introducing new ones. A compound collab is picked for audience overlap percentage, not reach. Done right, both sides net new paying subscribers that stick 60+ days.

Pick for overlap, not reach

Reach is the vanity metric for collabs. Overlap is the operating metric. The sweet spot:

Overlap below 15%: the audiences barely know each other. Introductions require more context, but when they land, both sides net new followers.
Overlap 20% to 40%: the ideal zone. Enough shared language that the collab feels natural, enough unexplored audience on each side to grow into.
Overlap above 60%: you are just reshuffling the same audience. Both sides see a short spike and no durable gain.

How we measure overlap:

Follower sampling. Pull 500 random followers from each account; compute mutual follows.
Tool-based estimation (SparkToro, Modash, or similar) for a cleaner read.
Content audit. If both creators cover the same five topics, treat overlap as 50%+ regardless of what the follower list says.

The compound test

Before saying yes to a collab, run three checks:

Will both sides gain followers who stick 60+ days? Not day-1 followers. Day-60 followers. If the content format does not give the new follower a reason to stay beyond the collab piece, it does not compound.
Will both sides see paid conversion lift, not just views? A collab that spikes views without touching revenue is a wash.
Will both sides come out with content they can keep posting? A one-shot collab locks the gain to a single piece. A format you can rerun unlocks compounding.

If any of the three is no, downgrade the collab from a format partnership to a one-off shoutout.

Formats that compound

Joint livestream with a shared offer. Both audiences see the same drop simultaneously. The offer gives new viewers a reason to signup inside the window.
Reverse POV series. Same event, each creator posts their angle. Viewers follow both to get the full picture.
Interview series. Each guest is also an audience bridge. Repeat the format with new guests; the format itself becomes a growth loop.
Co-produced course or guide. The highest-ceiling format. Durable asset, split revenue, mutual promotion for months.

Formats that cannibalise

Shoutout trades. Short-term vanity. The followers who move across often churn within two weeks.
Hostile dueting. Engagement spikes; trust drops. The new followers are rubberneckers who do not buy.
Pay-for-promo without story relevance. The audience reads it as an ad and discounts both creators.

Measuring whether it compounded

Three numbers to track after a collab:

Follower retention curve at 30 / 60 / 90 days. The gain you keep, not the gain you had.
Paid conversion lift attributable to the collab period. Revenue, not vanity.
Second-order content unlocked. Did the format spawn a repeatable series, a course, a new segment?

If all three move, you have a compound collab worth repeating with a new partner. If only the first moves, you have a vanity collab. If none move, you have a lesson.

Collaborations are leverage. Pick for overlap, run the test, and the format does the rest.