How To Tell If A Collaboration Actually Worked

The sale that showed up four months late

Somebody buys on a Tuesday in July.

Your dashboard says she came from the sales email. Which is true, in the sense that the sales email is the last thing she clicked before she bought.

It's also almost entirely wrong.

Because she got on your list in March, through a summit… a podcast… a bundle. She read for four months. She replied to something in May. She clicked a blog post in June. And then in July an email happened to arrive on a day when the thing she'd been circling finally made sense.

The sales email closed it. The collaboration created it.

And if you're measuring on last click, you'll conclude that sales emails work and collaborations don't, which is how people end up quietly killing the exact thing that was building their business.

So let's talk about what to measure instead, and how to actually do it without building a data science practice you don't want.

Why doesn't subscriber count tell you if a collaboration worked?

Because it measures how many people arrived, not how many stayed, engaged, or bought. It's the fastest number available and the least predictive one.

Two hundred subscribers is not a result. It's a raw material with a wide range of possible outcomes attached.

Two hundred people who are all gone by week three is a different event than two hundred people where sixty are still opening at ninety days. Same headline number. Completely different business.

The subscriber count gets treated as the score because it's the only number that shows up immediately, and because it's the one the host reports back to you. It's a starting position, not an outcome.

How do you attribute a sale to a collaboration months later?

Tag the source permanently at the moment they subscribe, and never overwrite it. Then look at buyer lists by source tag instead of by last click.

This is the whole mechanic. It's not complicated, it just has to be set up before you need it, which is why almost nobody has it.

Three rules.

One: the source tag is permanent. It records how someone entered your world, and it never changes, ever. Not when they download something else. Not when they buy. That tag is a historical fact, and it stays.

Two: source tags are separate from behavior tags. Where they came from, what they're interested in, and what they've bought are three different questions. If you're using one tag field for all three, you're overwriting your own history every time someone does anything.

Three: tag by specific collaboration, not by category. Not "summit." The actual event, with the actual date. Two summits six months apart can produce wildly different results, and if they share a tag you'll never know which one was worth repeating.

Once that exists, attribution stops being a mystery. You pull your buyer list for a launch, cross-reference source tags, and the answer is just sitting there. Fourteen of these forty-one buyers came in through one collaboration in March. That's the number. That's what it produced.

You cannot reconstruct this after the fact. There's no way to go back and figure out where someone came from once the moment has passed. Which is the argument for setting it up before the next yes rather than after.

What should you actually track after a collaboration?

Four things: revenue by source tag, engagement rate against your list average, survival rate at sixty days, and relationship outcomes.

Revenue by source tag, at 30 / 60 / 90 days. The direct answer. Check it three times rather than once, because the shape of the curve tells you something the final number doesn't. A segment that produces nothing at thirty and a lot at ninety is behaving normally. A segment that produces a little at thirty and then flatlines is telling you something different.

Engagement rate against your list average. Open and click rate for that segment, compared to your list overall. This is your leading indicator, and it's the most underrated number in the set. It tells you at thirty days what revenue is going to tell you at ninety. A segment engaging well above average is worth investing more nurture into. A segment engaging at a third of your average is not going to surprise you later.

Survival rate at sixty days. How many of them are still opening anything. This is the number that turns a headline into a real figure, and it's the input for the metric in the next section.

Relationship outcomes. Re-invites, referrals, introductions, contributors who became collaborators. There is no dashboard for this and it's frequently the largest thing a collaboration produces. Keep a running note. Not a spreadsheet, just a note. When you're deciding whether to do the thing again next year, that note tends to be more persuasive than any of the numbers above it.

What's the one number most people are missing?

Cost per engaged subscriber. Not cost per subscriber. The denominator is the people still opening at sixty days, not the people who arrived.

This is the metric I'd put in front of all the others if I had to pick one.

Add up what the collaboration actually cost you. Then divide by the number of people from that source who are still opening your emails at sixty days.

Not by the number who came in. That number includes everyone who unsubscribed in week one, and including them makes a collaboration that delivered volume look identical to one that delivered people.

The reordering this produces is frequently dramatic. The collaboration that brought in twelve hundred subscribers and the one that brought in three hundred can end up in the opposite order once you're dividing by who's actually there. And the second one probably took a quarter of the effort.

Run this across three or four collaborations and you'll have something better than any benchmark: your own pattern, for your business, at your price points. Which formats deliver people versus names. Which hosts send audiences that stay. That's the thing worth knowing, and nobody else can hand it to you.

What does a collaboration actually cost you?

Prep hours, promotion hours, the attention you spent from your own list, and whatever you didn't promote that week. Most people count zero of the last two.

ROI has two halves and the cost half gets skipped almost universally, mostly because collaborations feel free. No invoice arrives, so it registers as costless.

What it actually costs:

Your hours. Pitching, prepping, recording, creating the contribution, the follow-up. Put a real number on it, even a rough one.

Your list's attention. If you promoted the thing to your own audience, you spent something real. Attention is finite and you used some of it on somebody else's event.

The thing you didn't do. The offer you didn't promote that week. The launch you pushed. That's the largest cost in most cases and the one that never appears anywhere.

None of this is an argument against collaborating. It's an argument for knowing the actual figure, so that when you're comparing two of them, you're comparing complete pictures rather than two half-filled columns.

How long should you wait before deciding?

Ninety days, decided in advance, checked at thirty and sixty on the way.

Ninety because collaboration leads arrive further out than direct opt-ins and need time to close the distance. Thirty and sixty on the way because you want the shape of the curve, not just the endpoint.

The part that matters most: decide the window before the collaboration starts.

A measurement window set in advance is a completely different instrument than one you pick after you've seen the numbers. Set it beforehand and you get information. Set it afterward and you're just choosing the window that confirms what you already suspect.

What if it turns out it didn't work?

Then you learned which format, host, or audience isn't producing for you, which is worth roughly as much as a yes.

A no is a real result. It's most of what makes the next yes better.

Three questions worth asking before you conclude the collaboration was the problem.

Was the job named beforehand? A collaboration taken for relationship value and then judged on revenue was set up to fail a test it never agreed to take. Bundles get this treatment more than anything else.

Was there anything built to receive them? If people arrived and landed in a list that gets emailed occasionally, the collaboration did its part. The gap was downstream, and joining another one won't close it.

Was that audience actually yours? Sometimes the answer is a good host with a real audience who are simply not the people your offer is for. Nothing was done wrong. It was just a mismatch, and the useful move is noticing it in time to not repeat it.

The part I actually care about

Most people are running collaborations blind and then making strategy decisions off a feeling.

The feeling is usually formed in the first week, off the fastest and least useful number available, at the exact moment a normal unsubscribe wave is happening. That's a terrible time to be deciding anything.

Ninety days, four numbers, one permanent tag. That's the entire system, and it turns a year of collaborations from a series of hopeful shots into something you can actually steer.

You already did the hard part. This is just refusing to throw away what it taught you.

Next
Next

Are Bundles Worth It? What Bundle Subscribers Actually Do Next