In the Agentic PMM Field Guide, in the section on measuring PMM success, I wrote this line: “A PMM who produces a lot of the wrong things is failing, not succeeding.”

I stand by that, but in context, I was talking about leaders and founders who count decks, posts, and assets shipped and call that a scorecard (volume is the wrong bar). But I’ve been pondering that sentence for a while, and it’s only half the story. It treats “the wrong things” as one category. In my experience it’s two. One of them is a problem. The other is the whole point.

Failure Isn’t One Thing

There’s good failure and there’s bad failure, and a team that can’t tell them apart will punish both or tolerate both. Neither is the desired outcome.

Good failure is a bet you made on purpose. You knew it had a real chance of not working, you said so out loud, and you knew what you’d learn either way. The test didn’t pan out, and you now know something you didn’t before. That’s the purchase price of information.

Bad failure is just … failure. You got it wrong, it was unexpected, and nothing was learned on the way down. A launch date slipped because nobody checked a dependency. A claim went out that nobody verified (Zero-Trust fact-checking exists for a reason). The thing broke, and the only surprise was how long it took anyone to say so.

Here’s the test I use to tell them apart, and it’s one question:

Did you write it down before you tried it?

My Opportunity To Fail

That’s the idea behind a phrase I’ve used in several workplaces now: “my opportunity to fail.” I chose those words carefully. An opportunity is something you go after, where a risk gets managed and a confession gets extracted.

It does one job: it removes the stigma that failure is always bad. A team that believes every miss will be held against them behaves predictably. They stop trying anything with a real chance of missing. They pick the safe path, the proven message, the same campaign as last year. The work stays comfortable, and the learning stops.

I’ve been doing this long enough to have been on both sides of that, in both flavors. I’ve had the miss I’m still a little proud of. I’ve also had the miss that made my stomach drop and taught me exactly one thing: never let it happen again. That gives me a durable awareness of the difference, not one story to tell. And the difference is almost never the outcome. It’s what you did before and after it.

Good Failure: Say It Before You Do It

A good failure is declared in advance, and that’s what makes it positive. If you only call something an experiment after it flops, you’ve just told a better story about a bad failure. (We’ve all heard the “well, it was always kind of a test” speech. Nobody is fooled.)

So capture it in the plan, before the work starts. Five lines is enough:

Note: if you don’t run the team, write the five lines for yourself and hand them to whoever approves your plan. A bet your boss saw in advance is a conversation. A bet they find out about afterward is a surprise.

This is where AI changes the math, by the way. As I wrote in using versus adopting AI, AI makes exploration cheap enough to run in parallel instead of committing in sequence. When a test costs an afternoon instead of a quarter, the failure worth worrying about isn’t the test that flopped. It’s the one you never ran.

Bad Failure: Acknowledge, Capture, Diagnose, Prevent

Bad failure will still happen. The goal is to never fail the same way twice.

  1. Acknowledge it. Say it plainly and early. The longer a miss stays quiet, the more expensive it gets.
  2. Capture it. Write down what happened, while it’s fresh, in a place the team can find it.
  3. Diagnose it. Isolate it like a bruise and press your thumb into it. Find what went wrong and what let it happen.
  4. Put something in place. A checklist item, a gate, a changed handoff. Something that makes the next one harder to repeat.

Notice that none of the four steps is “find the person responsible.” Diagnosing a failure and assigning blame look similar from a distance and produce opposite results. One of them gets you the truth, and the other gets you silence.

The Gap in My Own Plan (see, told you)

In PMM Plan on a Page, I gave you three altitudes, an 80% rule, and a way to protect 20% of your capacity for the work you can’t predict. It’s a plan I use.

It’s also missing a section.

That 20% is for the opportunistic stuff, the inbound you know will show up but can’t specify in advance. An opportunity to fail is the opposite, because it’s specified by design, ahead of time. It deserves its own tagged line in the quarterly view, next to your commitments. Left in the buffer, it’s the first thing cut when the quarter gets tight.

If your plan has no place where a deliberate risk can live, you’ve built a plan that quietly rewards safe work. That’s a missing section in the template I’ll be fixing.

But Do You Mean It?

Now the part for the managers (and anyone who leads work without a title, see managing versus leading).

Almost every leader says “it’s ok to test and learn.” Almost none of us have checked whether we mean it. Your team has, though. They checked the first time something didn’t work, by watching what you did next.

So here’s a short test. Be honest, nobody’s grading.

One more, for anyone with a boss: could you explain a good failure to your CEO in two sentences? If you wrote the five lines first, you already have them. “We bet on X, we knew it might not work, and here’s what we learned.” In my experience, executives forgive a failure they saw coming far more easily than one they didn’t.

If you answered those and winced a little, good. That’s the useful part. I wince at a couple of them myself, and I just wrote them.

Founders, This One’s Yours

A word for founders, because you set the weather. You’re singularly driven for success, and that’s the job. It’s also why protecting every dollar and every week from anything that might not work can feel like discipline.

Here’s the catch. A team that only runs safe plays doesn’t learn, and a team that doesn’t learn stops getting better at the thing you hired it to do. Learning is how you get more success.

So make the space on purpose. Nobody on your team will ask permission to fail, and nobody will take it uninvited. Pick one bet per quarter, put a cap on it, and say out loud that it might not work. Then, when it doesn’t, notice what your face does. They’ll all be watching it.

Now go check your plan.

Adam