If you've ever had a readout where everyone said “good work!” or “how cool!” then seemed to forget everything you presented and go back to their work without even a slight change, I promise you aren't alone. When I look back at the studies of mine that went nowhere, the research itself was usually fine. What was missing was an agreement, made before the study started, about what the decision was meant to change and who was going to act on it.

If you missed the last two emails, you can catch up on decision research here and evidence strategy here. Evidence strategy aims a study at a real decision, and outcome strategy is the promise you make up front about what that decision should move. It has three parts, linking the decision to a metric, getting the decision owner to commit to what they'll do, and defining success before anything ships.

Most of us were taught to write research goals like "understand how organizers feel about paying for group bookings." That's a good goal for a discussion guide, but nobody in a leadership meeting tracks feelings about paying. They track conversion, retention, revenue and support costs, so a study that doesn't connect to one of those slides down the list as soon as priorities get tight.

So I build a metric ladder, which is three rungs that connect what users do to a number leadership already looks at:

Then I put the three rungs into one sentence:

If we [decision], we expect [user behavior], which should move [product metric] and, over time, [business metric]

Back at TeamTicket, the group ticket booking app from the last email, Priya is deciding whether to build split payments this quarter. My ladder for her decision reads like this:

If we build split payments, we expect more organizers to complete a group booking without paying for everyone upfront, which should move group checkout completion and, over time, gross booking value.

When I don't know which metrics leadership cares about, I look at the company OKRs or the last quarterly business review deck and borrow their exact wording. If the deck says "gross booking value," I write "gross booking value" and not "revenue," since that's the phrase people recognize when they see my plan. I drop the OKR doc and a few past review decks into Claude and ask which metrics come up most often and how each one is defined.

The question I ask myself here is whether every active study has a named metric written into the plan.

The evidence threshold covers what would change the decision owner's mind, while commitment covers who does what, and by when, once the results come in.

I used to assume that if the results were clear, someone would pick them up and run with them. [Add a specific example here of a study where the results were clear and nobody acted, and what that cost you.]

Now I write the commitment as three lines, one for each way the study could go:

If the result is [A], [owner] will [action] by [date]. If the result is [B], [owner] will [action] by [date]. If the result is mixed, we'll [next step] by [date]

I always include the mixed line, since most studies come back somewhere in the middle, and without a plan for that the decision gets pushed to "let's revisit next quarter." For TeamTicket, it looks like this:

If most organizers say they'd switch to split payments, Priya will add it to the Q2 roadmap and assign the payments team by May 15th. If most organizers say they'd keep paying upfront, Priya will move it to next year's backlog by May 15th. If the result is mixed, we'll run a two-week fake door test on the checkout page before the Q2 roadmap review.

I don't ask for this in a big meeting. I send a short message after the kickoff with the three lines filled in:

"Hiya [name]! So we can act quickly when the results come in, I've drafted what happens next for each outcome below. Could you check the owners and dates and let me know if anything needs to change?"

I've found people are far more willing to edit a draft than write one. And if the owner won't commit to an action for any result, I take that as a sign the decision has already been made, and I ask whether we still need the study.

The question I ask myself here is whether my decision owner has said, in writing, what they'll do if the result goes one way or the other.

This is the part I skipped the longest, since asking for numbers made me nervous and I didn't want to look like I didn't understand analytics. But without a starting number, there's nothing to compare against after launch, and the impact story I tell later turns into guesswork.

So I agree on success with the decision owner using this formula:

Success means [metric] moves from [baseline] to [target] within [time window] after [change] ships

Here's how I fill in each part:

For TeamTicket, success means group checkout completion moves from 58% to 65% within 90 days after split payments ships.

If there's no baseline at all, I don't let that stop the study. I ask whoever owns analytics to start tracking the metric now, so by the time anything ships there are a few weeks of data to compare against.

I used to worry that a number in my plan meant I'd look bad if it didn't move. But success belongs to the decision, not the research. If split payments ships and checkout completion stays flat, that's useful for Priya to know, and my research still did its job, since it got a decision made on time with evidence behind it. That's what the decision rate counts, which feels a lot fairer to me.

The question I ask myself here is whether I know the starting number before the study begins.

These are the three worries I hear most, and what I do for each.

"My research is generative, so there's no metric."

Discovery research still feeds a decision, only a bigger one, like which problem area to invest in next quarter. So I point the ladder at the business metric for that area. If the study explores why groups abandon planning, the ladder points at group booking volume, before I know what we'll build.

"I don't have an analyst or access to data."

I use the numbers I already have, like support tickets tagged to the problem, how often it comes up in sales calls, or a one-question in-product survey. If there's nothing at all, I ask an engineer to add one tracking event, which is usually a small ticket, and I start counting from there.

"I'm junior, and I can't make a director commit to anything."

I frame the commitment message as a favor to them, so their team isn't waiting around when results come in. Something like "To save your team time at the readout, I've drafted next steps for each result, could you check these look right?" feels very different to them from being asked to promise something.

Once I have all three parts, I add a short outcome block to the top of the research plan, right under the decision sentence, so everyone sees it at the kickoff:

The biggest difference shows up at the readout. My old readouts opened like this:

We spoke to 12 organizers and found three key themes around paying for group bookings

With an outcome block in place, the same readout opens like this:

Priya agreed we'd build split payments if most organizers would switch. Nine of the 12 said they would, so it goes on the Q2 roadmap, and we'll know it worked if group checkout completion moves from 58% to 65% by the end of summer

Nobody has to work out what the research means for them, since we agreed that before the study started. It makes impact strategy, the third practice, much easier too, since you already know which number to go back to.

You can do this today, even on a study that's already halfway through. Pick the one you're running right now, write the metric ladder, and send your decision owner the commitment message before lunch. If you only get as far as the ladder, that's plenty for this week!

There are two downloads this week, so use whichever fits how you work.

The Outcome Strategy skill is a Claude skill that walks you through all three parts, one question at a time. You give it your decision sentence and your OKRs or a recent business review deck, and it builds the metric ladder in your leadership's own words, drafts the commitment message for your decision owner, and helps you set a baseline, target and time window. At the end you get an outcome block to paste into your research plan.

The Outcome Strategy Guide is a fillable PDF with the same steps laid out as a worksheet you can type into, save and share with your team. Inside you'll find:

To add the skill, download the zip file, open Claude, go to Settings, then Capabilities, and upload it under Skills. Then start a new chat and tell Claude you want to set up the outcome strategy for a study. You can also upload the PDF to Claude and have it walk you through the process.

Please tweak the time windows and targets to fit your team!

Exclusively for paid subscribers: