5 Steps to AI Product Success

Welcome and Introductions

Kenny Kranseler & Ryan Cantwell | 00:00:00 – 00:07:23

All right, welcome everybody. Hello and welcome all. We’ll get started in a couple of minutes or so. And as we get everybody in and settled, if you want to share with us in our Zoom chat where in the world you are located, that’s always a fun game to play.

All right, we got a Wisconsin. Got the East Coast, the Midwest, and the West all covered. Idaho — welcome, Michelle. Oh, and my pal from Milwaukee. I think we’ve got two people from Wisconsin here. Joe’s here — hey, Joe. One of them I know quite well.

We are excited today. We have Ryan Cantwell joining us. Ryan is one of our principal consultants and trainers at Productside and has been doing a ton of work around AI and the AI maturity space. Today he’s going to be covering the five steps to AI product success, which is near and dear to his heart because he’s been heavily involved in developing the framework behind it. Ryan, over to you.

About the 2026 State of AI Maturity Report

Ryan Cantwell | 00:07:23 – 00:09:55

Thanks, Kenny. Yeah, I’m excited to be here today. This session came directly out of work we’ve been doing on Productside’s 2026 State of AI Maturity Report. We surveyed 250+ product leaders worldwide, and the data is blunt:

• Individual AI adoption is running well ahead of strategy, governance, and ROI measurement
• Most product managers are using AI — that part’s happening
• But at the organizational level, there’s a significant gap
• Most teams don’t have a documented AI strategy
• There’s no named owner of AI direction
• There’s almost no consistent way to measure whether any of it is working

The five steps we’re walking through today come directly from that data. These are not opinions or frameworks we invented in a conference room. They are patterns from real product organizations doing this well — and doing it poorly. So if you’re somewhere between “we’re experimenting” and “we have no idea if this is working,” this session is for you.

Part 1: The Readiness Gap

Ryan Cantwell | 00:09:55 – 00:17:08

Before we get into the five steps, I want to make sure we all feel the problem clearly. Because the gap we’re talking about isn’t subtle — it shows up in the numbers.

Here’s what we’re seeing across product organizations today:

• Personal AI adoption is high — PMs are experimenting, prompting, and iterating daily
• Organizational readiness is lagging far behind
• 40% of teams are still developing their AI strategy
• 22% are running fully bottom-up — no coordination at all, everyone improvising, nobody aligned

One quote from the report stuck with me — from a principal product manager in manufacturing: “Everybody is using AI in their own way. No common strategy, no shared goals.” That is the whole problem in one breath. Personal adoption high, organizational alignment missing. And there is that gap where the value starts to leak out. Not because people aren’t trying, but because nobody pointed at trying in the same direction.

It’s a constant push and pull where everyone is on a different rope. There’s lots of effort — and the flag just doesn’t move.

The question we need to start asking is: does our team share an understanding of what AI is for? Or is it still everyone running their own little pilot at their workstation?

Today’s session is built around five concrete steps to close that gap:

• Step 1: Fix your data first
• Step 2: Build an AI strategy, not just AI activity
• Step 3: Target the right skills
• Step 4: Build governance that people actually use
• Step 5: Connect AI to outcomes leadership cares about

Poll: How Are You Currently Measuring AI’s Impact?

Kenny Kranseler & Ryan Cantwell | 00:17:08 – 00:18:52

Before we dive into the steps, let’s see where the room is. We’ve got a poll coming up — go ahead and respond. We’re going to close it down in another 10 or 15 seconds.

All right, there is a winner. We’ve got productivity as the winner. I was picking the bottom one — not measuring. The bottom one was a close second. Not currently measuring doesn’t surprise me at all — we’ve heard consistently that that’s been a theme throughout the State of AI Report.

And in a minute, I’m going to make the case that productivity in particular is actually the weakest measure. If you’re not measuring and you jump to productivity, we’re going to see where that might be a little bit of a trap. I bet people are measuring productivity by usage too. Outputs instead of outcomes. Okay — let’s get into the five steps.

Part 2: The 5 Steps Overview

Ryan Cantwell | 00:18:52 – 00:20:17

Here’s how today is structured. We’re going to walk through all five steps in order, because they build on each other deliberately:

• Step 1 is about getting your data foundation right — because everything else depends on it
• Step 2 is about building an actual strategy, not just accumulating AI activity
• Step 3 is about targeting the skills that connect AI to outcomes — not the ones teams are already good at
• Step 4 is about governance that people will actually use — not a policy nobody reads
• Step 5 is the one that keeps all the others funded — connecting AI to the outcomes leadership already cares about

Let’s get into it.

Step 1: Fix Your Data First

Ryan Cantwell | 00:20:17 – 00:25:56

Step one is about data. And I want to be clear — this is not a technical talk. You don’t need to be a data engineer to care about this. But you do need to understand why it matters so much, because dirty data is the number one thing killing AI efforts before they start.

When it comes to data, it’s very easy to start thinking in terms of boiling the ocean. Data is housed in different places — different reports, different silos — and getting access to it is different across all of them. But mapping where data lives, how to access it, and who controls and owns it puts you in a spot to get started.

Most teams who go through this exercise are genuinely surprised. You’ll start to see that data lives in more places than you thought, and it’s accessed by more tools than you ever imagined possible.

Three things you can do for Step 1:

First, map it — don’t boil the ocean. Take the high-level view. Understand where your data lives. You’re not fixing all of it — you’re building the picture. Once you have that picture, you can start to fix what matters first.

Second, prioritize ruthlessly. Don’t deep-clean the whole house before your guests arrive — only clean the rooms they’ll see. Focus on the highest-leverage data sources your real AI use cases actually depend on. Not everything, just those.

Third, set a quality bar. Decide upfront what an AI output has to clear before it reaches a customer or influences a decision. For example: our recommendation engine needs an 80% relevance score before it surfaces anything. That bar will look different for different products — but you have to define it. Remember the Air Canada situation a couple of years ago? That did not come close to clearing any kind of quality bar. A human-in-the-loop check might be what gets you started — having eyes on AI-generated outputs before they publish, to give you the confidence you need.

This work is unglamorous. Nobody’s going to clap for it. But it is the foundation for the next four steps, and it will put you in a spot to do some really good work.

Step 2: Build an AI Strategy, Not Just AI Activity

Ryan Cantwell & Kenny Kranseler | 00:25:56 – 00:34:44

Step two is about strategy. And I want to be direct: without a strategy, every AI decision becomes a judgment call. And a hundred smart people making a hundred separate judgment calls don’t add up to a direction — they add up to a hundred directions. That’s not a team using AI. That’s a bunch of solo experiments wearing the same company logo. And that’s where most organizations are playing today.

Here’s what the report shows about where teams actually stand:

• 40% of teams are still developing their AI strategy
• 22% are running fully bottom-up — no coordination whatsoever
• Most cannot articulate how their AI activity connects to business goals
• Most do not have a named person responsible for AI direction

That last point is the one that matters most. Because when budget conversations come around — and they always do — “we’re using AI” is not an answer. “We used AI to reduce time-to-insight by 40%, which accelerated our roadmap decisions by two sprints” — that’s an answer.

Three things you can do for Step 2:

First, name what AI is for. As product people, we’re in a strong position to do this. Define what value you expect it to deliver — internally, externally, or both. Pick two or three outcomes it should move: revenue, speed, quality, risk. Write them down. Once everyone’s nodding at the same list, everything else becomes noise until those targets are hit. Aim at it.

Second, name an owner. A strategy with no name attached is not really a strategy — it’s a suggestion. Putting one person accountable for AI direction, for priorities, and for breaking the tie when two teams want to pull in opposite directions can be really powerful. And I want to address what I know a lot of product people are thinking: what power do I have to make that happen? You don’t need to go assign it and tell anyone what to do. Start having conversations. Evangelize. Show people what’s possible. Kenny, what’s your take?

Kenny Kranseler: Push from the bottom up and ask for the top down. Say — here’s where I think AI can help us, but we need to align it. Talk with your manager. Talk even skip-level at some point. Say: I don’t want to be an island out here — we can do this better together. Ask for some level of strategy or ownership within the organization. Maybe it comes back to you, and that’s a great opportunity to grab the brass ring and run with it.

Ryan Cantwell: Absolutely. And by becoming that advocate, you’re also helping to align stakeholders around the same outcomes. What is AI for? That’s a product superpower — aligning everybody on outcomes. Let’s do it now for AI.

Third, build a shared playbook. Take your scattered pilots and give them a common way of working:

• How do we test AI — what’s our assumption going in?
• What’s our win condition — is this helping us or not?
• How do we measure it — and is the investment worth it?
• How do we share the results so learning compounds instead of evaporating?

Treat your AI use like an experimentation program — the same way you experiment with your products and customers. Build an engine where learning compounds. And as product people, this is where we can have real influence: bring your experimentation philosophy to the broader organization. Guide other teams through surfacing assumptions, designing tests, and knowing what a win or a loss looks like. That’s how you make yourself more valuable to your organization — and help it mature.

The whole idea: set the direction from the center, establish a playbook, name an owner, and then let teams run locally. That means alignment — not micromanagement.

Step 3: Target the Right Skills

Ryan Cantwell | 00:34:44 – 00:40:30

Step three is about skills — and there’s an important distinction to make upfront that surprised me when I saw it in the report. The focus is on skills that will help the least. Companies are investing in the skills people need the least, and not investing in the ones that would actually move the needle.

Here’s what the data shows:

• Product managers score themselves 7.5 out of 10 on prompting and experimentation — high, because they do it constantly and have gotten good through repetition
• On value mapping and strategic assessment, they score 5.9 out of 10
• 28% rate themselves a 1 to 4 on connecting AI to actual business value — that’s the lower 40%

More than a quarter of product managers are essentially saying: I can drive this thing, but I can’t tell you where it’s going. The hands-on skills develop on their own through daily use. But the skills that connect AI to outcomes — they never just show up. They need deliberate investment.

And here’s something that felt genuinely important from the data: product managers inside companies with a clear AI strategy score nearly two points higher on every one of these skills. The organization is the rising tide, and that direction makes individuals better.

The three lowest-scoring skill areas — and where your leverage is hiding:

• Value mapping — understanding and articulating where AI is actually creating business value
• Cross-functional collaboration — bringing other teams into the AI work in a structured way
• Strategic assessment — evaluating AI opportunities against organizational goals

Three things you can do for Step 3:

First, target the low-score skills. Don’t invest more in what people already do well. Focus development on value mapping, cross-functional collaboration, and strategic assessment — that’s where the leverage is.

Second, make value the center of everything. Sit with that 28% figure. More than a quarter of PMs can’t confidently connect AI to business value — and that doesn’t fix itself through daily use. It needs product leadership to build it into:

• How you develop your team
• How you run your reviews
• How you incentivize and reward the behaviors you want to see

Assign tangible metrics to AI work — speed, revenue, efficiency gains. Show people the connection between the AI work they’re doing and a number leadership cares about.

Third, invest in strategy as a skill enabler. If skill building isn’t sticking — if training sessions aren’t changing behavior — it might not be a problem with how you train people. It might be that there’s no strategic ground for those skills to take root in. Clear AI direction lifts skill scores by nearly two points across the board. Strategy enables skill. Not the other way around.

Individual Skills Assessment

Ryan Cantwell | 00:40:30 – 00:41:35

And Kenny, I have something for everybody here. We talked about where product managers score high and where they score low — and naturally you might be thinking: where do I land?

So we built a free diagnostic that benchmarks your product management skills — AI fluency included — against a global cohort of product people. You get a personalized report at the end that tells you:

• Where you shine
• Where you need to focus
• How you stack up against everyone else doing the job

It’s the same skill data from the report, pointed at you specifically. The link is in the chat. It only takes a few minutes — and it’s worth doing this week while everything’s still fresh, before your next meeting swallows your calendar.

Step 4: Build Governance That People Actually Use

Ryan Cantwell | 00:41:35 – 00:46:45

Step four is governance. And I can feel it, Kenny — when I say the word governance, the energy leaves the room. So let me reframe it before I start.

If governance sounds like the meeting where good ideas go to fill out forms, you’re not alone. But if done right, governance doesn’t add friction — it removes it. The whole goal is guardrails people use, not a policy binder that lives on a shelf making everyone feel vaguely guilty they’re not looking at it.

Here’s where most teams are right now:

• 65% of product managers are working with no clearly documented AI policy — making judgment calls every single day with no reference point
• 9% have no guidelines whatsoever — it’s a total open road
• 80% are using AI regardless — so no policy doesn’t mean no AI, it just means harder

No policy just means it’s more difficult to:

• Assign accountability
• Establish an agreed-to bar for what “good” looks like
• Have an answer when something goes sideways — and things do go sideways

Think of it like a four-way intersection with no stop signs. Everyone’s still driving through it — you just don’t find out it’s a problem until two cars meet in the middle. People don’t freeze because they don’t have AI. They freeze because they have no clarity on how to use it responsibly. That’s the fight-flight-or-freeze moment — and governance is what gets people unstuck.

Three things you can do for Step 4:

First, start with one page. I am not talking about a policy binder. Start with guardrails people can actually remember — because a policy nobody remembers is a policy nobody follows. If it needs a table of contents, you’ve already lost. One page. Make it accessible, and make sure the reasoning behind each guideline is clear so nothing feels arbitrary or unnecessary. Engage your stakeholders in the process too — bring them in to help set those guidelines, and you win buy-in while you build it.

Second, make the safe path the easy path. This is the real trick. Don’t police people into compliance — design it in. Bake the good defaults right into the everyday workflow so the responsible choice is also the path of least resistance. People don’t resist guardrails. They resist what feels unnecessary.

Third, name an owner. Same drumbeat as every other step, for the same reason. Governance without a name on it quietly rots. Someone has to own keeping it alive — or in six months it becomes a document sucked into the black hole of SharePoint that nobody opens.

Keep it light, make the default easy, name an owner. That’s governance that survives in real work.

Poll: How Are You Measuring AI ROI?

Kenny Kranseler & Ryan Cantwell | 00:46:45 – 00:48:54

All right, let’s run our second poll before we get into Step 5. The question: how are you currently measuring the ROI of your AI investment? Take a look at the options and let us know where you are right now.

All right, it’s a two-horse race. Let’s close it down and share the results.

Productivity is the winner — and not currently measuring is a close second. Not surprising at all, given what the report found. And in just a moment, I’m going to make the case for why both of those might be a trap — and what to do instead.

Step 5: Connect AI to Outcomes Leadership Cares About

Ryan Cantwell | 00:48:54 – 00:52:16

Step five is the one that keeps the other four funded. You can nail all of it — clean data, sharp strategy, strong skills, governance people love. But if leadership can’t see what AI is worth to the business in plain numbers, the budget conversation gets ugly fast. Tools you can’t justify are tools that get cut.

Step five is all about making the case in the language leadership already speaks. Not your language — their language.

Here’s what the data shows about where teams are right now:

• 24% of teams don’t measure AI’s impact at all
• Only 16% connect AI to customer and business outcomes
• The vast majority are flying with no instruments

And measurement isn’t separate from skill — it’s tangled up with it:

• PMs who invest with clear goals score 6.7 on value mapping
• PMs who are still just experimenting score 5.1
• Knowing why you’re doing something makes you better at doing it

With agentic AI costs climbing — token limits and everything else — the pressure to justify the spend only goes up from here.

Let’s make this concrete. Take a team using AI to triage support tickets.

The activity metric: we are triaging faster. That’s fine — but nobody in the C-suite cares that you’re triaging tickets faster.

The outcome metric, rewritten: we cut resolution time by 3 days, which lifted 90-day retention by four points. That second sentence is a number the C-suite recognizes. Same work, same solution, completely different conversation with the people holding the budget.

Three things you can do for Step 5:

First, kill the activity count — replace it with outcomes. Faster is not a result. “Three days faster, four points to retention” — that is a result. Every time you catch yourself reporting an activity metric, swap it for the outcome it was supposed to drive.

Second, frame the return. Put AI spend in front of leadership as an investment with a return attached — not a cost you have to defend. Don’t say “this costs us $50K a year.” Say “this $50K returns $200K through these efforts.” Same spend, completely different sentence — and a very different budget conversation.

Third, bring one number. Not a dashboard, not 40 charts of AI activity — one number. Leadership already cares about:

• Revenue
• Retention
• Speed to value

Find one clean figure that connects to something they lose sleep over. That’s what gets budget renewed.

Three Things to Do This Week

Ryan Cantwell | 00:52:16 – 00:55:21

And here we are — that’s all five steps. Let’s land somewhere useful.

I acknowledge there is a risk with any framework like this. It’s energizing as we’re hearing it. But then you close your laptop, the inbox is on fire, and by tomorrow it’s gone. Good intentions, very little change. Not today. Let’s talk about Monday morning when you’re back at your desk, the webinar glow has worn off, and you’re wondering where to start.

Three things you can do this week — not this quarter, this week:

First, audit your data. Name what’s broken. You’re not fixing all of it by Friday — just find the mess and call it out loud. Naming it is most of the battle.

Second, put one person’s name on AI ownership. Not a committee. Not a team. One human being who owns the direction. If you only do one thing off this entire list, that is the one:

• If you’re an individual contributor — go talk to your manager about who that person should be
• If you’re a leader — go talk to your peers and figure out who it should be

Third, pick one outcome metric leadership cares about and commit to reporting it next month. Just one — associated with your AI usage. Find the thing they care about: revenue, retention, speed to value. Start tracking it. Share it.

A data audit, one owner, one metric. That’s it. Everything else in this talk grows out of those three. Start small, start soon, and you will find meaningful change follows.

If you forget everything else from the last 45 minutes or so, keep this: the teams that win with AI aren’t the ones moving the fastest. They’re the ones measuring value and pointing AI at making more of it.

The whole industry is screaming at you to go faster, ship more, fall behind, or die. But speed with no measurement isn’t progress — it’s just expensive experimentation with better branding. These five steps were never about slowing you down. You’re already fast — the data proves it. They’re about making sure all that speed is pointed at something that matters. Fast is easy. Fast in the right direction is the whole game.

Resources and Upcoming Courses

Kenny Kranseler | 00:55:21 – 00:57:08

If you’re curious about the numbers behind what Ryan just shared, click the link in the chat or use the QR code on screen to go into the full detail of what we learned in the State of AI for Product Management Report.

If you want to go further, here’s what’s coming up:

• OPM course in Chicago — live and in person
• Both in-person and live online programs to work through all of this in a structured way with your team
• An upcoming webinar on July 08 — helping Dean Peters do a deep-dive session on Vibe Coding: not just how to get running code out there, but how to use it to learn and push forward on product initiatives
• A full range of courses on Productside — both live online and in person — for Optimal Product Management and AI Product Management, separate for now but starting to come together more and more

Details are in the chat. Sign up, enjoy, and you will learn a ton from either myself or Dean.

Q&A and Closing Remarks

Kenny Kranseler & Ryan Cantwell | 00:57:08 – 01:01:30

And with that, we will go to some questions and answers. Feel free to throw questions into the chat or the Q&A screen. A couple have popped up already.

Question:
How do we identify the right skills to learn AI in the first place — where should we start?

Ryan Cantwell:
I think the starting point is: what is one thing I need to be using AI for? Plug into something like the Productside community and attend webinars like this one. But beyond that, just start playing with it. Our report found that most people are already experimenting, and you’re going to learn the most fastest by doing. Think about the things that take you a long time right now — the things you spend a lot of manual time on — and test whether AI can help:

• Make sure it has the right context
• Provide as much relevant data as possible
• Let it take a stab at it and see if it can help you gain insight faster

One thing I did on myself recently: the AI we use at Productside has hooks into our project management software, our email, our Slack channels — everything. I asked it to look at what I’m doing every day and tell me where I have the most opportunity to leverage its help. And it did — it told me exactly where it could add value. That was an easy place to start.

Question:
I understand the concept of identifying which metric to move. However, the key metric for internal usage is focused on efficiencies — and that’s one of the least-measured activities. How do we handle that?

Ryan Cantwell:
Internal product teams are always an interesting case. Even if efficiency isn’t a top-line metric, you can still quantify it:

• Correlate time saved to dollars — how much do we pay these people, and how much time are they no longer spending on that manual work?
• Frame it as opportunity cost — now that they’re freed from that task, what more valuable thing are they doing instead?

Whether it’s an internal team or not, those are the areas to look at to make efficiency measurable and meaningful.

Question:
What is the one thing a product manager should be using AI for right now?

Ryan Cantwell:
I’m going to recommend you use AI to sharpen your critical thinking skills. Critical thinking is so important for a product manager — and what I don’t mean is let AI do the thinking for you. What I mean is let AI test your thinking:

• Have it ask you questions to uncover things you might not be considering
• Use it to surface hidden assumptions
• Ask it to spot risks in your reasoning

It supplements the human judgment we bring to the table as product people. Start there — have it act as an assistant and coach as you’re working through a challenge. Take a document you’ve created, throw it at AI, and ask: how can I make this better? Then have it interview you — ask me questions so I can give you the context to be useful. That’s a great way to use it.

Kenny Kranseler | 01:01:00 – 01:01:30

I hope you all found this valuable. We’ve hit the top of the hour — keep the discussion going, join our LinkedIn group, and join our upcoming webinars. There’s one coming up in a couple of weeks. Have a great day, folks. Thanks, everybody.