Leader Spotlight: Designing personalization without losing the plot, with Steven Landau
Steven Landau is Senior Principal Director of Digital Experience at ADI Global Distribution, where he leads digital innovation and customer experience strategy across global platforms. With more than two decades in product management and digital transformation, he’s held senior roles at Resideo, Snap One, Wells Fargo, and Lowe’s, driving initiatives that connect business strategy with user-centric design.
In our conversation, Steven Landau makes one thing clear — that AI alone isn’t the innovation. It’s how you use it to serve real human needs. He shares hard-earned lessons on leading AI transformation at scale, building trust through transparency, and creating personalized experiences that feel magical, not manipulative.
Embracing AI as a partner in product management
You’ve said you see AI as a positive force for product and where it’s heading. Do you feel like that’s a common view or a controversial one?
It depends on where you sit and which groups you work in, but I’d say that for the people I work with, the majority view AI positively.
Now, the anxiety surrounding AI definitely exists, especially in product management. In my experience, it’s more positive than controversial. However, I know there are probably more people out there who would say they fear it. I have some thoughts on why it’s positive for product managers overall and why we, as a community, need to think that way.
At a high level, using AI properly — and without getting into whether it’s agentic AI, AI agents, etc., just using the term as an umbrella for a range of technologies — enhances data-driven decision-making. You can process far more data, be it customer feedback, market trends, usage metrics, and so on, much faster than humans can. That gives us richer, faster insights for product strategy, prioritization, and understanding customer needs better than before because you’re drawing more from this massive data lake.
Second, there’s the automation of tedious tasks. Even today, we have a team of five people working on thousands of data points daily just to get our KPIs and metrics boards updated. It’s time-consuming; there’s data entry, reporting, and summarizing feedback. Take feedback, for example: we have a team that reviews every screen recording and heatmap manually. That should all be AI-generated, so they can focus on higher-value activities like strategy, innovation, and stakeholder management.
Another advantage is faster iteration and time to market. Across the product lifecycle, from generating initial design concepts to A/B testing, identifying bottlenecks, and suggesting features, AI or agents can support the entire process. That means reduced time to market.
Improved personalization and innovation are another area. Highly personalized user experiences unlock product opportunities that weren’t feasible before. Think advanced recommendation engines, NLPs, and more.
And one more key benefit is better risk identification. AI can help us catch fraud risks, biases, and misalignments in the development process early before they affect the final product. That kind of proactive adjustment is only possible if we’re leveraging these technologies.
So at a high level, those are a few main advantages. Now, on the anxiety side, the biggest concern is job displacement. But I really think if product managers see AI not as a replacement tool, but as an augmentation tool — and I think that word is important — the fear becomes more manageable. Yes, the anxiety is real, but changing our perspective helps.
Skill gaps are another concern. Some PMs worry they don’t have the skills to leverage AI. That’s where upskilling and training come in. A PM should come into the office every day asking, “What am I going to learn today? How am I going to fail forward today?” That mindset is key.
There are also valid ethical concerns: algorithmic bias, privacy violations, and questions of accountability when AI makes autonomous decisions. Hallucinations, too. You’ve seen what that AIX engine did this past week, pushing out bad stuff. But that often comes back to training and how the systems are leveraged.
And finally, some people worry we’ll lose human intuition if we rely too much on AI. But the reality is, AI can’t replace human intuition. Not today. I don’t know what the future holds in 5, 10, 20, or 50 years, but right now, we still need the human side.
So yes, it’s a controversial topic within the product management community. But I believe the benefits and potential for empowerment outweigh the risks.
You took this perspective and used it to innovate your own role within your organization. How did that work and what was your perspective there?
I’d say it was less about changing what a product manager does and more about replacing how we do certain things and supercharging my capabilities and shifting focus to higher-value activities. I used a lot of AI agents and tools to offload or automate the repetitive parts of the role.
For example, a big chunk of a PM’s time typically goes into things like synthesizing market research, writing exhaustive business requirements, competitive analysis, and managing documentation. By leaning into AI tools, I was able to take those time-consuming tasks and streamline them. That opened the door for me to take on a more strategic and elevated role within the company and to become one of our resident AI experts.
How did I do that? Honestly, by diving in. Reading, experimenting, downloading tools, playing with them, learning through doing. That hands-on experience helped me accelerate research and synthesis, supercharge brainstorming, and ideate product concepts and solutions faster. If we had a problem, I’d use prompts like, “Here’s the issue. How would you solve it?” And AI tools could help jumpstart whiteboard ideation.
I also automated parts of our documentation and communication processes. Before, we’d spend time writing business requirements, translating those into technical specs, then into test scripts — all through the usual JIRA flow with epics, stories, tasks, and so on. Now, we’ve set up triggers and automation where the AI takes a business requirement and generates suggested test scripts to validate it. That used to take days or even weeks. Now, it happens instantly. All we have to do is review and maybe tweak a few things but it gets us 90% of the way there.
Rethinking the product lifecycle with AI
AI capabilities are shifting the way the job is done. Do you also see it changing the product life cycle and how the traditional build-measure-learn cycle works?
The traditional Lean Startup methodology — the build–measure–learn cycle, which is a cornerstone of product development — is undergoing a real transformation with AI. And it’s different from the digital transformations we’ve seen in the past. It’s almost like adding a supercharger to your car. We’re supercharging each phase of the Lean BML model to create faster, more data-driven, more intelligent feedback loops.
To go a level deeper: if you look at the build phase, the creation process, we now have AI helping with hypothesis generation and MVP development through automated prototyping and scaffolding. AI can even assist in generating code snippets or early-stage UI/UX markups. Now, that doesn’t replace a designer, but it can interpret the requirements and, say, your homepage concept or a feature change, and sketch out the boxes. Then your UX team can refine from there. That’s pretty cool.
We’re also using AI for things like predictive risk assessment and performance testing right from the start, which falls under the measure phase. You now get real-time, deeper, even predictive insights. As you build MVPs or code snippets, you can automate analytics, use pattern recognition, and even run A/B tests more intelligently. Sure, there have been heatmap tools for a while, but those required a human to analyze them. Now, by layering data and AI, we’re spotting patterns — sometimes ones we wouldn’t have found with traditional methods.
Take a behavior like people skipping the search box. In the past, you’d just ask, “Why are they skipping this?” But with AI, you might uncover a pattern across the whole purchasing cycle that reframes the issue entirely.
Then there’s sentiment analysis and qualitative feedback done at scale. With NLP, we can summarize, categorize, and extract sentiment from large volumes of unstructured data. That saves tons of time and leads right into predictive modeling. We’re talking about forecasting future user behavior, feature adoption, or market shifts, all while analyzing market research and A/B test results in parallel.
It’s personalized experimentation at its finest.
Making the business case for AI innovation
How would you recommend that people who share your perspective that AI can be used for genuine innovation make the case for its integration as a driver of innovation?
It’s a good point, and a critical challenge for product teams and leadership. Unfortunately, the natural inclination of many businesses, especially those under financial pressure, which is most companies these days, is to view AI’s gains primarily as an opportunity for cost reduction. And that often translates to cutting headcount and budget.
But I believe product teams have a strong case to make for reinvesting those gains into innovation. If you’re a company under pressure, you still want to be ahead of the competition; every company says that. That should be a core tenet, “We want to be the key player in our industry.” To support that, product leaders need to quantify efficiency gains and frame them as increased capacity.
So instead of saying, “We saved X dollars,” you can say, “We saved Y hours per person on Z task,” or, “This AI tool freed up X percent of the PM’s time on data analysis, which allowed them to do A, B, and C.” It’s not just about savings. It’s about what you can now accomplish with that freed-up time.
That’s how I’ve approached it. I don’t focus on automating tasks in a way that creates redundancy and invites leadership to cut people. I show how AI creates new capacity. For example, our AI-powered customer feedback analysis tool reduced time spent categorizing feedback by 70%, which freed up 15 hours a week for PMs. That time was then used to focus on new features, features that generated revenue, which we wouldn’t have had the bandwidth to build otherwise.
Let’s say one of those features improved search engine results. Even if it’s AI-driven, you still have to feed that search engine with data, prompts, and context. And beyond that, you might then want to personalize it further, for instance, by integrating pricing dynamically into results. All of that takes time and iteration.
The product backlog is often built through discussions with stakeholders, especially sales and marketing, who are focused on how features can drive revenue. If the company is financially strained, it should be open to anything that brings in more money. So if AI saves you 15–20 hours a week, and as a PM, you can now work through that backlog faster, you’re directly connecting AI-driven innovation to strategic growth.
This is where you can start tying innovation to your north star metrics, OKRs, and competitive differentiation. That connection is powerful. It shows leadership that AI isn’t just a cost-cutting tool. It’s an engine for top-line growth.
Have you seen AI-driven capabilities, like intelligent recommendations or demand forecasting, create new monetization opportunities that didn’t exist in the original product vision?
This is less about digital transformation in the traditional sense and more about AI transformation. I’m not sure what the current buzzword is, but it feels like a distinct shift from earlier digital changes.
If you look at companies like Amazon, Netflix, or Spotify, intelligent recommendations have clearly shown that the smarter and faster those systems get you to what you’re looking for, the more they increase conversion rates and average order value (AOV). And those two KPIs, conversion and AOV, are critical for any product manager working on a transactional digital property. That’s the key — transactional, meaning your site or app directly generates revenue.
AI-driven product recommendations are especially impactful. Let’s say a user types in a sentence like, “I want a nice-sounding, outdoor, waterproof speaker.” The AI doesn’t just return six product options; it can also bring in documents, articles, ratings, and more. You can really tailor this to your business and use case. That’s where it gets exciting.
And it goes beyond just the search bar. There’s real potential in bringing AI-powered recommendations into other parts of the user journey. For example, on the product detail page (PDP), or in the cart before checkout, with prompts like, “Customers also bought…” or “Recommended for you.” We’ve seen this on sites like Amazon, but many companies are still catching up.
For businesses concerned about the current economy or budget constraints, this is actually the time to lean in. If you can increase your conversion rate, you increase AOV, and with that, revenue. These improvements go hand in hand.
On top of that, you’re also enhancing user engagement and retention. It’s not just about AI search; that’s just one example. You can now do smarter cross-selling and upselling, introduce different pricing tiers, and even embed sponsored ads or recommendations. There are so many interesting possibilities here.
You can also apply this to demand forecasting. Imagine dynamic product catalogs that adjust in real time based on predicted demand, competitor pricing, and other signals. If you’re selling a widget and nine other companies sell it too, you could feed in pricing and margin data across the market and use AI to dynamically adjust your own pricing, within healthy boundaries, to stay competitive.
These are just examples, but this is a massive area for AI monetization. And when product teams are being asked to justify how they spend their time, this is a perfect response. If AI frees up time, use it to work on things leadership does care about. I could probably name 50 AI monetization ideas right now, and I guarantee not a single leadership team would say, “We don’t want that.” They’d say, “Yes, go build it.”
Designing personalization that feels human
How do you use AI to personalize experiences at an individual level without creating a confusing interface or straying from the product’s purpose?
The first step is making the personalization feel magical and helpful, not creepy or confusing. That’s a common fear. And it starts with how we frame AI internally. I prefer to think of it as augmentation rather than automation. We’re using AI to enhance the user experience, not to take away user agency.
It’s easy to over-engineer personalization, like trying to personalize an entire homepage or taxonomy just because we can. That’s when it becomes overwhelming. Instead, you need to ask: what real problems are we solving for the user? Start small. Use AI to address specific pain points like discovery fatigue, information overload, or irrelevant content. That’s where it can have a real impact.
It’s also critical to understand your users. Some people love hyper-curated feeds; others want more control and transparency. There’s no one-size-fits-all model. You need to rethink your personas. Ask who embraces personalization, and who resists it. Tailor your approach accordingly. If someone only shops for audio-video gear, don’t show them home security products. But also, don’t assume everyone wants a completely tailored homepage; some users don’t.
Internally, teams need to define clear success metrics. What does “successful personalization” mean for your product? Higher conversions? Lower bounce rates? Reduced time-to-discovery? Whatever it is, define those KPIs upfront; they’ll help you focus and avoid scope creep.
Transparency is also key. If AI is making recommendations, explain why. Use simple, respectful language. For example, “We recommended this based on your recent purchase. Would you like us to keep showing similar suggestions?” That gives users a sense of control. A lot of companies skip that entirely, and users are left wondering, “How do they know this about me?”
Finally, simplify the interface. Work closely with UX to ensure the experience is intuitive, with clear visual cues and progressive disclosure. Whatever version you launch, assume it’s V1. Plan to iterate. Use A/B testing and test with specific personas. You’ll likely go through many rounds before it feels right. That’s okay. Fail forward, but do it thoughtfully.
Building trust and transparency in AI experiences
With AI integrated into your products, how has your approach to collecting user feedback changed? Do you gauge how much AI users want in their experience?
Yes, we are doing this, though still in early stages and iterating. Our goal is to ensure that whatever we build, whether it’s a digital product, site, or app, makes users feel like we truly understand what they want to do without making them feel they’re losing control. That balance is crucial.
You can’t shortcut this process. When implementing AI personalization, it’s vital to provide both users and product managers with a feedback mechanism, because there are many nuances involved. Trust and transparency are key. The more users experience relevant content, features, and recommendations that save time, make their lives easier, and help them achieve their goals faster, the more they accept and welcome it.
Convenience is another important factor. Products that anticipate needs or proactively offer solutions create what we call “magic moments.” These happen when personalization feels intuitive and helpful without being creepy. We’ve seen this in some implementations where users say, “Wow, that used to take me an hour.” At that point, you’ve hit the apex of efficiency, convenience, and relevance, all delivered in a unique, innovative way.
What does LogRocket do?
LogRocket’s Galileo AI watches user sessions for you and surfaces the technical and usability issues holding back your web and mobile apps. Understand where your users are struggling by trying it for free at LogRocket.com.