Leader Spotlight: Some real talk on startups, growth, and AI, with Raj Singh

Raj Singh is Vice President of New Products at Mozilla. He began his career as a software engineer before starting his first startup as a grad student in college. Raj went on to create more than a dozen consumer-facing products, seeing them through the zero-to-one process and scaling them to tens of millions of users. Now a four-time exited VC-backed founder, Raj has seen virtually everything when it comes to product.

In our conversation, Raj shares hard-won lessons from building zero-to-one products — from validating ideas through personal pain points to navigating the murky middle of startup growth. He talks about the myth of green space, why timing can make or break a product, and how to balance intuition with iteration. Along the way, Raj reflects on common product traps, the evolving role of generative AI, and why great storytelling is just as important as great design.

---

Validating and building zero-to-one products

When you’re just starting out, what’s the first question you ask yourself, or your potential users, to validate whether a pain point is worth solving?

My career has mostly been direct-to-consumer, so I always start with, “Do I experience the pain myself?” All of my ideas begin there. I know that goes against the usual product advice, which is “don’t solve for yourself, solve for others.” And that’s fair. There’s an entire B2B ecosystem built on that. But I’d argue that many founders, even in B2B, are solving a pain point they’ve experienced themselves.

That matters because product management isn’t just about creative strategy or product sensibility. It’s also about domain understanding. And when you’ve gone through a pain point yourself, you bring a deeper level of understanding to it.

So that’s where I start. Then I ask, “Do I have a creative way to solve this? Do other people experience it too? And can I reach them?” I don’t worry much about monetization at that point. A lot of people ask, “Can you make money from this?” but I don’t focus on that early on. I’m in the camp that if I can solve a real problem and reach the users, I’ll figure out monetization later.

That’s a pendulum, of course. It swings back and forth. But historically, in consumer, that approach has worked well. The hardest part is getting users, especially since go-to-market is so different from B2B, where your buyer might be coming from the top down.

Do you ever feel like, because you're drawing from your own experience with a problem, that it can lead to over-engineering? Or do you still try to pull back and validate user needs before going too far?

Usually, once I know there's a problem space I want to work in, I just tell people, “Start swimming.” What I mean is: start building anything. It’s not over-engineering at that point. You’re just prototyping. Maybe you’re sketching a concept, drawing it out, or using GenAI tools to get something going faster than before. The idea is to get in motion.

While doing that, you’re also learning more about the space. You might talk to former employees of companies in the same area. You might study competitors or dive into market research. Through all that, things start to take shape. I call it “jello in motion.” Some parts stick, others fall away, but it slowly starts to solidify.

Almost never do you start at point A and end at point A. You usually start there and end up at point C. But you have to go through the journey. It’s like a series of small S-curves leading into a larger one. You keep learning, iterating, and adjusting as you go. And you never really know how long that process will take. Sometimes, something only unlocks after you come up with a new technical approach or a fresh UX idea.

I’ve had moments where I think I killed something too early. That’s why I don’t really believe in wide-open green space. What matters is leaning into a problem you genuinely care about. That passion drives the momentum.

As you go deeper, you rely on your intuition, product sense, and personal taste to guide you toward something useful and unique. If you don’t feel confident in those areas, it’s harder. Zero-to-one work isn’t easy. Some product managers thrive more in operational roles, while others are better suited to building new products.

When you say you don’t believe in green space, do you mean you don’t think there are truly untouched areas anymore? Everything's been done to some extent?

I think it's always good to assume there’s no green space. That’s a good starting point. The idea that you’ve come up with something totally unique that no one else has thought of is rarely true. Most of the time, someone already has. In fact, it’s often already been tested. It just may not have been the right time or the right approach.

I had an investor once say something kind of wild. They told me, “We never make the wrong investment, we just make it at the wrong time.” On one hand, it sounds a little arrogant, like they’re never wrong. But their broader point was about timing. And I think they’re right. Timing involves a bit of luck.

You might be building the next big digital whiteboarding tool, like Miro or Mural. But if you were working on it in 2015, it wouldn’t have landed the same way. We weren’t in the remote work era yet. Then COVID hit, and suddenly the timing was perfect.

There are parts of this process that you can’t control. You’re never going to be perfect. But in general, here’s how I think about it. I find a problem, sketch what I think the first version looks like, and talk to a lot of people. I start iterating with a simple prototype, and while I’m doing that, I build a mailing list. Then I begin asking, “Can I reach these users? Does this connect to something deeper?”

On past products, pivots, product surprises, and market timings

Can you tell about a time when something seemed really promising, but you had to pull the plug — or maybe you didn’t, and looking back, you feel like you should have?

I’ve had experiences on both sides. Right out of college, I started a file-sharing company during the Napster-Gnutella era. We ended up shutting it down because we were young, worried about the legal stuff, and didn’t know how to take the next step. Everyone in the music industry was getting sued. And now I look back and think, if I’d had more experience, maybe that could’ve become something huge.

Same thing with a dating site I built called Matchstudents — this was before Friendster. It was meant for college dating. It was making money, but I think looking back at how Facebook basically started in a similar space, our product could have been bigger. If I had figured out some of the technical social networking challenges or understood how to raise money, who knows where that could have gone. So yeah, there have been times when I probably pulled the plug too early.

There have also been times when maybe I didn’t move on quickly enough. You build something, raise money, start shipping, and it’s not clicking. Then you face that hard decision. Do you pivot? I tell people all the time, maybe five to ten percent of startups are clearly working. If you’re in one of those, you just know. You’re growing like crazy, and honestly, you could be the worst PM on the team and still look good.

Then there’s another small slice that’s obviously not working — you just shut it down. But the majority sit in this middle zone where it’s kind of working. Your growth is linear, not exponential and in startup land that would be considered failure. You’ve got to grow faster, break through that next step.

I’ve been in situations like that. My last company started in meeting summarization. But I was focused on the wrong part of the problem. If you look at meetings as a matrix, internal versus external, and one-on-one versus one-to-many, the people who need the most help are external one-on-ones. Think sales calls, where it’s hard to take notes. The people who need the least help are internal many-to-many meetings, because there’s usually a note taker, and the content has a short shelf life. Within a week, you’ve moved on.

But I was head-down in the wrong quadrant. And the thing is, when you’re just two or three people, pivoting should be quick, like steering a speedboat. But we treated it more like turning an aircraft carrier. It took too long. Eventually we did pivot and landed in a better place, but it was tough. And now, post-ChatGPT, meeting summarization is hot again. So maybe the idea was early. Timing is hard to predict.

I do think luck plays a role. I say that knowing that 86 percent of venture-backed startups, ones that raised real money, still fail. And that’s with smart people around the table. So how does that happen? It just shows how uncertain this all is.

Has there been a product in the last few years that really surprised you by taking off? Something you didn’t think had legs? Or something you thought would catch on but it fizzled?

I’ve been astonished by the growth of vibe-creation products, apps that are all about mood, atmosphere, or simulated interaction.

I thought RemixAI, which was like a generative AI version of Instagram, would do better. Every post was a generated image. It was entertaining, kind of like what Meta is doing now with generative posts through Meta AI, but Remix never really made it.

Then there was SocialAI, which was like generative Twitter. You’d write the first post, and bots would carry on the conversation. I didn’t think that would take off at all. But it went super viral, though it got acquired before it could really scale. I just didn’t think people would want AI-simulated conversations because it’s obvious it’s AI. But then you look at something like Character.ai, where people are having multi-hour conversations with bots and actually forming relationships. That really surprised me.

Same thing with AI voice bots or AI sales chat. I was convinced people would always want to talk to a real human. But now I’m hearing that some folks actually prefer the bots. You can say no to an AI without guilt. You can ask anything. And it’s always available. So that flipped my thinking a bit.

Ironically, I always believed Waymo would win over Uber just because people don’t want to talk to a human in the back of a car. So maybe I got that one right.

The whole AI space right now is wild. There is so much experimentation, tourism really, and so much money flowing into it. Some of these things like Bolt, Lovable, the AR stuff — the growth metrics are off the charts. Whether any of them actually stick is still an open question.

How do you balance short-term versus long-term growth? How do you decide where to focus your energy early on to show quick success but also plan for sustainable growth?

Growth in 2025 is harder than it’s ever been, and 2026 will be even harder. This challenge keeps increasing partly because tools have become democratized. Generative AI has made building products much easier. You no longer need to deploy your own servers or manage complex infrastructure. This has caused the signal-to-noise ratio to skyrocket, making it harder to stand out.

When people talk about growth today, they often focus on paid acquisition, buying traffic, and that’s a whole world of its own. But organic growth remains crucial. Organic growth is about finding gray areas where you can arbitrage or take advantage of small opportunities. These hacks don’t last forever, but they let you discover and exploit new edges.

You have to break growth tactics down by their timeline. If someone talks about SEO or “answer engine optimization,” that’s an investment with a longer payoff, often six months or more. A great example is the LogRocket Blog, which takes multiple months before inbound clicks start turning into measurable results.

On the other hand, if you find a short-term opportunity like being an early app on a new app store or platform with few competitors, that can drive quick growth for a month or two. But eventually, that space will become crowded and won’t work anymore. You have to constantly evaluate where your efforts are paying off and decide whether to continue or shift.

Resources are finite, even though AI helps us do more simultaneously. You must measure effectiveness and recognize that a lot of growth is operational work. Creative execution is important too, but you need to be focused on finding step functions — big jumps in growth, not just slow, linear progress.

As a startup, linear growth is often no better than what the biggest incumbents achieve, which likely leads to failure. The only way to win is to find those uncomfortable, sometimes crazy step functions. That might mean unconventional marketing, controversial moves, or finding new communities to engage. Step functions, not linear steps, are the key to sustainable growth.

Do you think it’s possible for users to use a product the wrong way, or is it the product manager’s responsibility to understand what users are doing and meet them where they are?

This happens all the time. I had a calendar startup two startups ago, and the product’s thesis or ethos was to help prepare you for meetings. Today, this is all built into the native calendar app, but in 2011 or 2012, calendar apps on phones were stupid. They didn’t do anything. They were dumb. So, this product would tell you about the person you were going to meet, pulling emails and things like that.

We built all this stuff and had a strong point of view, but when we finally analyzed our audience and got to the point where we could do that kind of analysis and data, we realized 70 percent of our users were using it as a birthday calendar and for their doctor appointments.

Now, you have a choice at that point. You could say, "Should I build for my largest audience?" I often tell people you need to have a point of view and a belief about where this is going. We knew we couldn’t monetize that birthday calendar audience, but we basically agreed internally that this audience was great for driving word of mouth.

We could use the birthday calendar audience to drive metrics. They would drive more adoption, create more users, which increases our mailing list and gives us more people to talk to. Meanwhile, the audience we really thought we could monetize was the one where we were showing more value. This is what I call leaning into your differentiators.

People often want to build lots of things, but I always tell them, "No, focus on one or two differentiators." Everything else you should just assume is noise. Your entire UI should focus on those two differentiators. Even if you think other things are table stakes, bury them or don’t spend time on them.

We leaned into our differentiators, and ultimately the company was acquired by Salesforce.

Common pitfalls in early-stage startups

In early-stage startups, what are some of the traps that you see product teams and product leaders falling into?

There are a lot of traps, and they come from different places. It could be the product itself, the culture, the people, the startup’s funding situation. There are dozens, but a few come to mind immediately.

One of the most common traps is starting with the technology instead of the problem. You see this all the time, especially now with hype around things like generative AI or Web3. Teams jump straight into building with the tech without asking what problem they’re really solving. I always say, you have to ask yourself, “What is the job to be done? And do you even need fancy tech to solve it?” Sometimes the fastest way isn’t the flashiest way.

Then there’s the cold start problem. Many startups build products that require a critical mass of user data to become useful, like a personalized feed or an email search tool. But if you don’t have enough users or data at launch, the experience falls flat. People just don’t want to wait days for a product to “warm up.” That lack of early engagement can kill momentum quickly.

For consumer products especially, I see teams missing the importance of connecting their core experience to what I call a consumer vice. Borrowing from Maslow’s hierarchy, the product has to tap into something fundamental, whether it’s helping people make money, find love, get famous, or save cash. If your product doesn’t speak to one of these desires, it’s really tough to get organic word of mouth, which is crucial for growth now that a lot of traditional growth hacks and API exploits are closing off.

Speaking of word of mouth, I have what I call the WOMO test — word of mouth optimization. If you ask five different people what your product does and get five different answers, that’s a warning sign. It means your messaging or positioning is off, and that kills viral growth.

Another trap is teams building products without thinking about growth from the start. Sometimes product and growth are seen as separate phases, but honestly, growth has to be step zero. It’s something you plan for alongside your product from day one.

Lock-in is often overlooked as well. Products that don’t create switching costs are vulnerable. Scheduling tools like Calendly are easy to switch because there’s little data lock-in, while CRM platforms like Salesforce or HubSpot keep you locked in with tons of data and integrations. Without lock-in, retention and long-term value suffer.

Finally, I often meet founders who are super focused on UI, which ties back to the classic vitamin versus painkiller dilemma. It’s great to have a beautiful interface, but if you’re not solving a real problem, the job to be done, it’s just a shiny distraction. So I always encourage stepping back to ask if you’re truly addressing a need or are just enamored with design.

These are some of the traps off the top of my head, but they show how complex early-stage product work can be.

Storytelling and fundraising in product leadership

How do you feel product leaders should engage with fundraising and investor storytelling? Are there ways products either over-participate or under-participate in that side of the business?

I see two sides to this, especially when you think about zero-to-one product work in startups versus large companies, because those environments are totally different.

In a zero-to-one startup, especially in consumer, almost always the founder is also the product manager. This often continues for quite a while. Even in B2B startups, founders should ideally be the ones making the first sales before you develop a repeatable process. So storytelling is absolutely critical.

When I hear a story that doesn’t quite feel right, I ask people to take a step back and explain how things are done today, walk me through a day in the life, step by step. Then have them explain how the product changes those steps. Seeing the “before” and “after” visually side by side really highlights what user experience is being improved. But if they can’t clearly describe the current process, or if it’s done in many different ways, then I start to question if the problem is too broad or if it needs more focus.

Founders need to be great communicators because they’re the visionaries and craftspersons behind the product. The only way to drive their vision is through storytelling. And this isn’t just for investors. It’s for users, teams, stakeholders. The story evolves, but it always needs to come back to the job to be done. Sometimes people get so caught up telling the story that they forget what problem they’re actually solving.

The role and impact of generative AI in product

Where do you think generative AI is having the most immediate impact right now in product? And do you think there are places where it’s getting overused, and it’s a mistake to overuse it?

I like to encourage people to document their day in 10-minute chunks and then ask themselves, “Could I have used generative AI to speed up that part of my day?” It’s a useful exercise to start understanding where AI can really help.

Product management, in particular, has been one of the earliest and biggest beneficiaries of generative AI because AI essentially democratizes fractional expertise. I think we’re past the point of denial about AI improving productivity. There’s enough data now from companies across different individual contributor roles that shows a clear productivity boost. The next question is how to integrate it smoothly into your workflow.

About overuse, one interesting thing I’ve noticed is the rise of what I call generative AI native individual contributors. This raises bigger questions around education and skill development. For example, if someone can only write with AI assistance and not independently, or if they rely heavily on AI for fundamental math, does that create gaps in their knowledge? Traditionally, foundational skills have been taught and then complemented by tools like Excel. But with generative AI, this balance is shifting.

I’ve observed that these AI native contributors, especially among younger generations, tend to produce more noise in their work because they may lack the deep intuition or foundational knowledge. But they move quickly and execute well with AI tools, so it’s fascinating to watch how this evolves. Ultimately, I think it will drive a major rethink of education and upskilling.

At the same time, this dynamic is reflected in hiring, where new graduates often struggle because senior folks have more experience, taste, and foundational knowledge. Those senior contributors are effectively editors or AI prompt engineers. They don’t necessarily write every line of code or article themselves, but guide the AI based on their expertise.

I also think of Vibe as a completely new interaction modality. Just like the keyboard and mouse, voice commands with Siri, or touch on the iPhone, conversational chat is a new way to interact with technology. Every tool is now being reimagined as if it started with a conversational interface.

There’s been some resistance to generative AI because people say it doesn’t produce work as high-quality as a trained human. How do you see the role of humans when using AI for product creation? Are there places where AI can fully replace humans, or is human involvement always necessary?

I think the assumption that AI can work without a human in the loop is flawed. Across areas like AI-generated code, design, content, and social media, human involvement remains critical because it adds judgment, sensibility, and quality control. For example, AI-generated code might insert errors or irrelevant parts, so engineers now focus on writing instructions and reviewing AI output rather than coding from scratch. Writers use AI for first drafts but still need to outline and edit carefully.

This shift means the nature of the work changes, but humans are not going anywhere. Without human oversight, AI output likely will not reflect a company’s or product’s unique taste or standards. So far, I have seen that pairing AI with human review, especially by CMOs and product teams, works best.

Also, AI is not perfect. I use the 95 percent rule as a benchmark, meaning AI tends to be about 95 percent accurate. For many applications like writing, that is fine when humans edit. But for critical tasks such as automated meeting scheduling or self-driving cars, 95 percent accuracy is not enough. Mistakes in those areas can have serious consequences, so human oversight is mandatory.

Many startups have tried to fully automate workflows with AI, but the reality is AI can only get you partway there. You need humans to fill in the gaps, and that changes the return on investment calculations. So while AI is a powerful tool, human expertise remains essential for quality, safety, and success.

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.