R087 - Product distribution in the AI era, with Carmen Madrazo

What should we know about you?

Today I work at Factorial and I've been there for five years. My experience is relevant for what I do now, but also for everything I've lived through during this time. For those who don't know Factorial, it's a software that started as a solution for human resources and has become one of the fastest-growing startups in Spain and Europe. I joined five years ago and have experienced all that growth story from the inside, changing roles and seeing it all.

Now I lead Product Growth and Product Marketing, which means being between product and marketing, and it's something that, in one way or another, I've been doing for the past few years. But it's also important how I got here. I studied Civil Engineering and worked for years as an engineer. There are highways in Dallas and Chile that I worked on. And, at the same time, in my free time, I was tinkering a lot with digital projects. That's what eventually brought me to the product world.

How were those first digital projects?

They weren't big products, they were small projects, some better than others. It all started because, although I made a living designing roads, it wasn't something that completely fulfilled me. I was exploring a lot and had always been attracted to the internet.

I fell into the whole sequence of “create your business,” “live off passive income,” and that led me to try many things. I started with something similar to dropshipping between eBay and Amazon. I made some money and thought: it doesn't add much value, but you can make money online. Then I learned SEO, digital marketing, and how to build websites, because I also fell into the world of niches and affiliate websites. I even built a barbecue website, even though I didn't own a barbecue. I managed to rank it and sold it. Technically, I can say I had an exit.

Then I learned to program on my own. I spent months doing a course on Udemy to build my own website, a flight search engine for friends. Looking back, it amuses me because now I would probably do something similar in two days, but back then it took me six months. I also set up a kind of marketplace for restaurants in Santander: I went, sold it, launched it, and there were orders. And, at some point, I discovered that the product role existed. Until then, I didn't even know that world existed. I had no close references in startups or tech. When I made that click, I started sharing what I was doing, talking to people online, and from there, I ended up at Factorial.

What did you take from that stage to your work in product?

On one hand, I took the hard skills. Having touched marketing, even if it was basic digital marketing; having gone out to sell something face to face; having received yeses and noes; having connected business, product, and distribution. All that, for me, was like the MVP of a product manager. In fact, that's what I sold when I tried to get into product: maybe I hadn't had the formal role, but I had already done many of the pieces.

And, on the other hand, I learned to be bold. You have to know how to sell yourself. In the end, today I sell products, but before I had to sell myself. Basically, I opened up opportunities through direct messages on Twitter: “Hi, I'm Carmen, I do this, would you like to talk?”. That's how I started moving.

What does it mean to distribute a product?

Distributing a product is, at its core, doing marketing and sales. It's getting your product to market and having someone buy it. I think that, in recent years, the product role has moved much closer to the go-to-market. It used to be more focused on delivery, on making the software; now it's much more connected to selling it.

It also depends on the type of company. In companies with a lot of growth ambition, like Factorial, everyone thinks about distribution, ARR, and how to take the product to market. There, the line between product and distribution is much smaller.

How does the go-to-market change when you go from a few products to more than twenty?

It has a lot of complexity. The company has to scale to be able to build those products, create teams behind them, and support them. But, at the distribution level, the challenge is also enormous. In our case, each product has a business objective, product marketing managers, product managers, and salespeople aligned with selling it.

The problem is that when a client enters the pipeline, it's very difficult for them to buy all the products at once. They usually come with a specific need. You can sell them some things, but not twenty at once. So you have to think very carefully about what type of client each product is for, where you distribute it, who knows how to sell it, and how you organize the subsequent expansion.

Also, a very important part of the business is in the existing customer portfolio. We work a lot on the logic of land and expand: the client comes in with a set of products and, over time, we sell them more. There, the role of customer experience and the teams that work on the existing account is key. The more you grow, the more business potential you have within your own installed base.

What is your ICP?

Everyone. We sell to companies of practically all industries and sizes. That's why we need to identify different customer tracks.

How do you design those journeys for different types of clients?

It's something we've started working on much more this past year. We know, for example, that a consultancy or agency usually has project management needs, understanding profitability, managing travel and expenses. So, to that type of client, you can talk about project profitability, expense automation, operational efficiency, and put them in a specific track.

For other clients, you talk about something else. Conversion improves a lot when you can build campaigns and specific journeys for each type of client.

How do you manage to sell more products to the same account?

Another way to do it is to create product bundles. If you know that a certain type of client needs several solutions that fit very well together, you can package them. Also, one of the product's value propositions is that the more modules you have, the more information is connected, and the more value you get. That's why it makes a lot of sense to sell groups of products that work well together.

How do you decide between direct sales, partners, or having the product sell itself?

It depends on the product, the goal you have, and the type of client. When you launch a new product, you're usually still validating the product-market fit, the pricing, and the real ability to sell it. In that phase, the distribution strategy is usually simpler: a small team, made up of product, sales, and marketing, directly validates with clients if it can be sold and gets the first fifty or a hundred clients.

Once that works, you can scale and open more channels. In more mature products, with larger business objectives, you already better identify what type of client it is, what channel they're on, and what deserves prioritization. In our case, once the product is validated, we try to take it to all channels as soon as possible. We don't have it fully segmented by channel because, in general, we sell quite well through all of them.

How many people are there in sales?

I don't have the exact number right now. I would have to look at the organizational chart. In product and engineering, there can be between 200 and 300 people.

Is it necessary to have everything perfectly systematized to grow?

No. I think sometimes people think that to succeed in the market, everything has to be perfect: the playbooks, the processes, the sequences, the systems. And the reality is not that. Of course, there's a lot of work done, but it doesn't have to be perfectly polished to work.

Also, when you grow fast and everything changes so quickly, it's very difficult to have 100% closed processes or completely reliable systems. There's a lot of work behind it, yes, but also a lot of constant adjustment.

Have you killed products during this time?

Kill, kill, I wouldn't say yes. But deprioritize, certainly. We are very aware of the return on investment of the product and engineering team. That's why every quarter we do a product review to evaluate what's generating impact and what's not.

And many times, decisions are made to move resources from one thing to another because it's not generating the expected impact. More than killing products, what we do is deprioritize them in favor of others.

Why does branding matter more and more?

Because the easier it is to make a product, the more good competitors appear. Software is becoming commoditized. That doesn't mean the products are bad; it means that, even if they're very good, they increasingly resemble each other in certain layers. And there, the brand becomes a differentiator.

I compare it to more B2C logics: Apple and Samsung, Pepsi and Coca-Cola. When the product tends to look more alike, the brand matters much more. Also, at the same time, some traditional marketing channels are also being impacted by AI. Think about SEO, inbound or organic traffic. If those channels lose effectiveness, then word of mouth, trust, and branding become even more important.

Is SaaS dying?

I understand why that narrative exists. If we think of SaaS as a technology that makes you more efficient within a workflow, I do think there's a big change. We're moving from tools that help you work better to systems where you can delegate part of the work. For me, that is a paradigm shift.

Now, I don't think SaaS will die because every company will make its own CRM in-house. That's not realistic. Building a CRM or any complex software is much more than programming something quickly. You have to understand the problem very well, solve it well, maintain it, evolve it. I don't believe in that scenario. I do believe, however, that what we will offer as technology will change radically and that the potential value that can be provided to the user is much greater.

What did you learn when changing the pricing?

The first thing I learned is that knowing about pricing in SaaS B2B is very valuable and uncommon knowledge. I've tried to find profiles with experience in this, and it's not easy to find them.

The second is that pricing in SaaS is very different from pricing in other sectors. When I talked to people from e-commerce or airlines, I saw that there the price is very linked to margins and competitors. In SaaS, until now, the marginal cost of serving technology was much less relevant, so the price had much more to do with the value you provide to the client and how you can explain it.

And there's another big lesson: the success of pricing doesn't depend only on who designs it. It's defended by sales and customer experience. That's why, when we made a significant change, moving from products that were bought à la carte to packages designed for certain ICPs, a critical part of the work was training hundreds of people to understand the new model, agree with it, and be able to sell it with confidence.

What changes in pricing when AI comes in?

With AI, margins matter a lot again. Before, many workflows were resolved deterministically, and the marginal cost was small. Now, when a user takes an action and that calls an agent or a model, there's a clear cost from the AI provider, tokens, inference.

That makes cost matter again in the pricing conversation. But, at the same time, the value you can provide to the user can also be exponentially greater. So the equation changes quite a bit.

How do you take a product to market that is still being built?

It's the nightmare of any product marketer, but it's also very common, especially in fast-growing companies. Ideally, you would launch something fully validated with many clients, but that rarely happens.

In our case, the big example was One. We launched the big campaign in October last year, but throughout that year, the product was being built in parallel. It was a deliberate decision. We knew there was a trade-off, but we wanted to position ourselves as leaders of change as soon as possible, and also, the fourth quarter was a very important time for business.

To do it well, we stayed very close to the product team and relied heavily on understanding the client through surveys. We wanted to know what concerns they had around AI and how they were using it. There we saw that most people were already using AI in their daily lives, but almost always limited to ChatGPT. That helped us build the message: go beyond the generic chatbot, explain why an integrated solution has more value, and how AI can multiply your capacity. The campaign was complex because it had to be localized in five countries, included creative pieces, spots, and a live keynote event for clients and prospects. But I think it went very well.

What differences did you find between countries?

We mainly sell in Spain, Italy, Portugal, France, and Germany. In terms of AI perception, we saw quite a bit of homogeneity. In all countries, the same feeling appeared: people wanted to use AI, were already using ChatGPT, but didn't really know what they could do with it beyond that.

That's why a part of the campaign was very educational: showing use cases, tools, and more concrete possibilities. Where we did find some more differences was in privacy, GDPR, and legal concerns. In general, in Northern Europe, those concerns were more present.

How important is compliance in the pitch?

A lot, especially the larger the client. In enterprise it's a central issue. Also, now European regulations on artificial intelligence are appearing, so for sales, it's very important to have a good legal Q&A and clear materials so the client understands how those risks are managed.

In fact, an important part of the value proposition is there: not only in security or regulatory compliance but in the peace of mind of knowing that your company's data is not circulating outside your environment. That is a major concern.

Does the value proposition influence enterprise and SMB equally?

Knowing how to sell the value proposition is essential in any segment. When you don't know how to do it, you enter price comparisons and competitor wars much more easily.

What changes is that the client is different, the budgets are different, and often, so are the competitors. The value proposition remains the number one piece, but it must be very well adapted to the segment.

How are you using AI within Product Growth?

In general, the whole company is shifting towards a much more agentic mindset. We have to build agentic capabilities within the product and also offer agents to our clients. In Product Growth, which is dedicated to using the product's surface to generate more business, we are already doing very concrete things.

For example, throughout the product, we have what we call upselling points, usually paywalls. These are points where, when a user doesn't have a functionality, we explain why it would add value and lead them to a call with sales or a self-serve purchase. What we developed in a day-and-a-half hackathon was an agent that analyzes how those paywalls are performing. It connects with Amplitude, with product data, and evaluates whether it makes sense to keep them, remove them, or move them.

Also, if you want to launch a new product, that agent can suggest where in the product it makes the most sense to place those upselling points, crossing traffic, context, and value proposition. That can save us a lot of time and make decision-making much more efficient.

Why is onboarding so critical?

Because in SaaS B2B, activation during onboarding is decisive. If a client doesn't activate well, you have almost guaranteed churn. That's why there's a dedicated team and why it's such a sensitive process.

How are you automating it with AI?

We are working on it together with the onboarding team. There are parts that can be automated for the client and others for the team. For example, we are developing an agent that guides the user in the basic setup, which can interact with voice, understand what happens on the platform, and push certain actions.

There's also another part oriented towards the onboarding specialist team. If a client gets stuck at a step, today a person can write to them to reactivate them. That can be done proactively by an agent, whether within the product, by email, or even, in the future, by phone. The idea is that the team can focus on higher-value tasks or more complex clients.

Is there a specific team for this, or is it a cross-functional change?

Similar things are happening in many teams because it's a general mindset change. These specific cases are being worked on from Product Growth in collaboration with the One team, which acts somewhat as a platform to enable these types of capabilities within the product.

But the idea is not to have a single team doing AI for others. The real change is that each team has to know how to develop with AI and think agentically. That is already an expectation within the organization.

Is everyone already developing with AI?

I suppose there are different degrees. Some people are very advanced, and others are just starting. But I would say that everyone has to have the mindset of wanting to do it. Even designers and product managers have access to tools that allow them to program, prototype, and go much further than before. Always with engineering review when necessary, of course. But the change is already happening.

Since when have you been experiencing this change so intensely?

I feel it especially from January until now. In this last stretch, the change has been exponential.

Where does human judgment remain when agents are making decisions?

I see it quite similar to working with people. Recently, I heard someone talk about how they managed their agents, and it seemed like a very accurate comparison: in the end, you also onboard an agent with context, review what it does, give feedback, correct, adjust, and gradually gain confidence in its judgment.So human judgment is still in how you train it, how you define the context, and how you supervise what it does. If you simply let it do anything without review, then it does disappear. But well-used, I see it more as a team member than as an uncontrolled substitute.Where are the real frictions today when working with AI?

I think a clear friction is that roles have to change. If an engineer can get something to production three times faster than before, the product manager can become a bottleneck. So teams have to reorganize, and probably other roles have to take on more product work.

New bottlenecks also appear, such as code review or clarity on what to build. And, above all, there's the fact that AI's potential is constantly changing. Every week changes what you can do, how you can do it, and how far you can take an idea. That forces you to move very quickly while learning in real-time. It's demanding, but also very interesting. I don't think any company that wants to come out well from this change can afford not to be making that effort right now.

The conversation with Carmen leaves a very clear idea: building a product and taking it to market are no longer two separate worlds. In an environment where making software is increasingly fast and accessible, the difference lies in knowing how to position it, package it, explain it well, and deliver it with the right message.

And there's an even more relevant second layer. When artificial intelligence starts participating in

onboarding

, commercial expansion, activation, and even decision-making within the product, we're not just talking about new functionalities. We're talking about a new way of operating. The question, then, is not just what we build, but how we design systems capable of amplifying that value without losing judgment, focus, or trust. I hope you enjoyed the episode.A big hug, and I'll see you in the real world., en la expansión comercial, en la activación y hasta en la toma de decisiones dentro del producto, ya no estamos hablando solo de nuevas funcionalidades. Estamos hablando de una nueva forma de operar. La pregunta, entonces, no es solo qué construimos, sino cómo diseñamos sistemas capaces de amplificar ese valor sin perder criterio, foco ni confianza. Espero que os haya gustado el episodio.

Un fuerte abrazo y os espero en el mundo real.