Operator Series: Nick Lafferty

I sit down operator to operator with Nick Lafferty, one of the first people to hold the Marketing Engineer title, to pressure-test what marketing engineering actually is.

I love doing this. Sitting down with an operator I trust, hitting record, and getting the real version of how they actually work. I did one with George from Ramp not long ago. Nick Lafferty was the one I wanted next.

Marketing engineering is exploding right now, and Nick Lafferty is one of the first people to hold the title. I’ve worked alongside him for about a year, and his work is some of the sharpest I know. So when I wanted the real answer on this role, he was the only call.

Here is the thesis I brought in. Marketing is not becoming engineering. The execution layer is. Building the pipeline. Cleaning the data. Running the agents. Winning the citation. That floor is getting engineered fast, and it is not going back. The judgment layer is a different story. What to build, what it should say, which bet is worth the chips. As the floor gets cheap, judgment absorbs the value.

I wrote the playbook for this role in April. Then Profound made it real. Idea on stage in April. A person in the seat by June. All in the open. The category caught up after: Figma, Plaid, Cloudflare, Ramp…all these companies are investing in the role. There is now a job board just for it.

Before Profound, Nick led growth at Loom and Mailgun through two acquisitions and spent six million dollars on Google Ads in a single year. Then he turned down Head of Marketing to build the seat he actually wanted.

So I hit record and asked him to show me the real thing. Operator to operator.

The spine

You ended up in a seat that did not exist yet. Nick, you turned down Head of Marketing to build it. Take me back to the moment the normal next step started to look like the wrong one. What did you see that the people around you did not?

I wanted to optimize my career, and my life, towards maximizing joy and fulfillment. It sounds simple but I knew having a head of marketing role wouldn’t make me happy. Sure the title feels good and it looks good on LinkedIn, but the actual work would make me miserable. I’m at the point in my career where I can make choices like that. So even though I knew what I didn’t want, the next step wasn’t immediately clear. I went to James, our CEO, and gave him a heads up, but I didn’t know what came next.

A few months later, this idea of the marketing engineer started to emerge internally and I was literally the perfect person for it. By turning down a job I knew wouldn’t make me happy, I ended up in exactly the right place.

I wrote that this role owns output, not pipes. That it closes the loop alone. Identifies the problem, builds the thing, ships it, measures it, kills it or scales it. So show me one. Walk me through a single system you shipped in the last month that a traditional team would have turned into a brief, a meeting, and three people. What did it replace, how long did the old version take, and what is the number that proves it worked? I want it specific enough that I could rebuild it from your answer.

I’ll give you a very meta example. We launched a podcast recently called The Marketing Engineer. I’ve interviewed Heads of Growth at Wispr Flow, Figma, Samsara and about 5 more companies that haven’t come out yet. We have this incredibly rich and content dense source material, in the video recording and the transcript. The smart play is to repurpose that content into 10-12 different formats across LinkedIn, Twitter, and Youtube.

So I built a system I’m calling the Profound Podcast OS that does exactly that. It takes the transcript and a bunch of shared context Claude skills for Profound’s tone of voice, our writing style, our social media guidelines, and a file for how I like to write in my own personal voice on social.

It helps me draft all the content we want to repurpose across all channels and then stages them in our social accounts for review by my team. It also sends a short Slack message to our internal Sales and Customer teams, giving them a very short summary of the episode and why it matters, so they can share with prospects and customers.

But I didn’t want to stop there. For each episode I write out questions ahead of time, and I wanted a way to analyze how good I was at asking those questions and getting the right response. So within this whole Profound Podcast OS there is a coaching section that analyzes my questions, compares them to the transcript, and then grades me on dimensions like Talk Share (how often I speak vs the guest), Improv ability (how well am I following conversational threads), and Specificity (am I getting exact details).

It runs that across every episode I’ve recorded so far and I can see improvements in some areas, and still some areas I need to work on.

Here is a thought of mine I want you to confirm or break. The tooling is the paintbrush and the marketing engineer is the artist. That a beginner with a great brush still paints a beginner’s painting. That someone who can only build becomes a backlog and someone who can only imagine becomes a PM. In your seat, which half is driving the work, and was I right that the taste has to lead, not the system?

Mostly right, with one correction. The artist framing works well. A beginner with Claude and every tool in the world still ships beginner work, because the tool will happily produce a confident, average thing. But average isn’t good enough.

The reason I can look at an output and know it’s wrong in four seconds is that I’ve shipped the wrong version of that thing before and watched it underperform. So much of that is driven by my experience of frankly screwing a lot of things up over the years. Someone who only imagines never earns that calibration, so their taste stays theoretical and soft.

So in my seat the judgment leads, yes. But judgment is a muscle you only grow by having your hands on the work. The pure builder becomes a backlog and the pure imaginer becomes a PM. The person who’s actually dangerous built enough to develop real taste and then started using it to kill things.

You ran the old playbook at a high level. Paid, funnel math, attribution, the whole thing. What part of that training actively misleads you now? Where does the instinct you spent a decade building still pull you in the wrong direction inside this seat?

The instinct I need to fight, and I think most marketers right now do too, is trying to do all of the tasks you used to do manually. Building campaigns, executing on individual work. That’s the kind of thing that does not scale and is the opposite of thinking in systems, which is how a marketer in 2026 should operate.

Instead of rewriting your ad copy manually for Google Ads, build a system that follows all your best practices, downloads your ad copy from the Google Ads API, makes changes, and then pings you for review. This can be a skill in Claude or something you code into a system with Claude Code or Codex.

It’s the systems thinking that every marketer needs to adapt right now.

I have a strong opinion about where this role belongs. It owns what it builds and influences everything else. It reports to whoever owns the GTM number, and it dies three levels down. You are the test case. Where do you actually sit, did the org try to hand you a single lane to own, and where does the rest of the team still not know what to do with you?

I spoke about this in my talk at Zero Click NYC this year, but we strongly believe marketing engineers need to sit within the marketing org and reporting to your Head of Marketing, VP, or CMO. That’s exactly how it sits at Profound, we have two associate marketing engineers that sit underneath me as well. I run twice-weekly standups with them, acting almost as an engineering manager, to help triage new tasks and direct them towards what are the highest priority areas to build in right now.

For the reader who wants to become this, or hire it. Output is cheap now, and the machine scales the average, so the demo everyone shows up with is no longer the signal. I wrote that the best ones are boring about the technology and precise about the output, and the worst are performing AI instead of using it. What separates a person who can do this job from someone who has read every thread and built one impressive thing? What is the part you cannot teach?

You really need someone with deep marketing experience first. They need to know what not to do and have a refined sense of taste and judgement from a career in marketing. Then you can layer on the AI builder skills to give them superpowers.

Not the market forecast. You have your hands in this every single day, which means you see things before the tooling and the rest of the industry catch up. What can you see coming from inside the seat that almost nobody outside it has clocked yet?

I think AI systems will quickly move from running locally on your computer into the cloud. We’re seeing this movement with Anthropic’s new Slack bot called Claude Tag. Everything will move into the cloud which in many ways makes a marketing engineer’s job easier – the hardest transition right now is going from a single player build (building something that works for you on your local machine) into multi player build (scaling that system so your whole team can use it).

We’re still in the very early stages of watching all of this play out.

You have a phrase for the build-versus-buy call: tolerance for jank. I want the unglamorous version. Where is the line between a janky system that works and a janky system that quietly poisons your GTM six months later? Give me a time the jank actually cost you something.

I think a lot of internally vibe-coded tools will need to be rebuilt in 6 months! And I think that’s ok, it needs to solve a short-term need now and then you can reevaluate down the road, because needs change and priorities shift and there is a real cost to maintaining these things that most marketers aren’t used to. This goes back to my singleplayer vs multiplayer analogy earlier, once you start building something that other people use internally you have a real responsibility to maintain it for all your users. And maintenance kind of sucks.

I think it’s ok if some internal systems are short-lived. I built a tool that syncs Email Unsubscribes across a few different marketing automation platforms we have. It works great now but we will eventually consolidate those systems into one, which is the right long term call. So my system is a band-aid by design. It didn’t take long to build (2-3 hours max), solves a real time suck for my team now, and gives us a bridge to figure out the right email system long term. Everyone wins.

Profound is building the marketing workbench for superintelligence, which means you cannot hide behind theory. You have to win the thing you sell. The Reddit agent that hunts for threads likely to become future citations. The engine that turns every old webinar into something built for how AI reads. Without me feeding you the numbers, what has building produced that headcount never could, and what is the one metric you watch that your peers are not?

Building these systems enables our entire marketing team to keep up with the massive growth Profound is seeing. Every marketer here needs to be a force-multiplier for the business, and AI systems help 10x every marketer who works here. Most people don’t know this, but through the end of 2025 there were only 4 marketers on the team here, and we kept hearing “Profound is everywhere”. That kind of scale was only possible because of the systems we built and because we hire world-class people. We’re still hiring, by the way.

The takeaway

The floor is getting engineered whether you like it or not. Your edge is not the tool. It is the calibration to look at an output and know in four seconds that it is wrong, because you shipped the wrong version once and watched it die. You only earn that by keeping your hands on the work.

So start there. Pick one manual task you repeat every week. Build the system that kills it. Ship it janky. Then go build the taste to know what to kill next.

Be on the lookout for more marketing engineer content. I will be hosting an upcoming event in a few weeks. Details to come.