Signal in the Noise: A 6-Month Experiment in Finding the Right Startup
Why the most rigorous approach to my job search was learning when to abandon rigor
I made a spreadsheet last year when I was evaluating which startup to join. This approach fell apart, and I’m writing this to reflect on that and detail what ended up working better for me.
I learned my biggest lesson while sitting at a bakery with two founders from MIT and Stanford, both with prior exits and a technical vision that made sense on every level. I’d scored their company a 9.7 out of 10 on my prospective job spreadsheet.
Yet as we talked through their go-to-market strategy and engineering roadmap, I felt something I hadn’t anticipated: absolutely nothing. The conversation was professional, thorough, and correct. They answered every question I’d prepared. On paper, this was the company. In the room, it felt like reading assembly code: technically accurate but somehow missing the point.
I realized then that I’d been treating my job search like an optimization problem when it was actually a signal processing problem.
Finding the right fit isn’t about collecting more data. It’s about recalibrating what you’re listening for. I built the wrong framework, watched it fail, and discovered what actually matters when joining an early-stage company.
The Initial Hypothesis: Engineering a Decision
Like any good engineer, I started with a framework. Above is a screenshot of the actual spreadsheet I had made.
After deciding to leave my last company, I approached the search the way I’d been trained to approach problems: systematically, rigorously, with clear evaluation criteria. The people I trusted most who hire and mentor told me that not jumping would be the surprising move. I took their advice.
The methodology looked sophisticated. I mined VC portfolios: Y Combinator’s latest batch, Andreessen Horowitz’s enterprise bets, Pear, Khosla, Sequoia, Menlo, and the sleeper companies buried in Founders Fund’s diverse portfolio. I subscribed to curated newsletters: The Generalist’s deep dives, Ben’s NextPlay (which is awesome), Lenny’s Product Jobs, newsletters from founders I admired. I built a spreadsheet that would have made even my consultant brothers proud.
My categories were comprehensive: Technical strength (engineering team pedigree, architecture decisions, tech stack modernity). Market opportunity (TAM, competitive positioning, timing). Funding runway (burn rate, months of runway, investor quality). Team culture signals (Glassdoor sentiment, LinkedIn tenure patterns, founder Twitter presence). Learning potential (complexity of problems, mentorship availability, scope for growth).
Each company got a score. Each criterion got weighted. I had a formula that determined which ones I actually chased hard.
This actually worked, at least on the surface. I got interviews—serious ones. I advanced through technical rounds. I did a week-long work trial at a developer tools company, pair programming with engineers who’d worked at Meta and Stripe. The system was producing outputs. Companies were responding. I had traction.
What I didn’t realize yet was that I was optimizing for the wrong variable entirely. I was measuring everything that could be measured while ignoring everything that mattered to me.
The Perfect Score That Felt Wrong
The company I mentioned earlier (the 9.7) was legitimately impressive on every metric. Their founding team had built and sold a company before. Their seed investors included names you’d recognize. The engineering team came from places like Databricks, Snowflake, and AWS. They had a working product, early revenue, and a clear path to scale.
They checked boxes I didn’t even know I had.
The meetings were professional, well-organized, efficient. The interview was challenging but fair. I met the team over coffee, toured their office, and sat in on a planning meeting. Nothing was wrong. That was the problem…
There was a hollow feeling I couldn’t name. Like watching a movie with the audio slightly out of sync: nothing overtly broken, but your brain keeps screaming “off.”
I kept returning to my spreadsheet as if more analysis would resolve the dissonance. How do you justify turning down a company that scores 9.7? What category captures “the vibe is wrong”? What weighting system accounts for “I don’t want to hang with these people outside of work”?
The spreadsheet was measuring legibility, not fit. I’d been evaluating what could be articulated (team pedigree, market size, funding status) while completely ignoring what could only be felt: energy, communication style, shared obsessions, the ineffable quality of whether these people wanted to build something together or just wanted to have built something.
It’s the same problem dating apps face: all the information is theoretically available, but context collapse strips out everything that matters.
It’s the same problem dating apps face: all the information is theoretically available, but context collapse strips out everything that matters. You can know someone’s height, job, and favorite books without knowing whether their laugh makes you want to tell more jokes. You can assess a startup’s metrics without knowing whether their problems are the kind that would keep you up at night in a good way.
I turned them down. It felt crazy. They were perfect on paper. It also felt right in a way I couldn’t articulate yet but was learning to trust. The decision haunted me for a week. Then I stopped thinking about it entirely, which told me everything.
Recalibrating the Instruments
If my spreadsheet was capturing the wrong signals, what should I be listening for?
The pivot wasn’t abandoning structure (I’m still an engineer at heart) but changing what I was structuring. I shifted from optimizing for data volume to optimizing for signal clarity.
From: Cold outreach to portfolio companies, scraping AngelList for “AI infrastructure” startups, responding to newsletter job postings, treating my search like a sales funnel.
To: Events, conferences, founder meetups, warm introductions, treating my search like relationship building.
The difference was profound. I think minutes in person reveal more than five rounds of video interviews. At an AI infrastructure meetup, I could observe how founders talked to each other when they thought no one was listening. At smaller technical talks, you get unfiltered takes from engineers already working at interesting companies: what frustrated them, what excited them, whether they’d join again knowing what they know now.
The research backs this up in unexpected ways. During this process I was reminded of Mark Granovetter’s work on “weak ties,” showing that most job opportunities come through acquaintances, not close friends. But here’s what he didn’t emphasize: those weak ties are only valuable if they’re real ties. A LinkedIn connection isn’t a weak tie; it’s no tie. A thirty-minute conversation after a meetup about the challenges of distributed systems? That’s a weak tie with signal strength.
The network effect compounded quickly. Each event led to introductions, which led to coffee chats, which led to “you should talk to my friend who’s building something in that space.” The search became self-reinforcing not through volume but through depth.
I started noticing patterns in what consistently excited me:
Founders who talked about problems, not solutions. The best conversations weren’t about their clever architecture or elegant API design. They were about the frustrating problem they couldn’t stop thinking about. One founder spent twenty minutes explaining why current approaches to a technical problem were fundamentally limited, barely mentioning his company until I asked directly. That obsession was the signal.
Teams that finished each other’s sentences. Not in an obnoxious “we’re so aligned” way, but in the way old bandmates play together. Someone starts a thought about a technical challenge and someone else completes it, builds on it, takes it somewhere unexpected. I think a classic Silicon Valley example would be Figma: Dylan Field and Evan Wallace had it. You can’t fake it, and you can feel it immediately. I ran into some unique teams that seemed to understand each other on a level others would never achieve.
Companies where technical challenges were the fun part. Some startups treat engineering as a necessary cost to reach product-market fit. Others treat engineering as the whole game. I wanted the latter. When engineers got excited showing me their nastiest debugging session or their most elegant hack, when “hard” meant “interesting” rather than “annoying,” that was signal.
Environments where people stayed late because they wanted to, not because they had to. There’s a visceral difference between a team grinding toward a deadline and a team that can’t stop talking about the problem even after the standup ends. The former looks the same as the latter on LinkedIn. In person, they feel nothing alike.
My framework evolved, but it wasn’t quantitative anymore. The criteria became qualitative and more predictive than “What’s their ARR?” “Would I be embarrassed to tell my friends I work here?” became more honest than “What’s their brand name recognition?”
Malcolm Gladwell calls this “thin-slicing” in Blink: the ability to find patterns in narrow slices of experience. But thin-slicing only works when you’re sampling the right dimensions. My spreadsheet was thin-slicing the wrong variables entirely.
The Network Effect: How Relationships Became the Algorithm
I first ran into my current company at an event called Founders You Should Know. A mentor steered me there originally because one of his friends had gone. The Founders You Should Know Showcase is a monthly event where a curated group of early-stage startup founders present their companies to a vetted audience of engineers, operators, and job seekers. Each founder gives a short pitch focused on why someone should join their team, followed by time for questions and networking. The event is designed to surface high-signal opportunities by connecting ambitious talent directly with founders who are actively hiring. Unlike traditional job fairs, it emphasizes storytelling, authenticity, and community trust to match people with startups where the fit truly matters.
FYSK gave me a free hoodie for joining a company that was at their showcase!
Brooke was there to give a lightning round pitch of Coval to the pool of engineers attending. I immediately took note when she mentioned she had come from Waymo, as we both shared a self-driving background (myself at Zoox).
Funny enough, this story doesn’t go as you expect. I didn’t even get the chance to talk with her at the event. I also wrote off the company because at the end of the chat she mentioned they were looking for senior engineers. It took a couple months for me to happen to be at a voice AI event at Coval’s office, where I got to chat with some of the team after. On the way out, my quick thank you to Brooke turned into an invite to stop by for lunch and meet the team.
The even more interesting part of this is that I had not gone to this event for my job hunt. It was just relevant to a side project I was working on.
That lunch turned into an invite to an interview. After a couple rounds of interviews, I got an offer for a work trial. I mentioned earlier in this article doing a work trial, and I want to pause for a sec to highly recommend this. I also made a dating analogy earlier, which is the most familiar concept to job hunting for a startup in my mind.
Because the company is so much smaller and intimate, it really matters that it is a good fit in terms of work and people. Would you go on one date with someone and then sign on to a multi-year relationship? I felt in big or medium tech, if there are coworkers or team members you don’t like, you can easily steer your work and projects and meetings to avoid them. There is no avoiding anyone in a five-person startup like Coval.
The opportunity to work trial and get past the introductory pleasantries is the best safety layer I found to getting past this, and it is mutually beneficial for both parties. Though, admittedly, one day into my week work trial I was sure this was the place for me.
The decision process was surprisingly simple. Not because the choice was obvious, but because I’d finally tuned into the right signal.
My spreadsheet couldn’t capture:
The founder who’s simultaneously a badass (one of five solo founders in her YC batch) and genuinely one of the most down-to-earth people I’ve met
How incredibly patient, curious, and helpful the founding engineers are, who spent hours answering my often basic questions
The way the team reacted to me taking down the entire pipeline in my first week: no blame, but instead a simple shrug and “thanks for fixing it yourself”
The transparency of every aspect of the company and culture. Many companies tout this but few execute real honesty
The efficiency of warm introductions compressed what could have been months of evaluation into weeks. Not because I skipped steps, but because the signal-to-noise ratio was finally right. When someone I trust says “you should talk to them,” they’re not just making a connection. They’re pre-filtering for values alignment, culture fit, the intangibles that no spreadsheet captures.
This is why platforms like On Deck’s talent network are so many steps better than cold adding on LinkedIn: they’re not just databases of people, they’re trust networks with built-in signal amplification.
The Broader Insight: What Early-Stage Companies Actually Require
Here’s the paradox that drove me crazy: the “inefficient” approach (networking, qualitative assessment, trusting gut feeling) led to the most efficient outcome. Once I knew what to listen for, finding the right fit happened quickly. But learning what to listen for required abandoning the illusion of rigor.
This matters specifically for startups in ways it doesn’t for established companies.
At Google or Meta, you’re joining a system. The company has processes, hierarchies, established cultures, defined roles, an org chart designed to parse down the complexity of your day-to-day responsibilities. You can evaluate it externally because what you see is roughly what you’ll get. The system is larger than any individual relationship.
At a twenty-person startup, you’re joining people. Success depends entirely on intangibles: communication style, trust, energy, shared obsession with the problem. These things can’t be measured in spreadsheets because they emerge from interaction, not inspection. You can read about a company’s culture all day; you won’t know if you fit until you’re in the room watching how they work.
A spontaneous half marathon with some of the Coval team
This is what makes startup hiring so notoriously difficult and why founder-market fit matters so much. The efficient market hypothesis breaks down when evaluating people. There’s no price discovery mechanism for “do these humans work well together?” You can only find out by working together.
Daniel Kahneman would call my initial spreadsheet approach System 2 thinking: slow, deliberate, logical. What I needed was System 1: fast, intuitive, holistic. Not because rigor is bad, but because the variables that matter in early-stage environments resist quantification. You can’t think your way to culture fit; you can only feel it.
The strongest signal isn’t in what a company says about itself (their vision, their values, their “why we’re different”). It’s in how you feel when you’re in the room with them. Do you want the meeting to continue or end? Do you find yourself thinking about their problems later that night? Do you feel energized or drained?
These aren’t soft skills or fuzzy metrics. They’re the hardest, most honest data points you have. Your nervous system is an incredibly sophisticated pattern-matching machine, running on millions of years of evolutionary optimization. When it tells you something is off, listen. When it tells you something clicks, listen harder.
What This Means If You’re Searching
If you’re looking for a startup role right now, here’s what I learned:
Build your signal processing skills, not just your search skills. Go to events. Have coffee. Join founder communities. Not because you’re “networking” in the gross sense, but because you’re calibrating your instruments. You’re learning what good looks like, what fit feels like, what problems make you lean forward instead of check your phone. You can’t vibe check anything if you don’t know how to pick up on good vibes.
Trust qualitative data. Make your spreadsheet if it helps you feel organized, but don’t let it override what you feel in the room. If something seems perfect on paper but wrong in person, it’s wrong. If something seems risky on paper but right in person, keep exploring.
Optimize for warm introductions. Your next opportunity probably won’t come from a job board. It’ll come from someone who knows you and knows them saying “you two should talk.” Build those bridges before you need to cross them.
Remember that early-stage companies are relationships, not evaluations. You’re not trying to pass their test or have them pass yours. You’re trying to figure out if you want to spend 70+ hours per week solving hard problems together. That’s a fundamentally different question.
The most rigorous approach isn’t always the most systematic one. Sometimes it’s knowing when to abandon your framework and trust the signal beneath the noise.
There’s a lot I’m leaving out here that I’ll need to revisit in separate reflections. For one: this approach of building genuine relationships is one of the only ways to actually reach real founders. The great ones rarely have time to read cold DMs. The ROI is vanishingly low compared to warm introductions and real conversations. The same logic that applies to us job seekers applies to the hiring party. Brooke once received a TikTok as a job application. There is no shortage of incoming interest for these founders.
What’s also under-stressed in this article is that the best connecting glue wasn’t just that I was looking for a job. It was that I was building and genuinely excited about the space. This curiosity, this knowledge, this excitement made so many of my connections more genuine and gave me much more to relate with founders about beyond “I’m looking for a job” and “we’re hiring.” My building on the side is a story I’ll cover in a separate reflection because I learned so much from it that I’d like to dedicate proper space to unpacking those lessons.
Six months felt long while I was in it. Looking back, I wasn’t just searching for a job. I was learning a new language. Learning how to listen for what matters instead of what’s measurable. Learning that fit isn’t something you can evaluate from a distance; it’s something you discover through proximity.
That’s not wasted time. That’s the whole point. Learn to listen for a different frequency entirely.
Want to chat or follow along as I continue building in the Voice AI space? You can find my other writing, projects and contact at www.loren.fyi, or just subscribe to my substack.