CRO Life: from a $20B Exit, to a Hyper-Growth AI Org (Michelle Donnelly, CRO @ Crescendo)
The Builder’s Playbook: Michelle Donnelly on Career Bets, Product-Market Fit, and the Fundamentals That Never Change
“I’ve never been strategic about my career. It’s a weird thing to admit as a CRO.”
That’s Michelle Donnelly, who spent a decade at Salesforce, then jumped to a pre-revenue AI chip startup that most people said would fail. It sold to Nvidia for $20 billion. Now she’s running revenue at Crescendo, a company that hit $100M ARR in under two years. Her secret? She followed the chills.
Not metaphorical chills. Literal ones. Every conversation with Groq’s founder Jonathan Ross gave her goosebumps. “Half the time I had no idea what he was saying. The guy is a genius. But I would get the chills.” So she jumped. Hardware. Pre-revenue. No product-market fit. Her brother worked at Switch, one of the largest data centers in the country, building infrastructure for OpenAI and Microsoft. His response: “Why are you going to do that? None of these startups win.”
He was almost right. What Michelle learned at Groq about building, breaking, and rebuilding nearly ended her. It also made her one of the most battle-tested revenue leaders in AI.
I wanted to talk to Michelle because she’s lived through contrasts most of us only read about: pre-AI and post-AI selling, big public company and pre-revenue startup, hardware and software, Salesforce’s legendary go-to-market machine and the chaos of creating a category from nothing. What’s changed? What hasn’t? And what do you actually do when the market moves faster than your org chart?
Here are five lessons from our conversation.
1. Your Network Is Your Oracle
Michelle didn’t find Groq on a job board. She found it through two phone calls.
The first was to Jessica, a friend who worked as the right hand to Deloitte’s CEO. “What’s next?” Michelle asked. Jessica’s answer: “Come to Deloitte. But if you don’t come to Deloitte, you need to go to Nvidia.” This was 2022. Pre-ChatGPT. Jessica saw what Deloitte’s largest customers were doing with GPUs and knew the wave was coming.
The second call was to Carla Stratfold, who had run global sales at AWS. Carla had just turned down the CRO role at Groq because she needed a break after a decade of hypergrowth. “You should talk to them.”
Two calls. Two people who thought about the future professionally. That’s it.
“This is the power of your network,” Michelle told me. “The future thinkers, the ones who really have a tap on what’s next.”
Mike Maples Jr., the legendary seed investor behind Twitter, Twitch, and Lyft, has a framework for this in his book Pattern Breakers. He argues that breakthrough startups are built on insights: non-obvious truths about how emerging inflections can be harnessed to change human behavior. The key word is non-obvious. If everyone sees it, the opportunity is already priced in. The path to outlier outcomes is being non-consensus and right.
But here’s the problem: insights don’t come from market research or trend reports. They come from people who are “living in the future”—immersed in a space deeply enough to see what’s missing before anyone else notices. Jessica wasn’t guessing about AI. She was watching Deloitte’s biggest enterprise customers pour money into GPU infrastructure. Carla wasn’t speculating about Groq. She’d spent a decade scaling AWS and knew what real compute demand looked like.
Most people use their network to find jobs. Michelle uses hers to find insights. There’s a difference. Job-seeking networks optimize for openings. Insight-seeking networks optimize for non-obvious truths about where the world is going.
Company picking has never mattered more. CB Insights data shows 42% of startups fail because they build something the market doesn’t need. The divergence between good and great outcomes is widening. Zero to $100M in two years is becoming common for the winners. For everyone else, the math is brutal.
Michelle’s filter combined intuition with sourced insight: “Every time I would talk to someone at Groq, I would get the chills. The people were incredible. The board was incredible. There’s something really special here.” Chills aren’t a strategy. But chills informed by people who live in the future? That’s a leading indicator worth following.
2. Product-Market Fit Is Binary (And Painful)
At Groq, Michelle built the enterprise sales organization from scratch. Public sector team. Commercial team. Finance and telecom verticals. She knew CX would be a huge market for real-time voice AI. She hired transformational leaders, experts in their crafts.
Then the market moved.
Meta released Llama. ChatGPT launched. Suddenly Groq could show side-by-side demos of how much faster their chip ran open-source models. Developers flooded in. The company pivoted to a developer-led motion overnight.
“Everything I built was gone,” Michelle said. “I actually left because I’m like, you guys don’t need me. You need an inside sales leader to do this developer stuff.”
This is the part of startup life nobody romanticizes. You can hire brilliantly, build systematically, and execute with precision. If product-market fit shifts, none of it matters. Your org chart becomes a memorial to a market that no longer exists.
Michelle calls it the hardest lesson of her career. And it nearly broke her. But it also gave her a non-negotiable for what came next: “If you didn’t have product-market fit and you couldn’t show me the number of customers that were buying and how customers were increasing their spend with you, I didn’t want to work with you.”
The research backs her paranoia. According to CB Insights, lack of product-market fit is the single biggest startup killer. Failory’s analysis of 80+ failed startups found 34% died because they built something the market didn’t want. And here’s the cruel part: 65% of failed startups built their product before validating the market. They got the sequence backwards.
Groq eventually found its fit. The developer motion exploded. Nvidia bought the company’s assets for $20 billion in December 2025. But Michelle wasn’t there to see it. She’d already moved to Crescendo, armed with scar tissue and a new rule: product-market fit first, everything else second.
3. Simplicity Is the Hardest Go-to-Market Problem
Michelle walked the floor at NRF, the largest retail conference in the country. Every booth was screaming about AI agents. Every pitch sounded identical. “It’s so noisy. It’s so confusing. It’s really hard to differentiate.”
She felt what her customers feel: overwhelmed, skeptical, and unsure who to trust.
Crescendo’s answer isn’t a better feature list. It’s radical clarity. “If you can’t understand who we are in 30 seconds, we have hashtag failed.”
Getting there has been a huge, focused effort. They’re on their fifth pitch deck in six months. They’ve tested messaging with BDRs doing 400 calls a day, run it past existing customers who told them the pricing was too complex, brought in outside consultants (some helped, some didn’t), and ripped apart their positioning in real-time workshops. “It took 20 minutes for customers to get the aha moment,” Michelle said. “That’s way too long.”
The research confirms what Michelle learned the hard way. Gartner found that B2B buyers spend two-thirds of any buying journey “gathering, processing, and de-conflicting information.” The buying process has become, in Gartner’s words, “nearly unnavigable.” What customers actually value? Suppliers who make the purchase process easier through the right information, through the right channels.
Simple sounds easy. It’s the opposite. “Simplicity is actually really hard to get to,” Michelle admitted. “I came from Salesforce. The messaging was amazing. But this is complex.”
Her forcing function: three priorities. Not four. Not five. “Everybody in my org can repeat to you on call, what are the three things we’re doing this quarter?” When your CEO has a thousand ideas a minute (Michelle’s does—he uses Whisper to track them all), someone has to be the filter. That’s the CRO’s job. Understand the vision, translate it to a number, simplify it for the team.
4. Show, Don’t Demo
In a low-trust market, everyone claims to be better. Michelle’s response: stop claiming, start proving.
When a gaming company called on a Monday needing help before the Super Bowl, Crescendo didn’t schedule a demo. Their head of product built a custom AI agent using the company’s publicly available knowledge base. Within 24 hours, they could show the prospect how their specific use case would work—voice and text—on their actual content.
“Wouldn’t you rather see an experience customized to how I’m going to solve your pain at Owner versus seeing a demo for the restaurant industry?” Michelle asked me. “You want to see how it works for you.”
This is only possible because AI has collapsed the build time. What used to take weeks of custom development now takes hours. Crescendo uses that speed advantage not just in their product, but in their sales motion.
They’ve extended the “show me” philosophy further. Michelle tells prospects to go become customers of Crescendo’s existing clients. “Go to Lovepop and buy a card. See how they upsell you to buy more cards for Mother’s Day and Father’s Day. Go check out Doc Martens and see how they increase your shopping basket.” One prospect went to test a client’s experience and hated it—the AI didn’t authenticate her as a gold member. Michelle’s team turned it into a teaching moment: “They chose not to do that. If you want to be authenticated, let us show you how.”
The underlying principle: in noisy markets, tangibility beats promises. Generic demos create generic trust. Custom proof creates specific confidence.
5. Talent Density Beats Headcount
“All of my BDRs are athletes. Every single one of them.”
Michelle hires grinders. Crew. Swimming. The sports that just suck. “We go after people who are grinders in this type of company.”
This isn’t random pattern-matching. It’s a deliberate bet on talent density over headcount. Reed Hastings coined the term at Netflix after the dot-com crash forced them to lay off a third of their workforce. What happened next surprised everyone: the remaining employees became more engaged and more productive. Hastings realized a small team of high performers will almost always outperform a larger team of average hires.
McKinsey’s research supports the bet: high performers can be 400% more productive than average employees. In complex roles like software development, that number climbs to 800%. Netflix now generates almost $3M in revenue per employee—twice Google’s rate, ten times Disney’s.
Michelle’s version of talent density goes beyond hiring. It’s about focus. “I don’t want to give them four priorities, I don’t want to give them five. They know that everybody in my org can repeat to you, what are the three things we’re doing this quarter?”
She also uses AI to compress ramp time rather than reduce headcount. The typical sales rep takes 11.2 months to reach full productivity, according to the Sales Management Association. Michelle’s goal: cut that from five months to three using Crescendo’s own AI assistant, Harmony, which her reps use to instantly pull competitor intel, find lookalike customers, and access the full knowledge base.
“Nine out of ten companies don’t hire enough capacity,” she said. But capacity without density is just headcount. And headcount without focus is chaos.
The Builder’s Throughline
Michelle’s career looks like a series of wild bets: architecture to furniture sales, Salesforce to pre-revenue hardware, a $20B exit to another hyper-growth startup. But there’s a throughline she named early in our conversation: “The thing that’s been consistent throughout my entire career is that I love to build. And I love sales.”
Builders and caretakers approach the same job differently. Caretakers optimize existing systems. Builders tear them down and reconstruct them when the market demands it. Michelle built an enterprise sales org at Groq, watched it become irrelevant overnight, and walked away. Then she built again at Crescendo—with the scar tissue to know that what she’s building today might not be what wins tomorrow.
The fundamentals she keeps coming back to aren’t complicated: Tap people who live in the future. Demand product-market fit before you commit. Simplify relentlessly in noisy markets. Prove value instead of promising it. Hire gritty people and give them focus, not chaos.
None of this is new. But in a world moving this fast, the basics aren’t basic. They’re the difference between scaffolding and structure.
Michelle ended our conversation with advice for new sales leaders: “Be really clear on what your goals are. Stay focused. Make sure your customers are really happy. Listen to them. Make sure you’ve got the right talent. And don’t be afraid to make changes quickly.”
Simple. Hard. And apparently, timeless.
If this episode helped you think differently about building in fast-moving markets, I’d really appreciate a rating or review on your podcast app! It helps more revenue leaders find the show.
Incredible breakdown of what actually matters in go-to-market. The insight about sourcing signal from people "living in the future" rather than consensus is gold. I've watched similar patterns where operators who tap domain experts early spot inflections everyone else misses. The Groq pivot story is brutal but honest, watching PMF shift overnight after buildng an entire org is the nightmare nobody talks aboutenough.