Why Utah Produces Elite Sales Talent + Building The GTM Machine @ Exa w/ Marcus Holm, CRO @ Exa

Marcus Holm is the Chief Revenue Officer at Exa, the search API built for agents, now used under the hood by Cursor and Cognition's Devin among others. He joined in 2026 after nearly three years at LaunchDarkly, where he was CRO and then President and took rep quota attainment from 18% to 54%. Before that he was a first-time CRO at Forter, joining sub-$100M and tripling revenue while the valuation 5x'd, and led go-to-market at Heptio, the Kubernetes company founded by Craig McLuckie and Joe Beda that VMware acquired at a ~40x revenue multiple. Earlier stops: Cloudera during the Hadoop wars through IPO readiness, and three and a half years at Cisco across Raleigh, Dallas, Singapore, and the Bay Area.

The origin story: number five of 11 kids in Minneapolis, with a dad who was a professional sales trainer. He spent two years in Lithuania, Latvia, and Estonia selling religion to atheists in Russian, a language he didn't speak a lick of when he arrived. Back in college he started knocking doors on commission, and by his third summer was running 150 reps across five offices. Then he took the Cisco SDR seat at a fourth of what he'd made the year prior to break into enterprise - and never looked back on technical sales.

Marcus's answer to the nature vs. nurture question: it's a training ground, not a gene pool. An 18-24 month mission runs on a near-military schedule - up early, study a foreign language, then out contacting strangers in parks and knocking 10-15 story Soviet apartment blocks (elevator's usually broken, so you walk to the top). In his case, that meant selling religion to atheists in the former Soviet Union, in Russian he didn't speak a lick of when he landed.

Come home from that and "doing sales of any form seems fairly natural." You're numb to rejection with thousands of at-bats already banked. His intimidation hierarchy explains why these hires keep outperforming: door-to-door is peak difficulty (in-person, on their doorstep, right at dinnertime), telesales lets you hide behind the phone, and field sales is basically the missionary motion - build rapport fast, discover, qualify, work the cycle.

His company-selection framework is channeling your inner venture capitalist: how big is the category, and what's this company's likelihood of taking an outsized share of it. But in the AI era there's a second filter that's "half of your decision criteria" - figuring out who Claude won't kill. Even the smartest VCs and CTOs give him polarizing answers (vertical app layer vs. horizontal infrastructure), so you have to underwrite the moat yourself.

What got him to Exa: years spent as a lab building a proprietary index from scratch, which means a steep barrier to entry. His adage is "easy come, easy go" - if a competitor sees you go 0 to $100M in two years and can mimic you with a wrapper and a few engineers, they'll drag the price to the bottom. He watched this exact movie in open source: a cemetery of companies that never built enough accretive IP around the free kernel, and the few (Red Hat, Databricks, Confluent) that did.

Participation rate tells you whether you're ready to scale. Feast-or-famine attainment means you haven't cracked the hiring profile, onboarding, or the motion itself, and hiring more reps just multiplies the problem. At LaunchDarkly he inherited 70-80 sellers with 18% over quota; by the time he left it was ~54% over plan, and closed opportunities went from sub-20% outbound-sourced to 70-80%, with larger deal sizes and ramp condensed by 30+ days.

His benchmarks: the SaaS-era model was 70% of reps at 70%+ attainment. Post-SaaSacre, 30-40% making plan is the norm, and north of 50% is best in class - at which point it becomes a recruiting weapon, because the club problem flips from filling seats to rationing them.

His recruiting pattern is the ICE framework: intelligence, character, experience - and the first two outweigh the third, because most of the rest is coachable. The interview goes layers deep on wiring: "where you spent 10,000 hours to get really good at something," how you cope with rejection and failure, your proudest nonprofessional achievement. Pass those and he'll happily hire outside dev tooling - AI is so new that everyone is learning on the job, himself included.

The thing he's changed his mind on: he used to bias toward sellers wired like him (high energy, enthusiastic, inspirational). Now he embraces the surgical, introverted, analytical profile - lower volume of outreach, higher conversion - because "to be trusted versus liked are two different things."

The architecture itself is changing. FDEs are going full-spectrum - one resource spanning pre-sale and post-sale - and the back office is compressing: his longtime RevOps partner Karan Singh built dozens of propensity-ranked territories in about two weeks using Claude, work that historically took a dedicated planning team. The result: revenue per head is way up, and quotas consistently run 2-3x SaaS levels by segment.

The unlock is that Exa is 95%+ PLG, so reps aren't fighting for meetings - they're engaging users already in the pond and converting pay-as-you-go usage into multi-year, dollar-committed contracts.

His phrase for it: "cherry pick and get the cream off the top."

(03:30) Why AI companies keep building sales teams in Utah

(12:38) The adversity that set his risk tolerance

(16:15) How he ended up a technical seller, and the superpower he found learning Russian

(21:34) Learn, earn, grow: how to pick your next company

(28:27) Who Claude won't kill, and why the open source graveyard is the analog

(34:54) Websites vs. the API, and humans becoming the customer

(43:56) The ICE framework for hiring

(48:17) The AI tools his reps actually use

(51:55) From 18% to 54% quota attainment

(1:02:04) GTM mafias, and why the AI era is too early to call