Why Anthropic Are Causing a Comp Crisis & Why You’d Never Hire From Salesforce or ServiceNow

Summary

• The sales-comp bubble is real and Anthropic is inflating it. Chris Degnan says CROs are now landing “$100 million-dollar packages” — when Harry floated $15-20M he was told “you’re not even in the ballpark” — and reps are being offered $1.2M in stock against a startup’s $600K on identical $400K cash OTEs. Peets can’t say Anthropic’s offer is wrong, while Degnan says, “I think Anthropic’s a four or five trillion dollar company,” so a rep looking at $6M in four years is hard to talk out of it. But the punchline is that “what sales people are currently being paid today is not sustainable” — when burn matters again, “all of these comp models have to change.”
• Never hire from a monopoly. The screen for order-takers vs. real sellers: demand two to three new logos opened in the last 24 months, then ask who the champion and economic buyer were — “if they start faltering, it means they’re lying.” Salesforce and ServiceNow reps are poor pipeline-generation bets: “You could talk to a guy at Salesforce. He’s like, ‘Yeah, I closed Wells Fargo.’ No, you didn’t. Wells Fargo’s been a customer for 10 years.” Hire instead from tier-three brands who won with “an inferior product” — that’s grit — and from orgs that develop people (MongoDB, Wiz), not famous logos.
• ARR opacity is the era’s head fake. Founders are lumping monthly on-demand revenue into “ARR” (“my peak day was 10 grand… I’m now at 3.6 million ARR”); with no booked contract “there’s no moat, it’s easy to move” — a competitor ships a better feature and customers quickly switch. Chris Degnan never paid Snowflake sellers until an annual contract was booked, and Degnan dismisses Jason Lemkin’s “multi-year contracts are deferred churn” line: “Is he a VC? That’s probably a typical VC saying.”
• Quota-setting is risk selection, not ambition. Set quotas too low and you overpay for a year while blowing through forecast — survivable. Set them too high and “the sales organization quits… you’re going to replace an A player sales org with a B player sales org.” If ramped reps are doing $3-4M, don’t celebrate — “they’re fat and happy” and you’re under-hiring. New standard kit: windfall clauses (Degnan invoked his roughly five times at Snowflake) so a $20M outlier deal doesn’t pay a rep $4M in commission.
• The Snowflake confessions are the episode’s spine. “We should have kept hiring. We slowed down… because we optimized to be public, and that was a mistake” — this at the first company they’d ever seen where hiring faster kept raising per-rep productivity. And Degnan on himself: “I made too much money… I got lazy” — managers skipping one-on-ones and forecast calls, a North America lead who hadn’t flown in five weeks. Is success correlated with flights? “Yes. 100%.”
• AI changes tooling, not the game. MEDDIC stands (“no pain, no deal; no champion, no deal”), AI prospecting is self-defeating spam (“I get three AI-generated recruiting emails a day… it’s embarrassing” — Peets’s recruiters are banned from sending notes and must call), FDEs are “a glorified professional services person” covering for product gaps, and killing SDRs kills “the future of your sales organization.” What genuinely changes: per-seat pricing is dead, comp must tie partly to consumption, and you now go global day one — Peets’s biggest change of mind in the last 12 months.
• The exit market is the unpriced risk. “Public markets are basically dead” for sub-scale names — “$200M or $300M — even a billion is not even interesting” — Thoma Bravo-style take-privates are “less likely,” so tender offers are substituting for IPOs, with CROs now negotiating rights to sell 20% of shares annually. On Databricks going out at $150B against Snowflake’s $55B: “Is somebody going to make the argument to me that Databricks is worth two and a half times Snowflake? Nobody’s making that argument to me.”

Deep dive

1. The news: Peets and Degnan merge into one sales-building machine — on equity, with a replace-your-CRO clause

• The announcement framing the episode: Chad Peets (30 years building sales orgs, most recently inside xAI/SpaceX) and Chris Degnan (“the only CRO in the history of technology to go from zero to four billion” — Snowflake, per Peets) are formally partnering. Degnan’s origin story: post-Snowflake board work revealed “pretty much the number one pain point of any startup CEO is they would need help building a sales team,” and every other recruiter disappointed him.
• The business model is conviction-priced: cash covers only recruiter expenses; the fee is an equity grant tied to four years. “You will see after a month exactly what you’re getting. By the way, if you don’t, no problem. You fire us.”
• Peets’s entry condition is brutal and stated upfront: “I’ve gone through the sales organization, particularly your chief revenue officer. I can’t work with him or her… you have to commit now that we’re going to replace that person. If you’re not prepared to replace that person, we can’t work together.” He refers skeptical CEOs to his old interviews as a self-qualifier — “a lot of people watch that and say yeah, we shouldn’t work together.”
• On scalability, Peets is candid: seven portfolio companies, “I can’t take on any more” — Degnan plus a recruiting-scale hire (Mike Hoss) make it scalable, but “will we ever be a company that’s got 30 companies in our portfolio? Never.”

2. The order-taker screen: never hire from Salesforce or ServiceNow

• The diagnostic every CEO asks Degnan for: “Give me an example of you opening up two to three new logos in the last 24 months,” then go a level deeper — who was your champion, who was the economic buyer? “If they start faltering, it means they’re lying.”
• Peets’s resume heuristic: monopoly companies can’t teach pipeline generation. “If a guy’s been at salesforce.com for the last 5 years, he’s never opened a new logo… You could talk to a guy at Salesforce. He’s like, ‘Yeah, I closed Wells Fargo.’ What? No, you didn’t. Wells Fargo’s been a customer for 10 years.” ServiceNow gets the same snort.
• What he wants instead: “somebody that works for a company that nobody’s ever heard of and actually has an inferior product and was actually able to go to the market… and win deals.” Degnan’s proof case: a head of sales hired from “a very mediocre company” who succeeded there seven years — “that’s grit.”
• Industry expertise is explicitly less important — “you hire the athlete.” It’s the quality of the sales organization you came from — who trained you, what was the DNA — “there’s great companies with very shitty sales organizations, and shitty companies with world-class sales organizations.” Security-space hiring gets special skepticism (historically channel-driven orgs), Wiz being the recent exception.

3. Quotas as risk management: overpaying is survivable, losing A-players isn’t

• On “we should carry $2M quotas because we’re an AI company”: “Who gives a [__] that you’re an AI company? Show me the data.” The evidence-based rule: if reps onboard and clear $1.5M within six months, hire faster and faster — and if they’re doing $3-4M, “don’t pat yourself on the back… they’re just sitting there and they’re fat and happy.”
• Peets’s asymmetry argument, worth quoting whole: quotas too low → “you’re probably going to overpay… but can you live with overpaying a sales organization for one year? …you’re blowing through the number.” Quotas too high → “none of them are making any money, morale is [__]… if you lose A players, you don’t replace A players with A players. As soon as the A players go, the rest of the A players know that they shouldn’t go there.” Historic 3x-OTE coverage is “probably going up” given PLG lead sources.
• The new-era addition: windfall clauses, because a single rep can now land a $15-20M deal. “You might still make two million, but you’re not going to make five.” Degnan: “At Snowflake I had that windfall clause and I probably invoked it five times.” Harry’s gloss — signing the Pentagon on a $30B contract and taking 20% — lands because the scenario is no longer absurd.

4. ARR is being gamed — book contracts, pay only on bookings, comp on consumption

• Peets’s first message to freshly-funded CEOs: “Raising a round doesn’t mean [__]. The fact that you were able to get three guys to write you a big check doesn’t mean you’ve accomplished anything.” Then interrogate the revenue itself: “Is it ARR or is it annual recurring? …Oftentimes they’ll take monthly recurring and they’ll lump it into annual recurring. You can’t do that.”
• Degnan’s Snowflake policy: sellers weren’t paid “until they got a booked contract, an annual contract.” Monthly on-demand revenue has “no moat, there’s no contract… what happens if XYZ company or Anthropic comes out and replaces you tomorrow?” The point of the booked contract: “you have time. They can’t just switch off you the second something goes wrong.”
• Harry’s pushback via Jason Lemkin — multi-year contracts without usage are “essentially just deferred churn” — is swatted: “I don’t agree with that at all. Is he a VC? That’s probably a typical VC saying.” The synthesis: in consumption models the rep “has to be invested in the long-term relationship” — tie most comp to bookings but some to consumption, so reps neither overbook nor walk away.
• Same logic kills per-seat pricing: “Per seat is dead” — every SaaS company is scrambling to add consumption because AI shrinks headcount, and Snowflake priced on consumption because its own COGS were consumption. To CFOs demanding predictable licensing: “Tough luck, Mr. CFO. This is the new world.”

5. The Anthropic comp crisis: group quotas vs. $100M CRO packages

• Peets lives this “quite literally every day”: “Anthropic in particular is offering sums of money the likes of which we’ve never seen. And I can’t say it’s wrong. They have endless amounts of money and they just care about speed.” The concrete matchup: same $400K cash OTE, but $1.2M in stock against a startup’s $600K.
• His counter-pitch attacks the structure, not the money: Anthropic runs a group quota, so “you can be the best guy in the world at doing what you do and you’re going to get paid as the shittiest guy… They don’t actually value having quality sales people. They just value having a lot of sales people.” Salespeople “are capitalists. They believe in meritocracy” — and if that doesn’t matter to you, “I’m not quite certain I want you in my sales organization anyway.” Degnan is blunter: “Do you think they really give a [__] how good their sales organization is? They probably don’t. Why should they?”
• Degnan concedes the pitch is decaying: “10 years ago I would sell on coming to work for a John McMahon company… That pitch is harder to land today because they’re like, yeah, I hear all that, but I can go to Anthropic and make five times as much money.” Degnan won’t even argue the stock: “I think Anthropic’s a four or five trillion dollar company” — at the 380 valuation offering a rep $1.2M, “you could look at that and say, ‘I’m going to have $6 million in four years.’ How do you tell someone in good conscience not to do that?” Degnan: “[__] $10 million will make you do a lot of things.”
• Up-stack it’s worse: “I know CROs getting 100 million-dollar packages… It’s a bubble and honestly Anthropic is inflating that bubble.” It’s now better-paid to be a startup CRO than a public-company one — driven partly by the shift away from sequential geographic expansion, which makes the rare CRO who has run global day-one “paid quite handsomely… enormous.”

6. The Snowflake confessions: optimizing for the IPO, and the anatomy of rot

• The costliest admission: Snowflake was “the first company we had ever seen where we couldn’t hire fast enough to bring the productivity number down. No matter how fast we hired, it kept going up.” And yet: “We should have kept hiring. We slowed down… because we optimized to be public, and that was a mistake.” Degnan owns it — “my hand was on that wheel.”
• Degnan on how a $4B CRO decays: “I made too much money… I wasn’t doing the performance management like I used to… There were people in my org that were not doing one-on-ones. They were not doing forecast calls… That was my fault. 100% my fault.” He brought Peets back in specifically to “call my baby ugly.”
• What inspection actually looks like, per Peets’s audit: weekly one-on-ones with hard questions (AI helps here — “get into Claude… tell me about their calendar, tell me about their forecast” — but “hey bro, how are things? is not a one-on-one”), eight face-to-face meetings per rep per week, MEDDIC rolled out consistently, managers on sales calls. The tell for a failing second-line manager: the travel report. “Dude, you run North America. You haven’t been on a plane in 5 weeks.” Success–flights correlation: “Yes. 100%.”
• Post-IPO wealth plus no individual quota is the terminal state: “I sell 50 million, you sell 10 million, and we make the same amount of money… There’s a word for that. It’s called socialism.” Big names famously run this way — “it’s raining purchase orders right now at these companies, but that day will change.”

7. Forecast bottoms-up, then break the scaling rules knowingly

• Forecasting is data — productivity per rep, hiring pace, attrition, ramp — “we can pretty much tell you with 90% certainty where you’re going to end the year.” What’s changed: “Doubling is not good enough. Going from 50 to 100 five years ago, everybody’d be super excited. Today, you go from 50 to 100 in a year and you might go out of business.”
• What Peets refuses to bless: reverse-engineering forecasts from fundraising targets — “we need to hit 15 million because then we’ll raise 100 million.” “I’ve threatened to walk away from companies… Forecasts are not supposed to be lucky.” His echo of 1999: “I moved to San Francisco in the dot com boom. It sounds a lot like the dot com boom.” And the margin question nobody asks: “How about you’re reselling a bunch of large language models at a negative margin? How does that go?”
• Scaling 100→300 reps in a year (Snowflake did it once): every ratio breaks. The rule is a manager needs five reps with three productive; at that pace you get managers with six sub-90-day reps, territories slashed from 50 accounts to 10, enablement strained. “Can it be done? It can, but the market better be massive and you better have a product that everybody wants to buy.”
• Ramp math founders get wrong: “productive” means the first quarter at full run-rate (a $1M/year rep doing $250K), and ramp = sales cycle + enablement — “if your sales cycle is six months, how do you get the ramp time less than six months? You can’t.” The keystone role: “the hardest job in technology sales is the frontline manager” — invest in enablement early, at 10-15 reps heading to 50.

8. Performance management: when you have doubt, there’s no doubt

• Degnan’s formative Slootman story, kept intact: told about a failing senior hire, Slootman replied, “Chris, I’m old and I’m going to die soon. So you got to do it now… The problem you have, Chris, is you have too much empathy. When you have doubt, there’s no doubt.” Firing well: “Say as little as possible, get in and out quick, but be kind.” PIPs? “I have seen them work, but it’s a rare occasion. It’s a sign for you to leave.”
• The steady-state numbers, which Degnan calls his most controversial advice because of cost: a manager per five reps, a second-line manager per four managers, six months of nothing from a hundred-person investment, and 25% annual attrition locked into the forecast — including firing the bottom 10% every year, executed quarterly (“you should be able to get rid of 2.5% of your sales organization every quarter”). A-players approve: “If you tell me we get rid of the bottom 10%, I’d be like, ‘You [__] better. If you don’t, I don’t want to come work there.’”
• Europe makes all of this harder, and Peets doesn’t soften it: “It’s very difficult to fire people in Europe… you want to fire somebody in Germany or France or Spain, it’s next to impossible.” Peets’s Amsterdam lesson: PIP’d SDRs immediately go on sick leave, “and then you’re negotiating with lawyers.” His German country manager’s workaround: mandatory four-hour Monday “development sessions” two hours’ drive into Munich — the underperformer resigned by Sunday night.
• The wider cultural read: “As a country, we’re soft” (the US — “I won’t speak to Europe cuz you know how I feel about Europe”), but it’s swinging back — “everybody I’m working with is working 70 hours a week. That was not the case a couple years ago.” The orgs that escape entitlement at scale: SpaceX and xAI, because “they hire for it… the belief that we are doing things to change the world.”

9. AI changes the tooling, not the craft: MEDDIC, phones, SDRs, FDEs

• MEDDIC survives AI, full stop. To a European sales head who declared it obsolete: “Well, dude — no pain, no deal. No champion, no deal.… People want to say AI makes a difference. I don’t believe that at all.”
• AI prospecting is drowning itself: “I get like three AI-generated recruiting emails a day… it’s embarrassing.” The edge reverts to the phone — Peets’s recruiters “are not allowed to send notes. They have to call.” Degnan’s condition for the human seller surviving: “as long as you’re selling something critical to the business — if it fails, people lose their jobs — you have to develop a relationship with a human being.”
• Against Harry’s “$60K SDR is dead, give me one $250K full-stack killer” thesis, the pushback is developmental: “That’s the future of your sales organization… I want to develop those people into my field sales organization over the next 2 to 5 years.” Will sales get smaller? “Potentially… right now it’s the opposite” — cf. Benioff: “No more developers. Salespeople welcome.”
• On FDEs, real disagreement worth keeping: Degnan uses them to drive consumption post-booking — “that’s the right approach.” The sharper counterview (Degnan): “The forward deployed engineer is a glorified professional services person. If you’re a really good engineer, you do not want to be a forward deployed engineer… There’s a lot of technical debt that forward deployed engineers are going to leave” — some big companies just deploy hordes of them to build the missing product on site. Customer success, per Degnan, was always “free professional services” and “could be completely automated to some extent” now that usage telemetry does the analysis.

10. Endgame: unsustainable comp, closed exits, and the Databricks question

• The thing nobody’s discussing, per Degnan: “What sales people are currently being paid today is not sustainable. You cannot pay sales people this amount of money and have a company that’s cash flow positive… Right now nobody cares because there’s so much funding out there… That has to change.” Downside scenario: “There is a world where there is five or six super relevant technology companies and not a lot else… in that world there’s not a need for a bunch of sales people.” Degnan says there’s a bubble even as he insists AI is “life-changing” — “build for speed and go fast, but this bubble will burst.”
• The liquidity contraction: “Public markets are basically dead. You can’t go out with $200 million or $300 million… even a billion is not even interesting.” Thoma Bravo repeating Coupa or Anaplan? “Less likely.” Hence tender offers as synthetic IPOs — CROs now “negotiate into their comp plans the ability to sell annually up to 20% of their shares.” Degnan’s caution from the other side: private Snowflake was judged on top line; public Snowflake on free cash flow. His personal advice: take 5-10% off the table whenever you can — “it made me less stressed.”
• The Databricks exchange, verbatim where it counts: “Snowflake today is worth what, 55 billion? Is somebody going to make the argument to me that Databricks is worth two and a half times Snowflake? Nobody’s making that argument to me… Can they go out at 150 billion, 10% above their last valuation? I’d like to see that.” Degnan adds: “When they start selling stuff at negative margin, I don’t know if they can.”
• Closing pattern-matches: 99% of VCs aren’t vocal enough because “they’ve never been operators” — the gold standard is Mike Speiser (“both a stock picker and an operator… Mike sees [] before anybody else sees it”), plus Sequoia’s Shawn Maguire and John Herring at Vi Capital. Best private sales org: Wiz under Dolly Rajic (the AppDynamics/Zscaler team, transplanted). And the misses confess the model: Lacework was “a world-class sales team, just a crap product” — “if they don’t have world-class product, you’re going to [] lose.”