Ron Gabrisko joined Databricks as CRO when it was doing under $1M in ARR. Databricks is now worth $200 billion and does many billions in ARR, and Ron has run sales the whole way.
He joined Sam Blond, co-founder of Monaco and former CRO of Brex, on the SaaStr AI CRO Confidential podcast to talk about how selling has changed since they last recorded in October 2024. Ron’s view is that the fundamentals of great selling are the same: preparation, relationships, account research, grit, EQ and IQ. Almost everything about how Databricks hires, trains, prospects, demos, and prices has changed around them.
Here are the top 10 learnings from Ron.
1. AE Candidates Build a Working Genie Demo for a Real Account in the Interview
Databricks runs a live test for the skills it wants reps to show customers. Candidates:
- Pick an account
- Deliver a CEO-level AI pitch for that company, specific to its industry and how Databricks would change it
- Get access to Genie, build a demo on the spot, and walk the panel through it and the outcomes it would drive
Back channels and referrals from known networks still count. Every candidate is also met in person.
Ron’s reasoning is simple. His AEs build custom demos for accounts like Pepsi and Rivian as part of the job, so the interview tests whether they can.
2. Reps Are Certified on the Pitch and the Demo All Year Long
Databricks starts with boot camps that cover the whole product line. Then it certifies reps on the CEO-level pitch and on industry-based Genie demos, and keeps certifying them throughout the year.
Ron says sales has drifted away from big-picture storytelling, and he wants reps who can tell a CEO what’s possible for their business. He played baseball, and he applies the same rule: practice 10x more than you play. He expects his reps to be the best sellers anywhere, and says that takes testing, enablement, training, and mentoring.
Every Databricks rep has to be certified on Genie. The company runs Databricks on Databricks, and reps use Genie on their phones.
3. 60% of Rep Time Goes to Customers: Prep on Monday and Friday, on the Road Tuesday Through Thursday
Ron’s long-time rule is “sell from your feet, not your seat.” He thinks many sales teams lost that habit when selling went virtual.
His target is at least 60% of rep time in front of customers. The weekly rule of thumb:
- Monday and Friday: prep and admin
- Tuesday through Thursday: on the road with customers, or more
Ron says reps who follow it are the ones who succeed.
4. Agents Now Do the Account Research, Presentation Prep, and Proposals
Databricks has built agents, all running through Genie, for:
- Account research and background
- Presentation preparation
- Exec prep docs
- Proposals
- Picking the right contacts on LinkedIn and customizing outreach to each one
Ron says he has seen this raise rep productivity, and the time saved is what makes the 60% customer-time target realistic.
5. SDR Headcount Has Stayed Flat While Databricks Keeps Adding AEs and SAs
Databricks still has a large SDR team. It hasn’t grown, even as the company adds more AEs and solutions architects.
Ron attributes that to productivity. Outbound, prospect and contact signals, and campaigns are automated, with personalization down to the individual person. He also notes Databricks sells into one of the hottest markets ever in data and AI, so inbound interest helps.
If he were starting a company today, Ron would build a demand gen team and the tools behind it, and keep that team small.
6. Outbound Reply Rates Went From 1-2% to 6-10%
Ron described the old outbound model as loading 5,000 contacts into “the automated spammer,” sending to all of them, and hoping for 1-2% response rates.
With outreach customized to each prospect, Databricks now sees 6-10% response rates.
His point is that this doesn’t make reps less necessary. As long as companies buy software, he says, they’ll need people to sell it, because the sale still depends on relationships, trust, and teaching customers how to get value from the product. Reps now spend more of their time on that.
7. Ron Keeps the SDR Team Because That’s Where His Future AEs Come From
Many Databricks SDRs move up to AE. Ron’s question for companies cutting SDRs to zero: who teaches new sellers how to sell your product?
He expects the SDR team may shrink over time, but he wants to keep some form of academy for younger sellers, with a path into AE or management. Without one, every AE has to be hired from outside, and no one mentors the next generation. He sees that as a gap.
Some of Ron’s best sellers today came up this way. They grew up with AI, embrace new technology, and write their own programs and apps using Genie and LLMs.
8. The AE Owns the Whole Account, and the Buyer Is Now the CFO and CMO Too
Ron expects AEs to reach new departments, people, and stakeholders themselves. His line: “You’re the CEO of your account.” Covering every part of the account is the AE’s job, not just the SDR’s.
The buyers have also changed. Databricks used to sell mostly into IT. Now AI is a CEO-level initiative, business leaders want to build with it, and Databricks sells to CFOs, CMOs, CIOs, and CEOs. Ron says getting into a company’s most strategic initiatives is where his team spends more time now, and the automation is what freed up the time to cover those buyers.
9. Show Up With a Working Build on Industry Data Instead of a PowerPoint
Ron says the forward deployed engineer model has taken off and customers now expect to see what you’ll deliver. In his view, showing up with a PowerPoint no longer works.
The build may not run on the customer’s production data, but it should run on dummy data that looks like their industry and company. Databricks closes by showing a customer what it built for a specific use case, such as supply chain predictions or stockout prevention, and then deploying it. It has done this with large retailers and pharma companies.
This matters more when selling to business buyers, who need to see the outcome. Ron also thinks most vendors’ messaging sounds identical, so buyers decide on what you can solve for them, how fast, and whether they trust you.
Databricks has always been consumption based, so this is the core motion: pick the top use cases for an industry, sell them to business users, and implement them as fast as possible with FDEs.
10. Databricks Charged $1,000 Per User. Ron Made Users Free and Consumption Went Way Up.
Ron calls consumption pricing one of the smartest decisions Databricks made. As customers’ data and queries grow, Databricks grows. AI agents add to that: Ron expects millions of agents querying data alongside people, and says Databricks has more data under management than any company in the world.
He told several AI coding companies early on they would never get huge on per-user pricing. They have since moved to usage pricing on tokens.
His framework:
- Find your base unit. It should scale over time and track customer value. LLMs use tokens. Databricks uses DBUs.
- Price added value as multipliers or add-ons on that unit. Security, accelerators, and premium support are examples.
- Match price to value. Overprice and no one buys. Underprice and you leave money on the table.
The mistake: early on, Databricks added per-user pricing and a platform fee on top of consumption. When Ron asked customers why they weren’t ramping faster, they said every user cost them $1,000. He asked if they’d open it to everyone if users were free. They said sure. Consumption went way up after the change.
Ron’s other pricing advice:
- Don’t require a commit before customers can use the product. Keep the entry point as low as possible with free tiers, trials, PLG, or a credit card. Once customers like it, usage spreads from one or two users to a department, then multiple departments, and then you ask for a commit.
- Don’t push for a big deal on day one. Early-stage companies need logos and new customers.
- You can fix the wrong model later. Ron thinks companies that started with the wrong pricing can adjust. The bigger mistake is not capturing value from customers at all.
- Set pricing early. Switching from per-user to usage pricing later is very hard.
How Ron’s Playbook Fits Together
Each of the 10 connects to the others at Databricks. Reps are hired by building demos, then certified on building demos. Agents take over research, prep, proposals, and outbound, which keeps SDR headcount flat and gives AEs the time to spend 60% of their week with customers, including the CFOs and CMOs who now buy AI. In those meetings, reps show working builds on industry data. And consumption pricing with free users means once a customer starts using what was built, revenue grows with usage instead of with seat count.
