A Deep Dive with Denise Persson (CMO) and Chris Degnan (Founding CRO): How AI Transformed Marketing and Sales at Scale at Snowflake
When Chris Degnan joined Snowflake as employee #13 and the company’s first sales hire, no one could have predicted the company would achieve the largest software IPO in Wall Street history. Now, after recently retiring as CRO to advise AI startups, Chris joined current Snowflake CMO Denise Persson (employee #120) to share how AI has fundamentally transformed their go-to-market motion.
Together, they’ve helped define the data cloud category and scale Snowflake to over 10,000 customers. Their new book, “Make It Snow,” captures the lessons, near-misses, and strategies that built one of enterprise software’s greatest success stories. But this session focused squarely on something more immediate: how AI is reshaping sales and marketing in real-time, and what it takes to succeed in this new era.
Top 5 Takeaways
1. It’s Not Nearly Enough to Tell Everyone to “Go Experiment With AI”
The companies winning with AI share one critical trait: a culture of curiosity combined with executive mandate. It’s not enough to tell everyone to “go experiment with AI” – that creates chaos and duplication. Snowflake’s approach: identify the naturally curious folks who lean in, formalize them into an AI council (30 people at ~20% time allocation across all marketing functions), and have them test, learn, and share quarterly with the broader organization. But if the CEO doesn’t make AI a top strategic priority, employees won’t treat it as one either. You need both top-down leadership and bottom-up innovation.
2. AI is a Task Automator – And That’s Powerful
Stop thinking of AI as magic, at least for many internal use cases. Think of it as an incredibly effective task automator. Snowflake’s global support team saves 418 hours per week using AI tools. Their marketing team of 450 sees 90% time savings on specific tasks like drafting video scripts, interview preparation, and localization. The DataCloud Now TV team reports 90% time savings in script creation. But the critical question isn’t “is it cool?” – it’s “is it generating more money or saving more money?” Without clear ROI, AI tools won’t survive in the enterprise long-term.
3. Proprietary Agentic Models Deliver Game-Changing Results
Snowflake built two proprietary agentic models that transformed their operations. The first is a campaign agent that provides real-time ROI data on every campaign, optimizing digital ad spend and channel allocation automatically – saving significant money while increasing returns. The second is a competitive intelligence agent that gives sales teams instant, customized talking points for any compete situation, specific to the use case, industry, and customer. In B2B, enabling teams to compete effectively across every competitor, use case, and industry combination is nearly impossible manually – this agent solves that at scale.
4. Data Security Isn’t Optional – It’s The Foundation
Large enterprises are terrified of sending PII and sensitive data to random AI tools. Snowflake’s differentiation started with their obsessive focus on trusted, governed, centralized data storage from day one. Their approach: customers can use any LLM they want, but only approved ones can access their data. Non-compliant LLMs get locked down automatically. This security-first mindset applies internally too – their AI tools like Raven (their go-to-market assistant) are built on Snowflake intelligence, keeping everything secure and governed. As Denise emphasized from Ad Week: the trust between vendors and consumers around data usage is paramount. You can’t just let departments “go loose” with hundreds of ungoverned applications.
5. Hire for Aptitude and Curiosity, Not Yesterday’s Skills
In the AI era, adaptability is a superpower. The skills relevant five years ago aren’t relevant today. Denise looks for people who are eager to learn, curious about trying new things, and lifelong learners. She asks candidates: “How do you learn? How do you advance your craft? How do you innovate yourself?” Meanwhile, Chris notes that recruiting for AI-relevant companies is surprisingly easy – younger talent especially wants to work at companies pushing the AI frontier. Position your company correctly in the AI landscape, demonstrate real ROI, and talent will come.
The Foundation: Customer Zero and Data Strategy
Before diving into specific use cases, both Denise and Chris emphasized a fundamental principle: Snowflake is always “customer zero” for their own products. When AI hit “like a fast-moving train out of nowhere,” Snowflake had thousands of customers asking how to handle it. The answer started with what Snowflake had focused on since the earliest days: storing and protecting data in a centralized, trusted, governed location.
“There’s no AI strategy without a data strategy,” Denise stated emphatically. “You need all your data unified in one place and governed to build AI experiences on top of your proprietary data.”
This foundation proved critical. AI is only as good as the data it receives. Snowflake’s platform evolved from a data warehouse to a comprehensive data platform handling structured, unstructured, and semi-structured data – allowing customers to apply AI across all their data sets securely.
Chris added the enterprise perspective: “Large enterprises are super nervous about sending their data out to random AI tools because they don’t want personal identifiable information or PCI data making it out to these AI tools that could get published to who knows who.”
Marketing AI Use Cases: 90% Adoption and Real ROI
Ninety percent of Snowflake’s 450-person marketing organization uses AI daily. Let that sink in. But this didn’t happen by accident or through top-down mandates alone.
The AI Council Structure
Denise launched the AI council, led by Hillary Corpio, focused on new technology innovation. Rather than asking all 450 marketers to experiment (creating chaos), they identified about 30 naturally curious people who raised their hands to spend 20% of their time diving deep into AI possibilities for their specific functions.
These council members attend conferences, test new tools, collaborate across functions, and every quarter host an “AI Day” for the entire marketing organization. They roll out new tools and use cases, share tips and tricks for using Gemini or ChatGPT, and educate the broader team. This alleviates stress for everyone else – they can focus on their work knowing a dedicated team is exploring what’s next.
Budget allocation stays within functional teams (creative tools funded by the creative budget), but the council drives strategy and shares learnings. Critically, any new technology must pass security reviews and meet governance regulations before broad deployment.
The Two Game-Changing Agentic Models
Snowflake built two proprietary agentic models using Snowflake Cortex and various large language models (Anthropic, OpenAI, and others):
Campaign Agent: This agent revolutionized campaign management. While it doesn’t automate every step yet, it provides real-time ROI data on every single campaign running. Most importantly, it helps optimize channels – especially digital ad spend – in real time. Previously, you might realize mid-quarter that you don’t have enough pipeline in California, but it’s too late for marketing to do anything about it. Now, with six-month pipeline forecasting powered by AI, Snowflake can reallocate resources and investments proactively to ensure every territory is healthy.
Competitive Intelligence Agent: In B2B, enabling both marketing and sales to compete effectively against every competitor, at every use case, in every industry is extraordinarily difficult. This agent solves that problem. Sales reps can input: “I’m competing against X company for Y use case at Z type of company” and receive comprehensive talking points instantly. Game-changing.
Additional High-Impact Use Cases
Pipeline Forecasting: Snowflake has used AI for pipeline forecasting for years, but the accuracy now in predicting where pipeline will be six months out has been transformational. This isn’t about current quarter scrambling – it’s about intelligent resource allocation quarters ahead.
Lead Scoring: With millions of leads generated annually, AI-powered granular scoring optimizes the entire customer journey dramatically.
Content Creation and Localization: Marketing owns localization for the entire company (documents, product content, everything). AI has delivered massive cost efficiency and execution speed improvements. For content creation, copywriting, interview scripts, and video scripts, teams report up to 90% time savings.
Digital Ad Optimization: Real-time channel performance monitoring and optimization. The ability to shift spend between channels dynamically based on live performance data has significantly improved ROI.
The Intelligence Team Structure
A critical organizational evolution: Snowflake consolidated previously siloed data and BI teams from across sales, marketing, and other functions into one shared intelligence team. Led by Chief Data Officer Anita Tasi, this team includes data scientists, analysts, and product folks (not product managers for Snowflake products, but product people developing internal data and AI applications).
These team members are embedded in marketing and sales, understanding the business problems deeply because many previously worked in those organizations. They report centrally to eliminate duplication while maintaining close functional partnerships. This consolidation accelerated development of tools like Raven and the campaign/competitive intelligence agents.
Sales AI Use Cases: From Certification to Custom Demos
The Solutions Engineering Revolution
CEO Sridhar Ramaswami asked Chris a deceptively simple question: “How do you know if your sales engineers are good?”
Chris couldn’t answer directly. So Snowflake did something remarkable: they certified every single solutions engineer (formerly called sales engineers), all the way up to the fourth-line leaders running the organization. Not just individual contributors on sales calls – every senior leader had to get technically certified.
This wasn’t ceremonial. Some people excelled. Others realized they needed to significantly up-skill (and Snowflake helped them do so). The result: a genuinely technical pre-sales organization.
Why does this matter for AI? Because with that technical foundation in place, Snowflake rolled out Cursor AI to the entire solutions engineering team in just six weeks. Now solutions engineers can create custom demos and custom content dramatically faster. More importantly, they can iterate in the field as AI evolves rapidly, customizing and changing on the fly with customers.
Credit where it’s due: Chris praised both Ramaswami and solutions engineering leader Moe for driving this transformation. The lesson: technical credibility in your pre-sales team unlocks AI capabilities that would otherwise be impossible.
Raven: The Go-to-Market Assistant
Perhaps the most powerful internal AI tool at Snowflake is Raven, their go-to-market assistant built on Snowflake intelligence. Every sales leader and customer-facing person uses it.
Here’s the problem Raven solves: Previously, before meeting a customer, you’d need to check multiple systems – dashboards over here, Salesforce over there, support ticket systems somewhere else. Time-consuming and incomplete.
Now, sales reps open Raven and ask: “I’m meeting with XYZ customer. Give me a 360 view.” Raven delivers:
- Current consumption trends
- Customer health/satisfaction scores
- Open support tickets and issues
- Opportunities and potential use cases for Snowflake expansion
- Detractors and concerns
Because Raven queries structured, semi-structured, and unstructured data, it provides genuinely comprehensive intelligence. It’s not just a productivity gain (though it is that) – it’s a qualitative improvement in customer understanding.
The C-suite uses it too. CEO Ramaswami meets with at least 10 customers weekly. Previously, the sales team created five-page briefs for each meeting. Now Ramaswami pulls out his phone 30 minutes before the meeting, asks Raven his questions, and gets everything he needs about that account instantly.
This exemplifies an important point: AI isn’t just for individual contributors. When executives adopt AI tools for their workflows, it sends a powerful cultural signal and demonstrates real utility at every level.
Support Team Efficiency
Snowflake’s global support team saves 418 hours per week using AI tools. That’s more than 10 full-time employees’ worth of time – time that support engineers can now spend on higher-value customer interactions rather than mundane tasks.
Governance: The Non-Negotiable Foundation
Both Chris and Denise hammered this point repeatedly: security and governance aren’t optional at enterprise scale.
“Security is the number one thing for Snowflake,” Denise emphasized. “Keeping customer data secure is the top priority. Like many larger enterprises, we cannot just go out and implement any application. They have to go through security reviews.”
This means experimentation happens, but production deployment at scale takes longer. The tradeoff is worth it. As Denise noted from her recent Ad Week visit, the number one topic in marketing AI discussions is privacy and trust. “The trust you have with your consumer regarding how you’re using their data when they’re being so generous giving it to you – there’s nothing more important from an AI perspective.”
The governance approach shapes product strategy too. Snowflake sees more AI applications being built directly on their platform, distributed through their marketplace. Why? Because large enterprises don’t want marketing and sales departments “going loose and bringing in hundreds of applications” – it’s a governance nightmare. Applications built on the data platform, where data never leaves the governed environment, solve this problem elegantly.
Hiring in the AI Era: Aptitude Over Aptitude
“If we’ve learned something over the last couple of years, adaptability is a superpower in business today,” Denise explained. “You need to adapt fast. You need to embrace change. You need to be a lifelong learner and curious.”
Her hiring philosophy has shifted dramatically: look at skillset less, hire for aptitude more. “You can learn the skills today. If you’re a curious lifelong learner, you can learn everything. Things you knew five years ago aren’t relevant today.”
Her interview questions reflect this: “How do you learn? How do you go out and advance your craft? How do you innovate yourself?” These questions reveal adaptability and curiosity far better than technical assessments of current capabilities.
Chris added the market perspective: “There’s a ton of people, especially younger people, super interested in working at companies that are AI-relevant.” He advises multiple AI companies now and consistently sees this pattern. One recent head of sales hire at Factory.ai thought early-stage recruiting would be difficult – instead, he had no shortage of people wanting to join an AI-relevant company.
The key: “It depends on how you market your company.” If you position yourself well in the AI landscape and demonstrate real ROI with your AI initiatives, talent acquisition becomes significantly easier. Chris noted how alive San Francisco feels right now as the “mecca for AI” – people want to be part of the revolution.
The Product Opportunity: Customer Zero to Customer Infinity
A subtle but powerful point emerged throughout the session: Snowflake’s internal AI tools are becoming external products.
Denise mentioned that the internal intelligence tool they use – the one she said she “never wants to see a dashboard again” because she can just interrogate data directly – is launching as a Snowflake product in November. Any customer using Snowflake will be able to deploy similar agents for their own use cases.
This customer zero approach creates multiple benefits:
- Internal teams genuinely use and refine tools (no vaporware)
- Real-world enterprise security and governance requirements get baked in from day one
- The product roadmap aligns with actual business value (ROI requirements)
- Marketing and sales teams can authentically evangelize products they depend on daily
It’s a flywheel: internal innovation drives product development, which attracts customers, which funds more innovation.
The Org Structure Evolution
The consolidation of data and intelligence teams deserves emphasis because it’s counterintuitive. Many companies jealously guard their functional data resources. Snowflake went the opposite direction when Ramaswami brought in Anita Tasi as Chief Data Officer.
Every data analyst and business intelligence person across functions moved into this centralized team. Chris noted: “Anita’s job was still to support Denise and me from a business standpoint, but we consolidated those sources so there was not any siloed applications.”
RevOps still exists at Snowflake – they haven’t been replaced by the intelligence team. Instead, RevOps acts as business stakeholders who partner with the intelligence team. Think of it as: RevOps defines the “what we need,” intelligence team delivers the “how we build it.”
This structure ensures consistency. Anything deployed in sales gets deployed in marketing. Both organizations look at the same data the CFO sees. Eliminating data silos doesn’t just reduce duplication – it creates a single source of truth that enables better decision-making across the entire company.
The Real Talk: ROI and Task Automation
Chris framed AI in refreshingly practical terms: “I view it as a task automator. If there are mundane tasks or things that humans have to do, AI is doing a great job of automating that.”
“There has to be a return on investment. We have to see cost savings. Otherwise, it won’t live in the real enterprise. Customers won’t buy something just because it’s AI. Initially they would, but over time, that’s changing.”
He noted feeling this shift across enterprises: “There are some tools that are cool and I’m using them, but what’s the actual business reason? Is it generating more money or saving me more money? Those are the two things we’re seeing in the enterprise.”
This matters enormously for AI companies and for enterprises deploying AI. The “cool factor” had a brief shelf life. Now it’s about demonstrable business value. Time savings that translate to cost savings or revenue generation. Efficiency gains that scale. ROI that shows up in the P&L.
Snowflake measures this religiously: 418 hours per week saved in support, 90% time savings on specific marketing tasks, real-time pipeline forecasting enabling better resource allocation, ad spend optimization delivering measurable ROI improvements. These aren’t vanity metrics – they’re business outcomes.
What It Really Takes: Top-Level Endorsement + Grassroots Innovation
Denise serves on boards and advises companies. Her observation: “There are some companies where the leadership demand doesn’t come from the leadership level and nothing is happening in those organizations. The CEO really needs to say this is one of our top priorities for the company.”
You can’t delegate AI strategy to a middle manager and expect transformation. It requires:
Top-down:
- CEO declaring AI a top strategic priority
- Executive team actively using AI tools (Ramaswami using Raven sets the tone)
- Investment in proper data infrastructure and governance
- Willingness to consolidate teams and eliminate silos
- Budget allocation for experimentation and deployment
Bottom-up:
- Identifying naturally curious team members
- Formalizing exploration through councils or similar structures
- Time allocation (20% for council members at Snowflake)
- Quarterly knowledge sharing and education
- Empowerment to test within governance guardrails
- Celebration of successful innovations
The magic happens where these meet. Executive mandate without grassroots innovation becomes theoretical strategy that doesn’t ship. Grassroots innovation without executive mandate becomes isolated experiments that don’t scale. You need both, working together, with clear communication channels between them.
Top 5 Mistakes They Made (Or Saw Others Make)
While Chris and Denise focused primarily on successes, reading between the lines reveals critical mistakes to avoid:
1. The “Everyone Experiment with AI” Chaos Trap
Denise explicitly called out what doesn’t work: telling everyone on your team to go out and test new things with AI. “It really creates a lot of unnecessary duplication of efforts and also chaos.” The mistake: assuming democratized experimentation automatically leads to innovation. The reality: you need structured exploration through dedicated groups (like their AI council) who can coordinate, share learnings, and prevent 450 people from testing the same tools in isolation.
2. Measuring AI by “Cool Factor” Instead of ROI
Chris noted customers initially bought AI tools “just because it’s AI” but that’s changing. The mistake: deploying AI without clear business metrics. Snowflake learned early to ask: “Is it generating more money or saving more money?” If you can’t measure 418 hours saved per week or 90% time reduction on specific tasks or real-time ad spend optimization ROI, you’re implementing technology without business justification. The bill comes due eventually.
3. Siloed Data Teams and Fragmented Tools
Before consolidating under Chief Data Officer Anita Tasi, Snowflake had separate data and BI teams in sales, marketing, and other functions. Chris said this created “siloed applications and stuff like that which was really” – he caught himself, but the implication was clear: it was problematic. The mistake: allowing each function to build their own data infrastructure and AI tools. This creates incompatible systems, duplicated efforts, inconsistent data, and makes cross-functional insights nearly impossible. They had to consolidate to scale effectively.
4. Skipping the Governance Foundation
While not explicitly stated as “their” mistake, Denise’s repeated emphasis on security and governance suggests this is where they see other companies failing. The mistake: moving fast with AI implementation without proper security reviews, data protection, and governance frameworks. “Like many larger enterprises, we cannot just go out and implement any application. They have to go through security reviews. It takes longer.” Companies that skip this step either face security incidents or have to rip out and rebuild systems later – far more expensive than doing it right initially.
5. Assuming Technical Roles Don’t Need Technical Depth
The fact that CEO Ramaswami had to ask Chris “How do you know if your sales engineers are good?” and Chris couldn’t directly answer reveals they hadn’t systematically evaluated technical capabilities in their pre-sales organization. The mistake: assuming sales engineering titles equate to actual technical proficiency. Only when they instituted mandatory certification for every solutions engineer (including senior leaders) did they create the technical foundation necessary to deploy AI tools like Cursor effectively. Without that technical credibility, AI enablement would have failed. Companies making the same assumption will struggle to leverage AI in customer-facing technical roles.
The Bottom Line
Snowflake’s AI transformation wasn’t about chasing the latest shiny object. It was about systematically identifying high-value problems, building secure and governed solutions, measuring real business outcomes, and scaling what works across a 10,000+ customer base.
The lessons are clear: culture matters more than technology, security and governance aren’t obstacles but foundations, ROI must be demonstrable, and the companies winning with AI aren’t necessarily the ones moving fastest – they’re the ones moving thoughtfully with both executive mandate and grassroots innovation working in concert.
As Denise and Chris make clear in “Make It Snow” and in this session: there’s no AI strategy without a data strategy, no transformation without top-level endorsement, and no sustainable competitive advantage without building tools you’d be willing to use yourself.
Learn more about Snowflake’s AI journey and lessons from their path to becoming the largest software IPO in Wall Street history in “Make It Snow” by Chris Degnan and Denise Persson, available at makeitsnowsbook.com. Connect with Denise and Chris on LinkedIn.
link