Snowflake: Agentic GTM Deployment Call
Your notes for the deep-dive on AI tooling, 3,700-person deployment, the MEDDPICC to SPICED transition, and perpetual enablement. Structured around their seven asks, with a co-creation canvas for the best-in-class offering conversation.
"Most of the market is still training people on methodology. Snowflake's consumption model doesn't need better training. It needs a shared operating language that runs inside your agents, your CRM, and your management cadence. That is what SPICED is designed for. Today I want to show you what AI-first deployment actually looks like, walk through the mechanics for 3,700 people, and then spend the back half of this call structuring the offering with you, because this is a new motion for us too, and you are going to shape it."
PProof Point Bank (cite sparingly, land them)
Use these on purpose. Every number you say sticks. Flags show what still needs confirmation before you say it live.
| Proof Point | How to Land It | Status |
|---|---|---|
| 2,000+ companies deployed SPICED | "SPICED as a methodology is deployed across 2,000+ companies. What is new is embedding SPICED inside an agentic stack. That is where Snowflake would be the flagship." | CONFIRM LATEST NUMBER |
| Global Payments train-the-trainer | "Global Payments ran a train-the-trainer model across their GTM organization. Their frontline managers became the system of reinforcement. That is the same skeleton we are proposing for you, with the agent layer on top." | CLEARED |
| Jack, WbD's internal agent | "We run an internal agent called Jack that is trained on the full Revenue Architecture IP and SPICED methodology. We scored our own calls first. It is not a slide. It is how we work. I can show it to you live right now if that would help." | DEMO-READY |
| 53% more revenue per opp | "Reps who quantify Impact sell 53% more per opportunity. That is from our study across 50,000 opportunities over six months. Section V of the proposal, Figure 3." | FROM PROPOSAL |
| SPICED is machine readable | "Five dimensions, each mappable to CRM fields. That is why it works inside agents. MEDDPICC was built for a human-curated whiteboard." | FROM PROPOSAL |
| $1M vs $4M | Only if they raise price. Do not lead with price. "Our model is outcome-aligned. We get paid when your reps deliver impact on calls. Incentive alignment is the point, not the discount." | FROM PROPOSAL |
Say "outcome-aligned," not "outcome-based," in live conversation. Contractual outcome-based pricing is in the proposal, but positioning in dialogue should be incentive alignment, not fees-at-risk. Conor was clear on this.
1What does AI-first deployment actually look like?
"Is this engineered, or is it a prompt library in a deck? Show me the plumbing."
Your Answer (60 seconds)
AI-first deployment is three layers working together. Prompt library at the surface. SPICED embedded in your internal agents with scoring logic and CRM field mapping in the middle. An agent certification gate at the bottom before enterprise rollout. Nothing ships to 3,700 people until the agent passes certification.
The Three Layers
SPICED AI Prompt Library
Every Snowflake rep can build a SPICED-powered assistant in Gemini, Claude, Copilot, or ChatGPT on day one. No infrastructure dependency. Personal productivity tool for call prep, deal review, email drafting.
SPICED Embedded in Snowflake Agents
WbD works directly with your AI and IT team (Phase 2 of the deployment plan). Includes prompt architecture, SPICED scoring logic, CRM field mapping, discovery, renewal, and expansion agent flows.
Agent Certification
Phase 6 in the plan. Before your agents go live to the field, WbD certifies they meet quality standards. This protects Snowflake from deploying a subtly wrong methodology at scale.
Anchor with Jack
"We do this internally. Our agent is called Jack. Jack is trained on the full IP library and scores our own customer calls. We found our own gaps in Impact and Decision before we ever ran this with a customer. The Snowflake deployment starts with your AI team building your version of Jack, for Snowflake, with Snowflake's CRM, inside Snowflake's agents."
- "We have built agents for 2,000 companies." We have not. Methodology yes. Agents no.
- "Our agent is better than yours." Their agents are theirs. You are the methodology layer on top.
- "We will build your agents for you." WbD guides, IT builds. Read Phase 2 again.
2Where is this deployed today?
"Am I an R&D pilot? Who else has done this? Who can I reference?"
Your Answer (45 seconds)
Two answers, and it matters that you separate them. SPICED as a methodology is deployed at 2,000-plus companies. Agentic SPICED, embedded into a customer's own internal agents, is a new and emerging motion. Snowflake would be the flagship. That is not a weakness. It is why the pricing is structured the way it is.
The Clean Distinction
| Deployment Type | Where Today | Proof |
|---|---|---|
| SPICED as methodology (training, CRM fields, coaching) |
2,000+ companies across all segments, including global enterprises running train-the-trainer | Global Payments, case studies across RevOS cohorts CLEAR REFS LIVE |
| SPICED AI at the surface (prompt library in any AI tool) |
Available to every WbD client today. Reps use it in Claude, Gemini, ChatGPT, Copilot | Live in the IP library. Dan can demo the prompt library if asked |
| SPICED embedded in customer agents (agentic rollout, certification) |
Emerging. WbD runs this internally with Jack. Customer-side is a Snowflake-pioneer motion | Internal at WbD. Snowflake becomes the flagship case study per Section VII co-marketing terms |
If you collapse all three into "deployed at 2,000 companies," they will catch it in week 6 and you lose the relationship. If you lead with the distinction, you get credit for candor AND you position them as the flagship. That is why the proposal has the $500K outcome-aligned cap plus $500K IP transfer. You are sharing the risk of a pioneer deployment.
- "We have done this exact thing with Company X." Only if Matt has cleared it.
- "2,000 agentic deployments." See above. Not true. Don't do it.
- "You would be our pilot." Say "flagship." Different word, different contract posture.
3Enablement of a 3,700-person global org: walk me through it
"Do you have the mechanical capacity? Or do I end up with 300 SVPs trained and 3,400 untouched?"
Your Answer (90 seconds)
Top-down and systems-first. Leaders first, managers second, agents third, field absorbs through the work. Methodology adoption fails when it starts with the field and hopes management reinforces it. We invert that. Here is the cohort math.
The Cohort Mechanic
| Phase | Audience | Format | Cadence | Weeks |
|---|---|---|---|---|
| 1 | ELT + Senior VPs (~30) | ELT Alignment Workshop | 1 cohort, intensive | 1-3 |
| 2 | AI + IT Team | Agent Training Guidance (pair with Snowflake engineers) | Working sessions | 4-12 |
| 3 | Rev Ops + Finance | GTM Diagnostic across 5 selling motions | Parallel to Phase 1 | 2-6 |
| 4 | Frontline Managers (~150) | FLM Certification, remote cohorts of 20-25 | 6 parallel cohort streams | 6-14 |
| 5 | Enablement team | Facilitator Certification (train the trainer) | Intensive | 10-12 |
| 6 | AI + IT Team | Agent Certification before field go-live | Quality gate | 17-18 |
| 7 | All GTM (optional) | Jacco keynote | One-time | TBD |
The IC Question (they will ask this next)
"How do you train the 3,400 ICs?"
Three answers running in parallel:
- Asynchronous IP library. All six Revenue Architecture models as self-paced modules. LMS-hosted. Reps complete at their own pace after their manager is certified.
- Manager coaching cadence. FLMs coach SPICED in weekly 1:1s, deal reviews, forecasting, and QBRs. This is where methodology becomes behavior. Not the workshop.
- The agent is the enablement. Reps do not need to memorize SPICED if the agent surfaces the missing element in real time. The agent IS the training. Forever.
- "We will train 3,700 people live." We will not. Do not commit to that volume of synchronous time.
- "The LMS does the heavy lifting." LMS is one leg. Managers and agents are the other two.
4How do we manage the transition from MEDDPICC to SPICED?
"I am worried about sunk cost, political cost, and dual-adoption chaos. Tell me candidly what has worked and what has not."
Your Answer (90 seconds)
SPICED and MEDDPICC are not competing frameworks. MEDDPICC is a qualification scorecard, built for the sales cycle. SPICED is a customer-facing operating language across the Bowtie, built for a consumption model. They can coexist. Where they compete in a CRM, you have to pick one. In practice, SPICED wins for a consumption model because the sale is not the finish line.
Where it has worked
- CRO coached SPICED personally in forecast calls and QBRs, in the first 60 days
- SFDC fields migrated from MEDDPICC to SPICED in the first quarter. No dual fields. No optionality.
- Frontline managers certified before the field heard about the methodology
- Enablement team owned the content, not the rollout. Managers owned the rollout.
- Impact definitions carried from discovery through renewal. Same language in CS that the AE used.
Where it has stalled
- Ran a workshop, called it done. No change to coaching cadence. Reverted inside 90 days.
- Kept MEDDPICC fields in SFDC alongside SPICED. Reps picked. System did not.
- CRO delegated it. Managers did not coach it. Methodology did not land at the 1:1 level.
- Marketing and Product did not adopt the shared Impact language. Discovery Impact did not carry to the case study or the product roadmap.
- Agent or AI layer trained on old methodology persisted silently. Created friction with new expectations.
The Reframe for Snowflake
"You do not have to fire MEDDPICC. You have to decide which one is the operating language and which one is the tooling layer. Command of the Message and MEDDPICC were built for the left side of the Bowtie. SPICED spans both sides. In a consumption model, revenue is re-earned on the right side. That is the business case."
- "MEDDPICC is outdated." You don't know who championed it internally.
- "Force Management is failing." Stay factual. Their proposal was $4M. Let the math talk.
- "Rip and replace." Use "transition" or "evolution." Political safety.
5Beyond launch and Agent Copilots: how do you leverage methodology with AI?
"I don't want a launch. I want an operating system. Does this compound, or does it sit flat after week 18?"
Your Answer (90 seconds)
Launch is week 18. Everything after week 18 is where the real value sits. Perpetual methodology is a closed loop: every call scored, scoring data trains the next agent version, coaching dashboards update weekly, IP updates flow into the agents continuously during the contract, cross-customer benchmarks emerge as the multi-tenant data grows. This is the Titleist R&D story. You are not buying a 2026 methodology. You are buying continuous R&D.
The Compounding Loop
| Element | What It Does | Compounding Effect |
|---|---|---|
| Call scoring | Every customer call scored by the agent on SPICED dimensions | Immediate coaching signal per rep per week |
| Training data loop | Scoring outcomes retrain the agent quarterly | Agent gets sharper on Snowflake's actual customer conversations |
| Coaching dashboards | FLMs see SPICED progression across their team weekly | Management layer coaches on pattern, not anecdote |
| IP flow | WbD IP updates (new playbooks, role IP) flow into agents during contract | Agents stay current without re-contracting |
| Cross-customer benchmark | Snowflake sees their SPICED adoption vs. peer cohort, anonymized | Network effect. Benchmark is the moat. |
| Enablement content | Best and worst calls become exemplars in the LMS and roleplay tools | Enablement content auto-produces from the work (see Q7) |
- "It gets smarter over time" (too soft). Be specific: quarterly retraining, weekly dashboards, continuous IP flow.
- "Network effect" without grounding it. The benchmark IS the network effect. Say that.
6AI automation vs human touch. Where are we now, where should we be in 1-2 years?
"Are you ahead of us, or are you a 2024 methodology vendor in a 2026 wrapper?"
Your Answer (90 seconds)
The honest map. AI is already strong at preparation, scoring, inspection, and content generation. Humans still own the moment of commit, executive sponsorship, the difficult retention conversation, and the novel impact thesis. In 18 months, AI moves from post-call coaching to in-ear coaching during the call, and the management layer shifts before the rep layer does. Managers feel the change first. Reps feel it second.
- Discovery call prep and account research
- Post-call SPICED scoring and summary
- Next-best-action recommendations
- Deal-by-deal inspection and forecast anomaly detection
- Email and follow-up drafting
- Onboarding checklist adherence in CS
- The moment of commit with the economic buyer
- Executive sponsorship and board-level relationships
- The difficult "we are losing you" retention conversation
- The novel impact thesis that does not exist in the training data
- Pressure-testing a Critical Event with a skeptical CFO
- Reading the room when the room contradicts the call transcript
The 1-2 Year Trajectory
| Horizon | What Shifts | Implication for Snowflake |
|---|---|---|
| Today | AI coaches after the call | Weekly dashboards, post-call rubric scoring |
| 6-12 months | AI coaches during the call (in-ear, Gong-style) | Real-time SPICED prompts to the rep |
| 12-24 months | Agent runs portions of the call autonomously (discovery, upsell, renewal pre-calls) | Humans move from deal execution to impact architecture. Manager layer changes most. |
7Using AI to build perpetual methodology enablement content
"Does your enablement content author itself from the work, or do I have to keep commissioning new content every year?"
Your Answer (75 seconds)
Enablement content is no longer authored. It is extracted. Every scored customer call becomes a candidate exemplar. The best calls become positive training data. The worst become the coaching cases. This is the production flywheel. Your enablement team stops writing content and starts curating it.
The Production Flywheel
- Customer calls recorded in Gong, Zoom, or directly through Snowflake agents
- Transcripts scored by the SPICED agent on five dimensions
- Exemplar selection: high-scoring and low-scoring calls flagged and anonymized
- Training content generated: SPICED good vs. bad clips, role-play scenarios, manager coaching cases
- LMS auto-updated: enablement team reviews and publishes. No authoring from scratch.
- Roleplay tools (Yoodli, internal agents) use the exemplars as scoring rubrics for reps in practice
- Manager coaching dashboards surface pattern clips to each FLM for their team
- Agent retraining: exemplar data feeds the next quarterly agent version
Your enablement team's job changes. They stop being content producers. They become curators, coaches, and system designers. That is the growth governance model. Same headcount, 10x the leverage.
- "AI will replace your enablement team." Will not happen. Will not land.
- "No more content creation." You still need strategic content. You stop producing workshop decks.
8Structuring the Best-in-Class Offering (Co-Creation Canvas)
After you answer their seven questions, pivot to this. "We have a proposal on the table. I want to use the back half of this call to structure the offering with you. Four pillars. I will draw them, you tell me which ones matter most and which ones I am missing."
The Four Pillars
The question: What counts as adoption? The proposal says "SPICED elements attained per call." Is that the right signal for Snowflake?
Why it matters: Adoption is not rep behavior. Adoption is buyer response. The buyer answering the Impact question with real numbers is the ground truth. Reps parroting SPICED language is noise.
What to ask Snowflake: "If we measured only one thing in year one, what would convince you this is working?"
The question: Where is accountability in your behavior change model today? For most GTM orgs it is a gap: classroom, roleplay, content, but no closed-loop adoption measurement.
Why it matters: This is the 4th leg most deployments are missing. The agent layer IS the accountability mechanism. Every call, every rep, every deal.
What to ask Snowflake: "Who is accountable for methodology adoption in your org today? The CRO? Enablement? It is usually nobody by name, and that is why it stalls."
The question: How do we know what good looks like for Snowflake specifically, not for the general WbD benchmark?
Why it matters: A consumption model has different dynamics than a seat model. What looks like good Impact in a subscription deal may look thin in a consumption deal. Calibration is the first 90 days.
What to ask Snowflake: "In the first 90 days, would it help if we ran the SPICED scoring on your existing call library before we train anyone? That sets the baseline."
The question: How does this investment get more valuable in year two, year three, without Snowflake paying more?
Why it matters: The multi-tenant benchmark. The IP flow. The agent quarterly retraining. This is the Titleist R&D moat. The unit economics of the $1M cap only work if year two sees compounding returns.
What to ask Snowflake: "When you think about the 24-month horizon, what would make you confident this compounds rather than flatlines?"
9Discovery Questions to Turn Monologue into Conversation
Use these throughout the call, not just at the end. They shift posture from vendor to advisor.
- "When you look at your current Command of the Message deployment, what is working and where is it stalling?"
- "How is MEDDPICC surviving the shift to a consumption model? Where is the language breaking down?"
- "If your agents could do one thing perfectly in every customer conversation, what would it be?"
- "What does a great SPICED deployment look like to the Snowflake ELT two years from now?"
- "What is the deadline pressure on deciding this? Is it tied to your FY planning cycle or a specific launch?"
- "What has to be true by Q3 for this to be considered a success?"
- "Who else on your side should be in the next working session? AI and IT? RevOps? CRO?"
- "Walk me through how a decision like this normally gets approved at Snowflake."
OObjection Anticipation
| Likely Pushback | 1-Line Response |
|---|---|
| "Outcome-based pricing feels risky for procurement." | "Position it as outcome-aligned. $500K cap means your exposure is fixed. We get paid more when your reps are delivering impact on calls. That is incentive alignment, not fees-at-risk." |
| "How do we know your agent scoring is accurate?" | "Agent Certification is Phase 6. Before go-live, WbD certifies the scoring. You and we agree on a calibration set. Nothing ships at scale until the scorer is reliable." |
| "Our AI team will not want WbD in their agent architecture." | "We don't build your agents. Your team does. We provide the prompt architecture, SPICED logic, and field mapping. Your team owns the stack. Phase 2 is explicitly paired work." |
| "MEDDPICC is tied into our SFDC fields. Migration is a nightmare." | "You do not migrate reps. You migrate the coaching cadence and the reporting fields. FLMs change how they run 1:1s. SFDC follows, not leads. 90-day plan, not a cutover." |
| "Why should we pick WbD over hiring this in-house?" | "The IP library, the cross-customer benchmark, and the continuous R&D. Hiring builds your version. Partnering with WbD gets Snowflake the benefit of 2,000 companies of pattern recognition. The moat is the benchmark." |
| "Can we scope it smaller? Start with one region?" | "Yes for the field. No for the leadership layer. The ELT alignment and Agent Training Guidance are global by design. The FLM cohorts and GTM Diagnostic can phase by region if that helps the rollout." |
CClose + CTA
"Three next steps. One: we set up a working session with your AI and IT team to walk through the agent architecture. Two: we run a calibration scoring exercise on a sample of your existing calls, so we have a baseline before we commit. Three: we come back in two weeks with a refined offering based on what you just helped me structure. Which of those can we schedule today?"
- Schedule the AI/IT working session before you leave the call
- Get the calibration sample committed (even 10 calls)
- Name the date for the refined offering. Specificity closes.
- Ask: "What did we not cover that matters most?"
- Ask: "What is the one thing that would make this a no?"
- Ask: "Who else needs to be in the next conversation?"
Winning by Design · Internal Prep Brief · Not for Snowflake Distribution
Generated from WbD_Snowflake_Proposal.md · April 2026