Jeanne DeWitt Grosser on Go-to-Market Engineering, the Buying Journey as Product, and Building Sales Teams Engineers Trust

Jeanne DeWitt Grosser — COO of Vercel, formerly CPO and sales lead at Stripe — delivers a comprehensive GTM operations guide centred on the rise of the Go-to-Market Engineer, treating the buying journey as a product experience, and building technically credible sales organisations.

Key ideas

The GTM engineer is a structural role, not a trend. A former sales engineer who writes code, the GTM engineer shadows the top performer in a GTM function, encodes their workflow into an AI agent, and iterates against KPIs. At Vercel: one engineer spent six weeks automating the inbound SDR workflow, maintaining lead-to-opportunity conversion at human parity, and redeploying nine of ten SDRs to outbound. The agent costs ~$1,000/year.
Go-to-market is a product. As software commoditises, the experience of being sold to becomes a competitive advantage. Map the buying journey the way you map a product flow — from first awareness to five-year retained customer — and optimise each step for experience rather than transaction. At Stripe, replacing the discovery interrogation with a whiteboarding session gave customers an architecture diagram they’d never drawn before.
Segmentation is a company concept, not a GTM concept. Every PM should know which segment their feature targets. Build the segmentation matrix empirically: partner with data science to find which customer attributes predict revenue, then add axes specific to your product’s economics (growth potential for consumption-based models; business model or workload type for Stripe and Vercel respectively).
The 10-minute test for sales teams. The litmus test for a sales org that engineers trust: put an AE in front of ten engineers and it should take them 10 minutes to realise the AE is not a PM. Deep product knowledge, signal-vs-noise discipline, and a GM mindset make the sales org an R&D extension rather than a revenue-extraction function.
80% of buyers buy to avoid pain. Most customers — especially enterprises — are not buying the art of the possible; they are protecting their revenue target, their job, or their brand. Pitch to the risk being avoided, not the upside being gained.

On the GTM engineer

The role’s prehistory: Stripe’s Project Rosland (2017) tried to build a “company universe” database and mad-libs outbound emails without LLMs. It mostly failed — false-positive rates were too high. The concept is now viable.

Modern GTM engineer workflow:

Shadow the highest performer in the target GTM function
Document every data source, tab, and judgment call they make
Build an agent that replicates the workflow (deterministic + LLM steps)
Keep a human in the loop to review every output and hit send
Track the same KPIs as the human baseline
Once KPIs are maintained, redeploy the humans up the value chain

At Vercel: built on their own AI cloud (workflows, AI gateway, fluid compute). The agent runs fully in production for inbound qualification. Lost-bott and deal-bott (built on Gong transcripts) provide retrospective and real-time deal intelligence.

Ideal profile: former sales engineer (understands GTM + writes code) rather than pure engineer or pure AE.

On the deal-bott and lost-bott

Standard loss analysis relies on AE self-reporting, which is systematically biased. In Vercel’s largest Q2 loss, the AE reported “lost on price”; the deal-bott analysed all Slack, email, and Gong call data and returned: “You never reached an economic buyer and failed to demonstrate ROI.”

Real-time deal-bott: posts deal-health alerts into per-customer Slack channels. High iteration velocity creates a parallel enablement problem — Vercel ships something nearly every other day. The deal-bott identifies where the go-to-market team isn’t handling new-feature objections effectively, and weekly sprint reviews fix those “bugs.”

On the buying journey as product

The thesis: when technical differentiation narrows, buyers differentiate on how they feel during the sale. The whiteboarding discovery session at Stripe replaced a standard interrogation with a session where the prospect drew their own payments architecture — often for the first time. They left with an asset; Stripe left with the full competitive picture.

Grosser’s principle: add value at every touchpoint regardless of outcome. The nine-year Stripe perspective: half the customers lost in any given cycle came back five years later and bought.

At Vercel: outreach includes peer-benchmarked core web vitals data and AEO content even when there’s no immediate conversion path. Authority compounds.

On segmentation

The XY matrix approach:

X-axis: size (SMB / mid-market / enterprise) — always relevant because it predicts buying complexity
Y-axis: growth potential — critical for consumption-based businesses where a 200%-growth customer is worth disproportionately more
Third axis: something product-specific — business model (Stripe: B2B / B2C / marketplace / platform → different products); workload type + Crux traffic rank (Vercel: e-commerce / SaaS / crypto; traffic rank = promote tier)

Segmentation is delivered to every new hire in the first week at Vercel as part of “Know Your Customer.” PMs must know which segment they’re building for.

On sales-product alignment

The 10-minute test implies a specific sales team profile: deep product knowledge, GM mindset, ability to distinguish signal from noise when surfacing customer feedback to product. Grosser’s hiring mix: classic sales profiles (pipeline discipline) paired with consulting/banking profiles (quantitative, TCO analysis, CFO conversations). The two cohorts learn from each other.

Sales comp limitation: annual plans lock in incentives 12 months ahead. In a fast-moving product company (AI cloud launched mid-year at Vercel after plans were already written), this creates structural misalignment. No clean solution yet, but worth examining.