GTM Strategy Manager: Case Study
Six hypotheses from 50 accounts. Each one becomes a play. Each play earns its way into the system by proving it works at small volume before it scales.
Steve Schnorr · May 2026
Part 0: What the data says
Before designing anything, I spent two days looking at what was there. Six patterns emerged. Each one becomes a play to test.
(Directionally: signal-driven deals went 19/19, generic outbound went 2/5. Small samples, but the pattern is consistent with the threshold analysis below.)
Hypothesis: Paid accounts with workspace growth > 5% and an active admin will convert to expansion deals.
First mover. Shortest feedback loop produces growth thresholds and signal benchmarks that sharpen every other play.
Meridian Healthcare: $86K → $83K expanded → $42K in negotiation. One account, nearly $300K.
Hypothesis: Free accounts with high MAU, 3+ workspaces, and fit ≥ 50 are ready to buy. Someone just needs to ask.
Intelligence from Expand: validated growth thresholds sharpen Convert's PQL scoring. But if a Convert signal fires first, act on it.
Hypothesis: Closed-lost "No Decision" deals with re-emerging usage aren't dead. They're warm accounts with a timing problem. The product proved itself after sales walked away.
Intelligence from Expand: validated signal benchmarks prove that usage patterns predict outcomes. Reclaim applies that to post-loss accounts.
Hypothesis: Accounts running a competitor alongside active Scribe usage have already chosen with their behavior. Consolidation saves them money and unlocks analytics the competitor can't offer.
Intelligence from Reclaim: timing proof and win-back case studies build the displacement playbook.
Hypothesis: High-fit, low-engagement enterprises have a gap between potential and adoption. 4 users found Scribe. Nobody connected the dots to enterprise. Multi-threaded outbound closes that gap.
Intelligence from all prior plays: growth thresholds, PQL model, timing proof, competitive wins, and TCO benchmarks feed the full enterprise playbook.
Hypothesis: Paid accounts showing activity decline, power user departures, or operational friction patterns are at risk. Early CSM intervention prevents churn.
Uses the same Snowflake usage signals as Expand, inverted: growth decline instead of growth acceleration.
All six plays draw from a shared intelligence layer. Expand tests first because it has the shortest feedback loop, not because the others depend on it. As each play produces evidence, the intelligence sharpens. Nothing scales until the data makes it obvious.
Prioritized by impact on revenue decisions × frequency × downstream dependency count. These five scored highest on all three:
OPP-041: test record with negative ACV. Walmart: 50 employees at $611B revenue, tagged Velocity SMB. Vanguard: $180K close tagged VSB. Multiple empty LeadSource fields. Artisan Legal: signup date after last active date.
The framework
Each hypothesis becomes a play. Each play enters the system and proves itself at small volume before it earns the right to scale: Expand existing accounts, Convert high-signal free users, Reclaim lost deals with re-emerging usage, Displace competitors, Activate dormant enterprises, and Retain at-risk paid customers.
Don’t scale what you haven’t proven.
Every play follows the same sequence:
capture what's working →
optimize the signal →
scale what's proven.
All six motions draw from a shared intelligence layer. As each play produces evidence, the intelligence sharpens — but no play is blocked waiting for another. Expand tests first because it has the shortest feedback loop, not because the others depend on it.
50 accounts through the model. Routing is automatic. The rules behind it are human decisions, implemented as code.
Retain: existing paid customers with declining usage signals. Same infrastructure, different trigger direction — growth decline instead of growth acceleration. Not shown above because Retain operates on the installed base, not the 50-account prospect dataset.
Each metric matches what you'll see on the decision cards below. Hover for details.
Every threshold in the qualification strips maps to a measurable win-rate spread from the dataset (see Hypothesis 3). Growth > 5% = +33 pts. Workspaces ≥ 4 = +29 pts. Engagement > 45 = +63 pts. Fit ≥ 50 = +19 pts. Each motion's entry criteria combine the signals most relevant to that play. Nothing is a guess.
TrueNorth Software qualifies for both Convert (Free, high usage) and Displace (runs Tango alongside Scribe). The system needs priority logic.
Rule: the system evaluates which play best fits the account's current signals. TrueNorth's strongest signal right now is free-tier usage volume, so Convert runs first. If they convert, Displace is irrelevant — they'll drop Tango on their own. If they don't convert, the competitive intel from the attempt informs the Displace approach. If a Displace signal fires before Convert validates, act on it. No play is blocked waiting for another.
Those are the six plays and the criteria that route accounts into each one. Next: when each motion earns its start, and what a rep actually sees when an account lands in their queue.
The 30 / 60 / 90 as a system
Each motion enters as a hypothesis. Nothing graduates until the data makes it obvious. This is a testing ground, not a launch plan.
Shortest feedback loop in the system: 28-day cycles, warm relationships, zero CAC. The upsell vehicle is Scribe Optimize (usage analytics, centralized admin, SSO).
PQL-sourced deals close at 72%. Free accounts hitting governance ceilings. The pitch is Pro (workspace admin, analytics) or Enterprise (SSO, Optimize).
"No Decision" is the #1 loss reason. These aren't failures, they're timing mismatches. The evaluation work is done. Re-engaging is the cheapest pipeline you can build.
Accounts running Trainual/Tango/WalkMe alongside Scribe. Scribe Optimize (usage analytics, auto-capture) is the differentiator competitors can't match.
Dormant enterprises (Fit 80+, Engagement <45). Multi-threaded exec outreach with full Optimize + Enterprise pitch backed by proof from all prior stages.
Paid accounts showing activity decline, power user departures, or support friction patterns. Same Snowflake signals as Expand, inverted: growth decline instead of growth acceleration.
Six hypotheses in action
Each hypothesis produces a decision surface: the evidence a rep needs to act. The left side is what they see. The right side is where that evidence flows.
88 inactive users = unrealized value. Enterprise agreement with SSO turns organic growth into a managed rollout. Incident response docs are the lead use case.
The decision
Should the AE initiate an expansion conversation?
“Your team added 3 new workspaces this quarter. Are security ops, engineering, and CS coordinating or building in silos?”
AE (account owner) · Call or wait
Validation: signal-to-meeting rate > 20% sustained for 30 days.
Graduation: proves growth signals are predictive, not noisy. Unlocks Convert.
Trigger
WS growth > 5% AND admin active AND WS ≥ 4
Target
Paid, Tier 1–2. Cloudvault (92), Meridian (87)
Enrichment
Clay contact mapping, SFDC renewal, Snowflake usage
Offer
Optimize + seat expansion. SSO for Enterprise
Why first? Shortest feedback loop (28-day cycles), lowest risk (warm relationships), and the strongest directional signal in the dataset.
Cloudvault enters the Clay table when Snowflake flags workspace growth > 5%. Clay runs three enrichment steps:
Output: a row with account context, 3 named contacts with titles + LinkedIn URLs, and a routing recommendation. Total enrichment cost: ~$0.30 per account.
The AE types @meeting-prep Cloudvault in Slack. Dust's agent runs a retrieval chain across:
The agent returns a structured brief:
Agent instructions and data source architecture are designed and API-tested. The brief format above is what the agent produces.
Under 14% penetration on a free plan. This is a timing problem, not a demand problem. Someone needs to ask.
The decision
Should the BDR reach out or let self-serve run?
“You're running 11 workspaces on our free plan. At your usage level, you're hitting limits on centralized admin, SSO, and analytics. Has governance become a challenge?”
BDR · Reach out or nurture
Validation: does PQL score predict free-to-paid conversion?
Graduation: 25+ PQLs processed. Competitive intel from free stacks feeds Displace.
Acme enters the Clay table when Snowflake flags a free account crossing 200 scribes/mo. Clay enriches in three steps:
Output: scored PQL row with contacts, tech stack, and a “ready to convert” flag routed to the BDR queue.
BDR types @meeting-prep Acme Corp in Slack. Dust synthesizes:
Usage exploded after the deal died. A technical team embedded Scribe into their workflow after sales walked away. The product proved itself. Find the engineer who built it. They're the new champion.
The decision
Should the AE re-engage a closed-lost account?
“When we last talked, you had a handful of users. Now you're at 195 scribes a month across 8 workspaces. Something clicked. What changed?”
AE (original owner) · Re-engage or pass
Validation: do reclaimed deals close faster than net-new?
Graduation: 3+ deals reclaimed. Timing > qualification proven.
Frostline enters the Clay table when the closed-lost re-scan detects post-loss usage growth > 10%. Clay runs a different enrichment path than Expand:
Output: the original deal contacts + 2 new engineering contacts + a “usage-after-loss” flag that routes this to Reclaim, not Convert.
AE types @meeting-prep Frostline. Dust pulls from a different data mix than Expand. It includes Gong transcripts from the lost deal:
NovaTech already chose Scribe with their behavior: 567 scribes/month vs. whatever Trainual produces. Consolidation saves licensing cost and unlocks Optimize analytics Trainual can't offer.
The decision
Should the AE pitch tool consolidation?
“Your team creates 567 scribes a month while also paying for Trainual. What's each tool doing? We often see teams consolidate once they realize Scribe handles both capture and training.”
AE · Pitch consolidation or hold
Validation: does the consolidation pitch win?
Graduation: 1+ competitive win closed. Enterprise proof assembled for Activate.
NovaTech enters the Clay table when tech stack enrichment detects a competitor (Trainual) alongside high Scribe usage. Clay runs a competitive-specific enrichment:
Output: competitive intel row with both tools mapped, 3 stakeholders, and a TCO comparison the AE can use in the pitch.
AE types @meeting-prep NovaTech. Dust pulls competitive-specific data:
Perfect ICP, near-zero adoption. 4 users found Scribe. Nobody connected the dots to enterprise. Multi-thread: find the active users (likely claims ops), map their org chart, identify VP of Operations as exec sponsor.
The decision
Should the AE multi-thread into the enterprise?
“Four people at Pinnacle are using Scribe to document claims processes. In insurance companies this size, that usually means someone found a compliance shortcut. I'd love to understand what they're building.”
AE (enterprise) · Multi-thread or skip
Validation: do enterprise wins generate expansion signals?
Graduation: enterprise logos become Expand candidates. The cycle restarts.
Pinnacle enters the Clay table when ICP fit score hits 100 but engagement stays below 40. This is the “perfect fit, no adoption” trigger. Clay runs enterprise-specific enrichment:
Output: 4 existing users mapped to claims ops + 3 enterprise contacts (Director, VP, Procurement) + department-level expansion estimate. When usage spans multiple departments, output includes per-department contact maps for multi-threaded outreach.
AE types @meeting-prep Pinnacle Insurance. Dust pulls enterprise-specific context:
Meridian expanded to $300K but usage dropped 23% in a month. The original admin champion moved roles. Three workspaces went silent. This is the same account from Expand — the signal infrastructure that detected growth now detects decline.
The decision
Should the CSM intervene proactively, or let usage normalize?
“Your team's Scribe usage dropped 23% this month. Did something change with how you're documenting incident responses?”
CSM (account owner) · Intervene or monitor
Validation: do proactive interventions recover usage within 60 days?
Graduation: 5+ at-risk accounts saved. Intervention-to-recovery rate > 30%.
Full meeting prep briefs, outreach sequences, and detection scripts for all six plays available on request.
Measurement
Each hypothesis has a graduation gate. But does the system compound once plays start stacking? Measurement surfaces the question. A person answers it. You can't wait six months for ARR data. The cascade gives you early signal at every stage.
Every play moves accounts through the same stages. Using Expand detection as a live example:
The shape of the funnel tells you where things break. Signals good but reps aren't touching them? Adoption problem. Meetings happen but opps don't create? Wrong accounts or weak meeting prep.
Compounding rate: Does each motion make the next one better? Does Convert improve as Expand matures? Does Reclaim improve as Convert matures? If not, the plays aren't connected yet. Adjust thresholds, retune enrichment, tighten routing until they do. This is the single number that tells you whether the system is compounding or just running in parallel.
System-level kill criteria: If total signal-sourced pipeline < 2x the cost of running the system (Clay credits, rep time, tooling), the system isn't earning its keep. Measure monthly from Day 60.
The learning loop: Every 30-day cycle feeds back into the system. Thresholds that don't predict meetings get retightened. Enrichment steps that don't improve conversion get cut. Entry criteria for each motion adjust based on what the prior motion learned. The system doesn't just run the plays. It rewrites them.
Five numbers on the weekly exec dashboard.
Do plays create pipeline or just touch accounts that would have converted on their own?
PLG cannibalization risk: The biggest attribution danger at a PLG company: touching free users who would have converted self-serve. Holdback testing is the only way to know. Until we have that data, Convert runs conservatively: high-fit, high-signal accounts only.
What this presentation is also showing you
The bar
The first play ships in week 2. By Day 60, the shared intelligence layer tells us if the motions are compounding or just running in parallel. If signal-sourced pipeline doesn't reach 2× system cost, we kill it.
If it’s not feeding a human decision, it’s waste.
Steve Schnorr · steven.w.schnorr@gmail.com · 507-210-4406
Appendix: AI & tools
I used AI heavily. Here's where I disagreed with it.
Four lessons that change how you scope sprint one: FLS before fields (permissions break silently), deploy in dependency order (record types cause atomic rollbacks), validate types at every boundary (CSV string "true" is not boolean true), and document Bulk API sharp edges (encoding, line endings, field order).
Data analysis, Python scripting, Salesforce metadata, ICP scoring logic, this presentation.
Custom fields, permission sets, bulk import of 50 accounts + 100 opportunities.
Expansion candidates table with enrichment formulas and signal detection columns.
CSV cleaning, ICP score calculation, stage mapping, import formatting. ~200 lines.
Meeting prep agent designed and API-tested. Rep types @meeting-prep Cloudvault and gets a full brief: company snapshot, usage signals, talk tracks, objections. Agent instructions produce the same brief format shown in this deck. Data source architecture: accounts + opportunities uploaded via REST API, chunked and embedded for semantic search.
Each one closes a gap in the system.
Pricing and approval data feeds back into signal scoring. Which expansion offers close fastest? Approval velocity becomes a signal itself.
Competitor mentions, budget language, and champion changes become play triggers, not just call notes.
When Convert closes, CS onboarding milestones become the first Expand signals. The handoff is a signal source.