Sales operations case study · Real problem, synthetic reconstruction
From portfolio data to a coverage decision.
Which accounts should the bank prioritize first—and why?
The case asked where a finite sales team should spend attention across a much larger portfolio—without disrupting existing relationships or mistaking product whitespace for demand. I turned the source views into a proposed decision system built around relevance, permission, capacity, and evidence.
01 · The brief behind the charts
I reframed the problem before I ranked an account.
Which accounts deserve attention first—and how should coverage be tested without inventing impact?
The source exercise gave me several ways to describe the portfolio: business group, company stage, deal type, geography, relationship owner, and product balance. That was useful context, but a sales team cannot act on concentration alone.
Prioritization was not just a scoring problem. It was a constrained operating system: relevance, permission, capacity, and evidence. I turned those constraints into four questions before I ranked an account.
- 01Which accounts have fit and credible timing?Priority
- 02Who owns the relationship and may approve activation?Permission
- 03How much new work can the team absorb now?Capacity
- 04What early outcome would justify continuing?Evidence
02 · How I interrogated the data
I treated every chart as a question—not an answer.
A large segment is not automatically the best segment. A large book does not mean its owner has capacity. A missing product does not prove demand. I used each view to decide what needed validation next.
- 01Concentration
Where does account volume sit—and is that strategic, or simply historical?
- 02Timing
Which lifecycle events could make a conversation relevant now?
- 03Coverage
Does regional account density match the ownership and service model?
- 04Relationship risk
Who owns the account, what commitments exist, and how much can that book absorb?
- 05Customer need
Does product whitespace match a real use case—or is it only an empty field?
03 · Public reconstruction
I rebuilt the data to show the method—not the source.
The original case data is not reproduced here. This model uses 2,400 synthetic accounts with mutually exclusive categories within each lens, so every share and count can be audited.
One denominator
Segment
- Technology
- 43% · 1,032
- Life sciences
- 27% · 648
- Professional services
- 19% · 456
- Commercial and other
- 11% · 264
Stage
- Established
- 51% · 1,224
- Expansion
- 28% · 672
- Early operating
- 15% · 360
- Pre-revenue
- 6% · 144
Region
- West
- 38% · 912
- Northeast
- 27% · 648
- South
- 21% · 504
- Midwest
- 14% · 336
Primary relationship
- Deposits
- 41% · 984
- Credit facilities
- 31% · 744
- Investment services
- 17% · 408
- Payments and treasury
- 11% · 264
04 · Synthetic pressure test
I built three conditions to pressure-test the decision.
These illustrative synthetic conditions may overlap. They are not findings about SVB; they make the operating consequences of concentration, product depth, and missing ownership concrete.
- 45%1,080 accounts · Top three owners
I saw a capacity risk—not a performance ranking.
Before assigning more accounts, I would verify each owner’s active book, open opportunities, service obligations, and SLA load.
- 61%1,464 accounts · One product
I saw a hypothesis—not a cross-sell list.
One-product relationships might contain relevant whitespace. I would still require a real need, fit, timing signal, and relationship-owner approval.
- 8%192 accounts · No owner
I saw an operating blocker—not a prospecting pool.
I would stop activation until the record was deduplicated, ownership assigned, and consent or policy requirements resolved.
05 · My prioritization logic
I used the score to organize judgment—not replace it.
Fit and timing lead because they are closest to relevance. Whitespace and engagement help order the work, but this is only a reviewable v0 queue—not a production model or an automated verdict.
- 40%
Fit
Does the account match the segment, use case, and service model?
- 25%
Timing
Is there a credible lifecycle, funding, renewal, or other verified trigger?
- 20%
Relevant whitespace
Would an absent product solve a real account need?
- 15%
Engagement
Is there recent response, intent, or relationship evidence?
07 · Coverage design
I would earn the right to increase intensity.
Cadence comes after evidence, permission, and capacity. The stronger those signals are, the more tailored the motion can become. When timing is weak, the system waits.
08 · How I would test the decision
I defined the evidence before I imagined the result.
Because this case was not deployed, I cannot say the proposed coverage system lifted pipeline or revenue. I can show how I would learn whether it deserved more investment. Qualified-meeting rate is the first decision gate; pipeline and revenue remain lagging and unobserved.
Can the process run safely and consistently?
Illustrative capacity: six owners × 20 net-new active accounts = 120 accounts. Measure assignment, SLA adherence, usable dispositions, delivery and reply diagnostics, plus relationship guardrails.
This phase establishes operational feasibility—not causal impact.
Does activation outperform a fair deferred comparison?
Calculate sample size from the verified baseline and a decision-relevant minimum detectable difference. Where appropriate, randomize within score band, segment, and owner-capacity block into activate-now versus deferred activation.
Predefine business-as-usual contact, spillover, deferred timing, and guardrail thresholds. Analyze every account by randomized assignment. Expand only if the prespecified estimate and uncertainty support a decision-relevant effect while every guardrail stays in range.
- Qualified meetingPrimary · proposed
- Reply and conversationDiagnostic · proposed
- Opportunity createdDiagnostic · proposed
- Qualified pipelineLagging · unobserved
- Revenue or retentionLagging · unobserved
09 · What this case shows
The analysis improved when I stopped treating the charts as the answer.
The original exercise summarized the portfolio. This reconstruction makes the next layer explicit: the decision, constraints, operating handoff, and evidence boundary. That is the work I would want a hiring team to evaluate.
How I think
- Start with the decision the team must make.
- Turn descriptive data into questions that can be validated.
- Separate account rank from permission and readiness to act.
- Make capacity, ownership, and CRM handoff part of the analysis.
- Design the measurement before seeing a result.
What the evidence allows me to say
- I completed the original analysis as a hiring case for an SVB role.
- This page independently reconstructs the problem with synthetic data.
- The operating system and experiment are proposals, not measured results.
- The portfolio demonstrates reasoning—not attributed business impact.
Revenue operations · Analytics · Experimentation
I turn ambiguous data into a decision a team can execute—and a test it can trust.
If you are hiring for growth, revenue operations, experimentation, or analytics, this is the standard I bring: clear assumptions, executable controls, and honest evidence boundaries.