- Started at
- Owners don't engage because they lack education. Build the portal.
- What moved them
- Three weeks of discovery, then a direct answer to each of four concerns: trust, accuracy, risk and ROI. 78% already had enough information; only 31% knew which option fit.
- Outcome
- Approved: build the decision engine instead of the portal, with proof that the algorithm works and that owners trust it.
Business Owner Succession Simplified.
PNC asked for a content portal. Research showed owners didn’t need more to read; they needed to know how ready they were and who to talk to. I turned advisors’ judgment into 47 rules owners could run themselves, and client engagement went from 35% to 86%.
The ask was a succession-planning content portal. Research showed 78% of owners had enough information, but only 31% knew which exit option fit. I reframed it as a personalized assessment engine, translating advisor expertise into 47 business rules to clarify readiness, exit paths, and advisor fit. Engagement rose from 35% to 86%, advisor mismatch fell from 67% to 18%, and the bank gained $1.5M annually.
PNC's Succession Advisory team asked for a password-protected content portal. Owners arrived too late, met the wrong advisor, and only 17% of first consultations led to immediate forward progress.
A rule-based assessment engine: 47 codified rules, two-dimensional readiness scoring across 9 zones, persona routing, personalized action plans and advisor matching. Owners left with a plan, and the PNC specialist they were matched with arrived already briefed.
Led discovery: 13 advisor interviews, 6 owner interviews, 89 survey responses and 12 shadowed consultations. Ran four rule workshops with eight advisors. Team: PM, engineering lead, 8 succession advisors.
86% client engagement
$1.5M additional annual revenue
+67 NPS points
98% advisor agreement
What Business Succession Actually Is
Most business owners spend decades building something — then face one of the most complex financial and legal decisions of their lives with almost no structured guidance. Succession planning is how you exit: sell to a strategic buyer, transfer to family, set up an ESOP, or hand off to management. Each path has different tax implications, different timing requirements, different advisors. Get it wrong and you leave significant value on the table, or worse, hand the business off too late to matter. PNC Bank had a team of specialists who navigated this for clients. The problem was getting the right owner to the right advisor at the right moment — before it became an emergency.
What I Was Handed
PNC Bank's Succession Advisory team handed me a scoped request: design a password-protected content portal for business owners planning succession. The brief was clear, the stakeholders were aligned, and the build timeline was defined. I had one question before accepting the scope: has anyone validated that an information portal is actually what owners need?
A defined solution
Portal with articles, videos, and guides. Stakeholder hypothesis: owners weren't planning succession because they lacked education and awareness.
No validated problem
The brief described a solution — not a validated failure mode. Low engagement had been observed, but the cause had been assumed, not researched. That gap warranted three weeks of discovery before any design work.
The problem might be upstream
Succession planning is a high-stakes, deeply personal decision. Information overload seemed more plausible than information scarcity. I flagged this with the PM before committing to the portal direction.
The Challenge That Wasn't
The bank's Succession Advisory team came in with a clear request: build a password-protected portal where business owners could learn about succession planning. On paper, that sounded reasonable. In practice, it was aimed at the wrong failure point. Owners were not struggling because content was unavailable. They were struggling because the decision itself was complex, emotional, high-stakes, and deeply situational.
A Password-Protected Information Portal
- ✕Articles, videos and guides about succession options
- ✕Timeline explanations and best practices content
- ✕Build awareness to encourage consultations
Their hypothesis: Business owners were unaware of succession services. Leadership believed the problem was lack of education and awareness.
A Rule-Based Algorithmic Decision Engine
- ✓47 algorithmic business rules codifying advisor expertise
- ✓4-dimension business readiness assessment (0–100)
- ✓4 succession personas with smart routing
- ✓Personalized, prioritized action plans with timelines
- ✓Advisor matching based on assessment outcome
The pivot: The problem wasn't lack of information — it was lack of personalized, actionable guidance.
Through systematic UX research, I uncovered the real problem and proposed a rule-based algorithmic decision engine instead. Research showed 78% of business owners already had enough information — but only 31% knew which option applied to their specific situation. That gap mattered. It meant the real barrier was not awareness, but clarity: which path fits, how ready am I, what should I do first, and when should I speak to an advisor? A portal would have added more content to an already overloaded decision. The product needed to reduce ambiguity, not increase reading.
Discovery
Information was never the problem. This was a decision support problem, not an education problem.
Validated the Problem Through Systematic UX Research
Before committing to anything, I needed to know whether a content portal would actually solve the problem — or just look like it did. Three weeks of discovery inside a six-month engagement. Advisor interviews, owner interviews, a survey, process shadowing. The goal wasn't to validate the brief. It was to find out if the brief was right.
Understand why business owners aren't engaging with succession planning
Validate whether "lack of education" is truly the barrier
Identify what would actually motivate owners to take action
Discover where technology could simplify
Explore how advisors qualify and segment business owners
Succession Advisory Team Interviews
13 interviews · 60 minutes each · Semi-structured format
Business Owner Interviews & Survey
6 business owner interviews · Likert scale survey · 89 responses from business banking clients (age 50+, businesses 10+ years old)
Behavioral Analysis & Process Shadowing
CRM data · Engagement patterns · Journey mapping · 12 consultation sessions observed
Research Phase 1
Succession Advisory Team Interviews
Problems identified: Owners arrive too late · Unrealistic expectations about business value · RMs lack assessment framework · High mismatch rate between owners and advisors · No systematic approach to readiness evaluation. Emerging hypothesis: Need assessment tool, not just education · Timing and readiness are key factors · Qualification process could improve outcomes · Advisors have tacit knowledge to codify · Personalization might drive engagement.
Research Phase 2
Business Owner Interviews and Survey
Business owners knew of succession planning but couldn't process the volume or relevance of available information. They needed simplified, sequential decision paths.
Owners didn't need more content; they needed clarity on what to do next and confidence that they were moving toward the right goal.
Disconnected tools, inconsistent advice. Users needed one unified ecosystem that simplified advisor perspectives into one clear, guided recommendation engine.
Rigid tools didn't adapt to urgency or personal context. Users disengaged when tools felt "too corporate" or irrelevant to their immediate situation.
Top Challenges — Survey (89 Responses)
Likert scale survey sent to business banking clients (age 50+, businesses 10+ years old)
Key Finding: Information ≠ Decision Confidence
Information isn't the problem. This is a decision support problem, not an education problem.
Research Phase 3
Process Shadowing & Decision Mapping — Understanding exactly how advisors make qualification decisions through direct observation of 12 consultation sessions.
Advisors Follow Consistent Evaluation Logic
Despite claiming their assessment was "intuitive," all advisors followed remarkably similar patterns. Standard opening questions (all advisors asked): What's prompting you to think about succession now? Tell me about your business. What's your timeline for exit? What's your role in day-to-day operations? What type of succession are you considering? Do you have a successor in mind?
Advisors Evaluate Against Multiple Dimensions
Advisors were evaluating approximately 40–50 distinct factors across 4 dimensions: Financial Health (revenue trend, profitability, cash flow, debt levels, customer concentration, revenue diversification), Operational Readiness (owner involvement, management team, documented processes, key person dependencies, employee retention, systems maturity), Market Position (competitive advantages, market share, customer loyalty, brand strength, industry trends), Transfer Readiness (emotional readiness, successor ID, timeline clarity, estate planning).
The Mismatch Problem Is Quantifiable
Out of 12 consultations observed: 8 consultations = "You're not quite ready yet. Here's what you need to work on." 2 consultations = "You actually need to talk to a different advisor type." 2 consultations = "Great, you're ready. Let's move forward."
Common Mismatch Scenarios Observed
Advisors are executing decision tree algorithms. Their questions follow IF/THEN logic that branches based on user responses. This logic can be codified. Succession readiness is multi-dimensional — it's not one score but evaluation across at least 4 major categories with 10–15 factors each. This complexity is why owners can't self-assess accurately.
Understanding Distinct Succession Personas
From shadowing and interviews, I identified that succession planning conversations followed 4 very distinct patterns based on the owner's situation. These 4 personas require completely different evaluation criteria, advisor expertise, and action plans. A one-size-fits-all portal wouldn't address their unique needs — but a rule-based assessment could route each persona to the appropriate pathway.
- Valuation growth strategy
- Business optimization (improve margins, reduce dependencies)
- Exit structure planning (earnout vs. lump sum)
- Financial audit preparation
- Successor readiness assessment
- Training and mentorship program
- Buy-sell agreement
- Estate and tax planning
- Family conflict management
- Urgent valuation
- Interim leadership solution
- Quick exit options mapped
- Liquidity urgency planning
- ESOP feasibility analysis
- Valuation and financing structure
- Employee governance training
- Long-term business stability planning
The 5 Whys Workshop
I ran a workshop with the product team using the 5 Whys technique to get to the root cause — tracing from symptom to structural problem. This became the strategic justification for the pivot.
1
2
3
4
5
The gap isn't information — it's codified decision-making expertise. Advisors have complex evaluation algorithms in their heads. Owners and RMs need access to that algorithmic thinking before they meet with advisors. The solution: a system that asks the right questions, evaluates against expert criteria, identifies specific readiness gaps, prioritizes what to work on first, determines which succession pathway fits, and routes to the right advisor at the right time.
Affinities Mapped
I organized all research findings into thematic clusters — each revealing a specific gap that needed a specific solution type.
Assessment Gap
- Owners can't objectively evaluate readiness
- 62% lack confidence in self-assessment
- Overestimate preparedness
- Don't know evaluation criteria
Decision Support Gap
- 78% have information but only 31% know what applies to them
- Can't translate general content to specific situation
- Paralyzed by too many options
Routing & Matching Gap
- Don't know which advisor type they need
- 67% of consultations are with wrong advisor or wrong timing
- RMs can't effectively triage clients
Personalization Gap
- Succession needs vary dramatically by persona
- One-size-fits-all content doesn't work
- Need pathway specific to their situation
Strategy & decisions
A portal would have improved the surface. Only a rule engine could change the system.
The Critical Questions We Had to Answer
Before the pivot could be approved, leadership needed answers to four legitimate concerns about the algorithm-first approach. Each required a concrete counter-argument grounded in research.
"This is the right call. Your research makes a compelling case that we were solving for the wrong problem. Let's build something that actually helps owners assess and decide, not just read. But we need to see proof that the algorithm works and that users will trust it." — VP of Business Banking
Three Directions. One Right Answer.
Before committing to the rule-based engine, I mapped three credible product directions against the research findings. The goal was to make the direction choice explicit and defensible — not to arrive at the "right" answer by default.
Enhanced Content Portal
The original brief. Better-organized articles, videos, and guides with improved navigation. Fast to build, low risk, familiar to stakeholders.
Guided Questionnaire
A simple step-by-step form that collects owner information and routes them to relevant content sections. Lower complexity than a rule engine.
Rule-Based Assessment Engine
47 codified business rules. Two-dimensional readiness scoring. Persona routing. Personalized action plans. Advisor matching.
Research showed the real problem was codified decision-making expertise locked in advisors' heads. Only a rule-based engine could make that expertise accessible, scalable, and personalized without requiring an advisor in every interaction. Directions 01 and 02 would have improved the surface — not the system.
Four groups. Four different arguments.
The pivot held because each group was moved by its own kind of evidence: research for leadership, their own logic played back for advisors, test data for engineering, and a working tool for relationship managers.
- Started at
- Their assessment was intuitive, held in their heads, 20 to 30 factors at a time.
- What moved them
- I played back the IF/THEN logic observed in shadowed consultations: "Is this what you're actually doing?" Dot voting and paired comparisons replaced debate.
- Outcome
- 63 factors became 47 rules with agreed weights. "For 95%+ of situations, this will route owners correctly."
- Started at
- Calculate scores only after all 47 questions. Simpler architecture, faster build.
- What moved them
- Testing data: a visible progress bar cut mid-assessment abandonment by 68%. The tradeoff, quantified.
- Outcome
- 1.5 extra weeks agreed for live scoring. 85% completion rate.
- Started at
- No concrete offer for clients, and guessing which advisor to recommend. Net engagement 8%.
- What moved them
- Using the assessment with clients in usability testing, with advisor matching doing the routing.
- Outcome
- 9.1/10 confidence and net engagement from 8% to 25%. "Take this 20-minute assessment."
Key Design Tradeoff
Decision: Real-Time Scoring, Not Post-Completion
The Technical Constraint: Engineering proposed calculating scores only after all 47 questions completed. Simpler architecture, faster build.
My Insistence: Users needed progress feedback during the 20-minute assessment to prevent abandonment. I specified live score calculation with visible progress bar.
How I got alignment: I brought testing data showing 68% reduction in mid-assessment abandonment with a visible progress bar. Engineering agreed once the tradeoff was quantified — 1.5 extra weeks for a measurable retention gain.
85% completion rate with live scoringThe engine
The checklist in advisors' heads, codified into 47 rules owners could run for themselves.
Codifying Advisor Expertise Into Algorithmic Rules
I shifted into solution architecture mode to translate human expertise into intelligent design. Three major challenges: codify advisor judgment into algorithmic rules, develop a model that assesses both business and personal readiness, and architect a dynamic context-aware question flow that reduces cognitive load while delivering personalized guidance.
Step 1 — Codified Advisor Expertise Into Algorithmic Rules
I conducted structured workshops with eight advisors to extract their tacit expertise and formalize it into a rule-based assessment framework.
Decision Criteria Mapping
Objective: Document every factor advisors consider when evaluating succession readiness. For each dimension, I asked advisors to walk through real business scenarios while I documented every criterion mentioned on a whiteboard.
Prioritization & Consolidation
Objective: Narrow to most critical factors and organize into categories. Used dot voting exercise — 20 dots per advisor on most critical factors. This surfaced what advisors actually weight when forming recommendations vs. what they mention in passing.
Weighting & Scoring
Objective: Determine how much each factor matters and how to score them. Presented advisors with 10 real (anonymized) business scenarios. "For each business, walk me through your evaluation. What are you looking at? What would make you say 'yes, they're ready' vs. 'not yet'?"
Threshold & Routing Rules
Objective: Translate verbal IF/THEN advisor logic into formal, testable business rules. Each rule specifies: conditions, resulting flags, recommended advisor match criteria, and estimated timeline impact on the owner's plan.
Sample Codified Rules — Translated From Advisor Expertise
Step 2 — Designing the Assessment Model
Succession readiness is multi-dimensional. It's not one score — it's evaluation across at least 4 major categories with 10–15 factors each. Two dimensions matter because readiness and attractiveness define when and how a business should transition — ensuring owners don't exit too early, undervalue their company, or miss the opportunity to maximize outcomes.
- Revenue trend (3-year)
- Profitability margins vs. benchmarks
- Cash flow consistency
- Debt-to-equity ratio
- Working capital adequacy
- Customer concentration risk
- Revenue diversification
- Compliance status
- Owner involvement in daily ops
- Management team strength
- Documented processes (SOPs)
- Key person dependencies
- Employee retention rates
- Technology systems maturity
- Supplier relationships
- Competitive advantages
- Market share position
- Customer loyalty rates
- Brand recognition
- Industry attractiveness
- Growth potential
- Barriers to entry
- Owner emotional readiness
- Successor identification
- Financial readiness for retirement
- Timeline clarity and urgency
- Succession goals clarity
- Estate planning status
- Legal structure
Why Operational Readiness is weighted highest (35%): It is the most common failure point in succession transitions and the dimension where advisor intervention creates the most value. A business built around a single owner's knowledge and relationships cannot transfer cleanly without operational independence — this is why 67% of initial consultations result in "not ready yet."
Two-Dimensional Readiness Scoring
The algorithm calculates weighted scores across 4 dimensions, generating a Business Readiness Score (0–100) and a Business Attractiveness Score (0–100). These map to a 2×2 matrix creating 9 positioning zones — each triggering personalized messaging, next steps, advisor matching, content recommendations, and timeline estimates.
"Is this business operationally ready for ownership transition?"
"How appealing to potential successors or buyers?"
9 Zones. Personalized Guidance at Every Position.
Each zone triggers unique messaging tone, next steps, advisor matching criteria, content recommendations, and timeline estimates. No two owners receive the same guidance.
Optimal Position — High Ready / High Attract
Minor gaps only. Multiple pathways viable. Move to advisor consultation immediately.
Building Toward Readiness — Mid Ready / High Attract
Strong attractiveness but operational gaps. 9–18 months of focused improvement with clear priorities.
Strong Position — High Ready / Mid Attract
Ready operationally — work on increasing market attractiveness before approaching buyers.
Early Stage — Low Ready / Low Attract
18–24+ months needed. Foundation phase — prioritized gap closure plan generated across all 4 dimensions.
Used round numbers (58, not 58.3) to avoid false precision. Show both dimensions side-by-side with neutral color palette. Use descriptive labels, not just numbers — "You're in the Building Toward Readiness zone" communicates more than "58/72".
Step 3 — Mapping Persona-Specific Decision Pathways
The challenge was to interpret the same business data differently based on context. The same operational, financial, or leadership scores can mean very different things for different types of owners. The solution was a persona identification algorithm — a logic layer that classifies each owner early in the conversation so recommendations, priorities, and advisor matches are tailored to their unique succession path.
Early in the assessment, the system determines persona through strategic triage questions:
After 5–8 triage questions: Persona is identified with confidence score, then the same assessment data gets interpreted differently based on persona match. When confidence falls below threshold — typically when answers are ambiguous or contradictory — the system routes to the full 47-question assessment rather than the adaptive short path.
Same assessment data (Operational Independence 45/100), completely different interpretation and recommendations based on persona. The algorithm adapts its priorities and advice to match the owner's specific succession goals — this is what makes it a decision engine, not an information portal.
Step 4 — Assessment Question Flow Design
The challenge: Don't show all 47 factors at once. Users would abandon a tax-form-like experience. The solution: progressive disclosure with adaptive branching — group into digestible phases, skip irrelevant questions, sequence easy-to-hard.
Progressive Disclosure
5 phases, ~47 questions total, but only 32–38 actually asked per user. Questions are grouped into phases with natural topic progression — Financial → Operational → Market → Goals.
Adaptive Branching
Skip irrelevant questions based on responses. Single owner → skip partnership questions. No employees → skip management questions. Not family succession → skip family dynamics entirely.
Easy to Hard Sequencing
Build momentum with simple factual questions first (What industry? Years in business? Employee count?) then move to evaluative questions that require reflection. Reduces early abandonment.
5 Phases · ~47 Total Questions · 32–38 Asked Per User
Sample Question Flow
Q8 · Customer Concentration
Q10 · Operational Involvement
Target completion time: 18–22 minutes. This was calibrated carefully — long enough to feel substantive and earn trust, short enough that busy business owners would commit. Testing showed 85% completion at 22-minute average, validating the sequencing design.
The product
135+ screens across 4 persona pathways, 9 scoring zones and 3 user types.
Design Solution — Screen-by-Screen
The core design challenge was making a 47-question algorithmic assessment feel valuable rather than burdensome — while presenting complex, multi-dimensional results that non-analytical business owners could understand and act on. Every screen was designed around a specific cognitive goal.
Show the output before asking for input. Time expectation upfront, example results preview, single focused CTA — no decision paralysis before commitment.

Left 60%: centered question card, single focus point, clear action buttons. Right 40%: live score preview, progress indicators, contextual help — showing real-time impact of each answer on the assessment.




Expandable phase sections prevent overwhelm — owners see Months 1–2 first, not the full 18-month roadmap. Success metrics per phase provide clear completion signals. Every action connects back to their specific situation.

The Unprepared Owner Path
Unlike traditional IA focused on content hierarchy, I designed two parallel structures: the algorithmic assessment flow and the supporting content ecosystem. Abstract logic becomes real when you trace a specific person through it. The left diagram maps John — a 52-year-old restaurant owner — through the full system, showing exactly how the algorithm detects risk, routes intelligently, and creates a warm handoff. The right diagram shows the underlying assessment flow architecture: how 47 rules worth of data collection are sequenced using progressive disclosure and adaptive branching to feel conversational, not interrogative.
John is not an edge case — he is the modal user. Financially stable, operationally dependent, emotionally unprepared. The system's ability to surface his hidden critical moment and route him without human intervention is the core value proposition made visible.
He clicks "I'm not sure" — system provides context. He reconsiders: "No, I'm needed for most major decisions." Flag: High Owner-Dependence triggered.
Readiness 52/100 · Attractiveness 68/100 · "Building" quadrant. Gaps: Operational Independence (41), Systems (38). Dual-score captures what a single number would miss.
PDF generated. John segmented. Sarah (RM) gets a dashboard notification with full context — she arrives already briefed, not starting cold.
Preparing68Attractiveness
Moderate appealGaps: operational independence 41 · systems & documentation 38
Readiness −20 · adds management questions
Advisor matching was built for the owner, not for the sales pipeline. Before any advisor was involved, the owner already had their score, their gaps and a plan they could act on. Every match was to a specialist on PNC’s own Succession Advisory team, never a third party, and that advisor arrived already briefed. The first meeting started with planning, not qualification.
Screens Evolution & Layout Architecture
Single-page vs multi-page wasn't an aesthetic decision — testing showed it. Users needed focus during the assessment and density during results. The same layout couldn't serve both cognitive modes. So the architecture split to match what the user actually needed at each stage.
Users needed different cognitive modes at different stages: deep focus during assessment, comprehensive analysis during results review. A single layout approach failed to serve both needs. The assessment phase required singular focus with minimal distraction. The results phase demanded information density for complex decision-making and data synthesis.
Required singular focus with minimal distraction while maintaining progress awareness
Demanded information density for complex decision-making and data synthesis
Balancing Focus and Context
Users needed different cognitive modes at different stages: deep focus during assessment, comprehensive analysis during results review. A single layout approach failed to serve both needs effectively.
60/40 Split Architecture
The assessment screens employed a 60/40 split layout optimized for focused task completion while maintaining contextual awareness. Left 60% centered the question card with single-focus design. Right 40% displayed live score previews and contextual help.
Left 60% — centered question card · single focus point · radio buttons · clear action buttons.
Right 40% — live score preview · progress indicators · contextual help · real-time impact.
Outcome: 34% less assessment anxiety · 68% reduction in mid-assessment abandonment · 41% improvement in relevance clarity.
Design System — Ensuring Consistency at Scale
135+ screens across 4 persona pathways, 9 scoring zones, and 3 user types (owner, RM, advisor) created a real consistency problem. Without a system, the same component would render differently across assessment, results, and action plan contexts — breaking the trust signal the product depended on. The design system solved three specific problems: score presentation had to look neutral, not alarming; recommendation cards had to scale from 1 to 15 items without layout collapse; and advisor match components had to work across 5 advisor specialisms with different credential structures.
Conveys trust and professionalism. Used for all interactive elements and score displays to maintain neutrality.
Gray palette provides professional foundation while allowing semantic colors to communicate clearly when needed.
Red, orange, and green reserved exclusively for priority levels in action plans. Never used for scores themselves.
Never Use Red/Green for Scores
Avoids judgmental connotations. All scores use neutral Primary Blue to maintain objectivity and reduce assessment anxiety. A 41/100 score in red would trigger defensive reactions before the owner reads the explanation.
Semantic Colors for Priority Only
Red, orange, and green are reserved exclusively for priority levels in action plans and recommendations — never for scores or general UI elements. This creates a clear visual language: when you see red, it means "act on this first."
Blue for All Assessment Elements
Progress bars, score displays, and interactive elements all use Primary Blue for a consistent, neutral experience. This deliberately avoids making high scores feel "good" or low scores feel "bad" — the assessment is diagnostic, not evaluative of the owner's worth.
Typography rationale: Inter conveys reliability and professionalism appropriate for financial services without being corporate-stiff. Optimized readability: line height of 1.6 for body text ensures comfortable reading during long assessment sessions. Performance focused: Google Fonts delivery ensures fast loading and consistent rendering across devices. font-family: 'Inter', sans-serif
Actual Screens
Representative production screens from the 135+ screen system. Branding revised due to intellectual property considerations.
Screen examples shown are limited due to client confidentiality. These represent the design system and approach across 135+ production screens. Branding has been revised due to intellectual property considerations.
Impact & learnings
Measured by owners reaching the right advisor at the right time, not by pages read.
Three Rounds. Data-Driven at Every Step.
Three rounds. Algorithm accuracy first — does the engine match what advisors actually recommend? Then user trust — will owners believe a score? Then RM usability — does this make advisors' jobs easier or harder? Each round had a specific question. Each one changed something.
Does the algorithm match expert judgment?
Do owners understand and trust algorithmic recommendations?
Can RMs effectively use this tool with clients?
Measurable Results at 6 Months
Post-launch data from product analytics, advisor-reported funnel metrics, and validation studies.
Quantitative Impact
How mismatch was measured: the share of first consultations where the owner met the wrong advisor type or came at the wrong time, as reported by PNC’s succession advisors. 67% at baseline, 18% six months after launch.
The algorithm architecture was built to support ongoing refinement without structural rework — dimension weights, rule thresholds, and persona routing logic are all independently adjustable as new assessment data accumulates. The 11% of cases that fell outside the four core personas were tracked post-launch as the primary signal for future persona expansion. The product was handed off with a documented rule governance framework — a structured spec covering each of the 47 rules, their conditions, scoring impact, and owner — so the advisory team could refine criteria without requiring a design or engineering sprint. Engineering received annotated component specs, interaction states, and edge case handling documentation for all 135+ screens.
What I'd Do Differently
I'd codify the algorithm before the UI, not in parallel
We ran discovery and algorithm design alongside each other. That created friction — advisors were being asked to validate rules while I was still designing the screens those rules would surface in. The back-and-forth cost us two sprints. Next time I'd lock the full rule logic with advisors before touching a single layout. The algorithm is the product. The UI is just how you see it.
The Real Problem Is Often Hidden Beneath the Stated Problem
"Owners don't engage with succession planning because they lack education"
"Owners can't objectively assess readiness, don't know which path fits their situation, and meet with wrong advisors at wrong times"
Transparency Is Non-Negotiable for Algorithmic Trust
When an algorithm makes a high-stakes recommendation, the explanation UI is as important as the output. Owners who could see how their score was calculated trusted it. Owners who just saw a number questioned it. In regulated, financial-services contexts especially — show your work.
The most expensive mistake in a complex engagement is building the wrong thing with precision. Three weeks of discovery on this project redirected six months of build — and protected PNC from shipping a product that would have moved nothing. Validating direction before committing to it, and making a data-backed case for a different path to bank leadership, is due diligence I treat as non-negotiable on any regulated, high-stakes product.






