Bringing structure, data, and judgement to BTSE's people strategy

Five-lens diagnostic

Sequenced from context to capacity. Each lens answers a different question; together they form a defensible picture of where each team sits.

The process lens, illustrated

Why Lens 2 earns its place. Often, "team capacity" problems show up at the boundary, not inside the box.

If BD is "understaffed," the answer might be one more BD — or it might be that compliance review is the real constraint and BD only looks slow because of it. The lens decides.

Process

1. Map. One row per role: function, cost, span, manager, charter, OKR linkage, key cross-team handoffs. Pulled from people / project / finance systems where they exist (see closing on data infrastructure).
2. Listen. 1:1 with each department head (45 min): "where would one more person change the curve, where would one fewer not be missed in 90 days, and where in the flow with other teams do you lose the most time?"
3. Triangulate. Cross-check the leader's view against workload, shape, and the cross-team flow. Disagreements are the most useful information.
4. Co-design moves. Each recommendation has a name, a date, and change management measures attached. A re-org without change management is what damages morale, not the re-org itself.

Output — staffing & structure diagnostic by department

A view I would maintain quarterly with the COO. Each team gets both a staffing recommendation (capacity adds, reductions, redeployments) and, where relevant, a structural recommendation (process redesign, span/layer changes, scope shifts, automation). Click a department to see the team-level diagnosis. Sample data only.

The five levers — what each actually does

Only two of these cut labor cost; the others create capacity. Every plan starts from this honest base.

How I would diagnose productivity — two layers

1. Team layer. KPI achievement vs goals over 2–4 cycles, plus cost per FTE and per unit of output. Pulled from existing systems; light manual collection only where data is missing. Place each team on the matrix below.
2. Individual layer (within affected teams). Performance vs goals + strategic fit. Used to make per-person moves defensible.

Team level — what should happen to each team as a whole.

From team to individual

For teams in the "process / leadership fix" or "restructure" quadrants, per-person moves have to be defensible. Two inputs, mirroring the team layer: strategic fit (does the role / skills / seniority match the team's forward direction) on the Y, and performance (KPI vs goals, behaviour, impact) on the X — same axis ordering as the team matrix.

Individual level — what should happen to each person within the team.

From analysis to execution

The seven trackers

The "Source" column shows which Q1 diagnostic each tracker operationalises. Trackers without a source are independent monitoring layers that don't map back to a one-time Q1 output.

Tracker 3 (goal achievement) vs Tracker 4 (productivity): a team can hit all OKRs while being unproductive (over-resourced, soft targets), or be highly productive while missing OKRs (poorly set goals). Both are needed.

Operating cadence

• Weekly 1:1 with COO — exception report only.
• Bi-weekly with department heads — common dashboard view, short Start / Stop / Continue per owner. Layered into existing forums where possible — never a new meeting for a new dashboard.
• Monthly — workforce plan + productivity review.
• Quarterly — org shape + OKR review with the COO.

The data layer — likely a real project, not just plumbing

None of these trackers work without a canonical data view. Three workstreams:

• Connect what exists — HRIS, project tools, finance / ERP, engagement surveys.
• Build what's missing — capability profiles, manager time-use, qualitative project status. Lightweight templates and disciplined manual input where automation isn't realistic.
• Layer AI as the analysis surface — leaders ask natural-language questions of their own data ("show me my at-risk OKRs and cycle-time trend") instead of waiting for a custom report.

I would scope this as an explicit foundational project in the first 90 days — see closing.

Four operating rules

1. Co-author, don't impose. The first KPI conversation with every leader is a working session. Quarterly office hours after that — where leaders can retire, change, or argue for any metric — keep the dashboard a living tool, not a fixed one. Imposed metrics generate compliance, not improvement.
2. Outcome over activity; tiered visibility. Outcome metrics feel like accountability; activity metrics feel like surveillance. Across the leadership tier, share high-level outcomes (department OKR achievement, direction-of-travel on key business metrics) — that breaks data silos, aligns department goals to company strategy, and turns the dashboard from COO oversight into peer accountability. Keep granular and sensitive data private to the leader and COO (individual performance, named cases, comp). The aim is constructive transparency, not a competitive scoreboard.
3. Audit yourself first. The COO Office's KPIs (forecast accuracy, time-to-resolution on escalations, leader satisfaction with the office) sit at the top of the page. Reciprocity matters.
4. Be the first call when a number goes red. "I noticed your eNPS dropped 8 points; here are three hypotheses — 30 min?" — not "explain yourself in tomorrow's review."

Invest in leaders, not only in the dashboard

A dashboard only works if leaders know how to use it. In parallel:

• Goal-setting — many leaders set inputs, not outcomes. Coaching here changes the data the dashboard sees.
• Coaching skill — KPIs become useful only when they fuel real coaching conversations.
• Being a coachee — leaders modelling openness with their leader is the only way the cultural change travels down.

Five-step diagnosis

1. Quantify. Annualised attrition, regrettable share, tenure-at-exit, vs company & industry benchmark. Confirm there is actually a problem and how big it is.
2. Pattern. Cut by sub-team, manager, level, gender, geography, hire vintage, performance rating. Most "department-wide" turnover is concentrated.
3. Listen. Exit interviews; stay interviews across a representative cross-section (performance levels, tenure, sub-team) so the signal isn't biased by who you talked to; confidential pulse survey to the whole department.
4. Triangulate. Cross-reference with manager 360 scores, comp benchmarking on roles in question, workload signals, recent re-orgs, and — importantly — the cross-department interactions. Many "team" problems are actually friction at the touchpoints with another team.
5. Categorise root cause. Manager / Compensation / Career path / Workload & burnout / Mission alignment / Hiring profile mismatch / Cultural friction / Process friction with another team / Scope of work itself. The action plan is different for each.

Decision tree → action

Action menu — match move to diagnosis

Sequencing

Hypotheses, grouped

Investigation plan

1. Open conversation with Joe. "Walk me through the original headcount case. What changed?" Joe often knows more than the data shows.
2. Process scan. Map the team's workflow end-to-end. Where does work pile up between steps? Where are people waiting? Don't assume where the bottleneck is — let the data show it.
3. Performance distribution. Look at how output is spread across the team. Are a few people carrying disproportionately more than others? A long-tail pattern points to uneven capability, regardless of headcount.
4. Cross-functional view. Interview the 2–3 peer teams that depend on Joe's. Their experience is valuable input.
5. Onboarding audit for the 7 hires — time-to-productivity, 30/60/90 completion.

Metrics on the tracker

Improvement levers — match to diagnosis

• Decisions all funnel through one person → delegation coaching; deputy or RACI redesign so decisions don't all queue on the leader.
• Performance management has been avoided → COO-backed coaching for Joe; HR partner for the difficult conversations. Hiring is not a substitute.
• Workflow has bottlenecks (from the process scan) → adjust the process at the points where work piles up, before adding capacity. Adding people upstream of a bottleneck only grows the queue in front of it.
• Skill mix doesn't match the work → redeploy or cross-train; pause further hiring until the mix is corrected.
• Demand fell → repurpose the team. Don't fight the market.

1. Take it seriously, don't act yet. Three independent signals is meaningful. But "slow" is a subjective word — slow on what, vs what expectation, against what SLA?
2. Get specific with each leader. 30-min 1:1 with each of the three. "Tell me the last three concrete examples." Specifics reveal whether this is a process issue, capacity issue, or capability issue.
3. Understand the inter-departmental system. The 4 departments (Joe + 3 peers) form a system — the issue may be how they interact (handoffs, dependencies, intake quality), not Joe alone.
4. Hear Joe's view — on the same situations. Take the specific examples the three leaders cited and ask Joe about each, without framing them as complaints. "Walk me through how the [specific project / decision / handoff] played out." Joe may describe the same events very differently — unclear briefs, shifting priorities, dependencies he can't control. Comparing both sides on the same facts is more useful than asking Joe to defend himself against a label. The framing-back to Joe (if needed) is downstream — after the diagnosis, by the COO.
5. Verify with data. Joe's KPIs vs goals; cycle time on Joe's deliverables to peer teams; SLA breach rate on cross-functional requests; intake quality (are the asks well-specified to begin with?).
6. Categorise the root cause and provide suitable measures:
   • Process within Joe's department → standardise intake, prioritisation, internal workflow
   • Process between departments → SLA agreements at the touchpoints; intake-quality requirements from peers
   • Capacity → either reduce commitments or add capacity (Q1.1 framework)
   • Capability → coaching plan with the COO; if not closeable in 90 days, role change

Most "Joe and Jane" problems are stuck between storming (open friction) and norming (no agreed way to work together). The objective is not to skip storming — it's to move through it deliberately. Two strong professionals don't have to like each other; they have to know how to work together.

Five-step intervention

1. 1:1 with each, separately. Surface the friction — style, values, credit / territory, a specific past incident. Often one holds the bigger grievance.
2. Joint, facilitated session — interface design, not therapy. Outputs: shared definition of the work overlap; RACI on overlapping decisions; rules of engagement (no public disagreements; escalate to me before each other's manager). Make explicit that differences in style are normal — a broken interface for the company is not.
3. Structural moves to reduce unnecessary collaboration. Most "Joe vs Jane" friction is over-collaboration on work that should have a clean interface and async cadence.
4. Aligned shared OKR. One outcome they both win or lose together. Forces practical cooperation without pretending to like each other.
5. Manager coaching for both. Each gets specific feedback on how their behaviour is read by the other. Usually neither knows.

Concrete example — what the RACI looks like

Worked example: Joe leads BD/commercial, Jane leads Customer Onboarding / Ops. They share an interface around enterprise pricing exceptions and integration scope. The most common "Joe vs Jane" failure mode is both arguing on every deal.

Once written down, 80% of the disagreement evaporates — they were arguing about decision rights, not about the deal.

Anti-patterns

• Group hug. Forced public reconciliation. Adults find it insulting.
• Picking a side, even quietly. The loser becomes a flight risk; the winner becomes harder to manage.
• Ignoring it. "They're both senior, they'll figure it out." They won't — and the team beneath them is already paying the cost.

The three-layer system

Two often-missed elements

How AI makes this scale

AI is what makes brand consistency at 30-people-across-time-zones achievable rather than aspirational. Two highest-leverage uses:

• Auto-scored call analysis — every call rated on product accuracy, compliance phrasing, on-brand positioning. Coaching focuses on the 10% flagged, not random samples.
• Real-time canonical answers — BD asks "what's our position on stablecoin payouts in Vietnam?" and gets the approved answer with the source. Stops cowboy positioning at source.

Governance — who owns what