Five bands, one spine.
The internal ladder mirrors the offer ladder — same spine, built once, pointed at people instead of products. We grade on producing the weave, not on generic consulting skill. Below: what each band does on the core competency (problem-solving via T×E×P).
| Band | Problem-solving (T×E×P) | Mirrors |
|---|---|---|
| 1 · Analyst | Works one thread at a time. Structures evidence on a single axis when scoped; produces clean inputs; relies on a senior to integrate. | Rung 1 — free signal |
| 2 · Associate | Owns one axis end-to-end and sizes it defensibly — traces every number to an assumption. Sees where two threads interact and flags the tension; doesn’t yet resolve it. | Rung 2 — sized brief |
| 3 · Engagement Manager | Threads all three axes onto one decision and produces a verdict (act / wait / hedge / reprice / phase). Owns the synthesis. Names the binding constraint. | Rung 3 — pilot |
| 4 · Associate Partner | Frames the right decision to weave around — reshapes an ambiguous question into the decision that moves value. Pressure-tests others’ weaves for stale assumptions. | Rung 4 — retained |
| 5 · Partner | Sets the method. Defines what “decision-grade” means here, which weaves the firm will and won’t stand behind, and evolves T×E×P as IP. | Rung 5 — advisory |
Graded across five competencies: Problem-solving · Client impact · Leadership/people · Firm-building · Sector mastery.
The placement rule
Your current band = the highest band B where you score ≥ B on at least 3 of the 5 axes, and no axis sits below B − 1. One or two axes may spike higher, and one may sit a single band low — but a two-band gap on any axis holds you at the band below. That gap is your promotion blocker.
Worked example. Scores PS 4, Client 3, Leadership 3, Firm-building 2, Sector 4 → current band 3 (EM). Blocker: Firm-building (a two-band gap — priority). Spikes: problem-solving & sector (4). Lean on the weave while building a reusable asset and taking account ownership. Promotion is a sustained read — you operate at the next band before you’re titled it.
Score the band, then the proficiency.
Two questions, kept separate: which band’s work are you doing on an axis, and how reliably. You only “operate at” a band when you do its work at Reliable proficiency — doing it once, with help, on a good day, doesn’t count.
| Level | Name | Meaning | Operating there? |
|---|---|---|---|
| 0 | Not yet | Hasn’t demonstrated this band’s work on this axis. | No |
| 1 | Emerging | Can do it with guidance; inconsistent. | No — still the band below |
| 2 | Reliable | Does it independently, repeatedly, unsupervised. | Yes — this is the bar |
| 3 | Role-model | Sets the standard; develops others at this level. | Yes — a promotion signal |
The evidence rule
For every axis: an Evidence line (a specific, recent ≤2-quarter example — not “I’m good at this”), a Confidence (H/M/L), and what would change it. A band you cannot evidence is a band you do not hold — score down. A real case worked to an EM-grade verdict is direct proof of band-3 problem-solving.
The steps
Self-score
Per axis, tick the behavioural indicators; the highest band you hit ≥2 indicators at Reliable proficiency is your axis score (1–5). Attach evidence + confidence + what-would-change-it.
Place the band
Apply the placement rule (≥3 axes at-or-above, nothing two bands low).
Sponsor calibration
Someone one band above scores you independently against the same indicators. Where you differ, the tie-breaker is the recorded example, not seniority. No evidence → default down.
Sustained read
Hold the band across ~2 consecutive quarters before you’re titled it. Group moderation keeps a “4” meaning the same for everyone.
Four disciplines.
One and two are table stakes; three and four are what clients pay for. “Most weak work fails at Define, not in the maths — it answers a fuzzy or wrong question well.”
1 · Define rigorously
Name a decision, not a topic. A day sharpening the question saves a week of misdirected analysis.
2 · Work the process
Structure (tree) → prioritise (binding constraint) → plan → conduct → synthesise → recommend.
3 · Be distinctive
Expand · Link · Distill · Lead. Correct is the floor; these four are the edge.
4 · Harness collaboration
Obligation to dissent. Evidence settles arguments, not rank.
The Decision Brief — one page, before any work
| Field | What good looks like |
|---|---|
| Decision in play | A single decision with a verb and an owner: “Should [who] [do what], given [what tension]?” |
| Decision-maker & success | Who signs; what a good answer lets them do; the accuracy grade — directional / decision-grade / audit-grade — and timeframe. |
| Scope | The boundary — markets, assets, horizon, geographies in. |
| Constraints | What’s fixed or ruled out — capital, policy, what the buyer won’t move. |
| Stakeholders | Who shapes / blocks — the T/E/P actors (operators, capital, regulators). |
| Sources of insight | Which models, the registry, named experts, the desk. |
A good brief: two analysts read it and work the same problem; it names a decision, not a subject; you can state what answer would be wrong — it’s falsifiable.
Issue tree vs hypothesis tree
Issue tree (question-led) when you have no hypothesis — exhaustive, slower. Hypothesis tree (answer-led) when you have a lean — faster, but ask “what would disprove this?” at each branch. The AIA default first cut is always T×E×P.
The binding constraint
Score each branch on impact × decisiveness. High-impact + knowable → work it now (the binding constraint). High-impact + unresolvable → name it as the key uncertainty and scenario it. The “so what” test: if a branch came out either way and the recommendation doesn’t change, it isn’t binding.
Expand · Link · Distill · Lead
| Expand | Construct multiple perspectives — options and rival hypotheses before converging. What else could be true? Marks band 4. |
| Link | See how the threads move each other — second-order effects, spillovers (a fuels-import shift reprices port, rail, storage). Band 4 / sector mastery. |
| Distill | Compress to the core — the one-sentence verdict, the decisive chart. Don’t distill away the binding caveat. Marks band 3. |
| Lead | Frame the problem, type the decision (big-bet / cross-cutting / delegated), manage the room. Marks band 5. |
The two load-bearing links in the 7-step method
Step 2 = T×E×P — our standard issue tree is always the three threads onto one decision; disaggregation is where the weave lives. Step 3 = the binding constraint — pruning three plausible threads to the one that decides it. That’s the EM move.
Same decision, five depths.
Five worked cases — each takes one decision and shows what every band delivers on it. The decisive thread is different each time, which is the point.
| Case | The decision | Decisive thread |
|---|---|---|
| 01 · Mothballed refinery | Fund a refinery restart, or write it off and build for imports? | Economic (publishable) |
| 02 · R98bn corridor | Commit ~R98bn to a greenfield port + rail, or sweat brownfield? | Political / fiscal (publishable) |
| 03 · Gas monetisation | Which plant × port × mode monetises the gas — single project or phased portfolio? | Portfolio (anonymised) |
| 04 · LNG political read | Commit to the terminal now, or wait — is the political path clearing or narrowing? | Political (genericised) |
| 05 · Freight-mode | LTL / FTL / hybrid per shipment — and renegotiate the carrier rule? | Organisational (anonymised) |
The costume line
“A confident number with no provenance is band-1 work wearing a band-2 costume.” Every material number carries a value, a source, a confidence and a named owner.
Case 01 — the binding constraint
“The binding constraint is structural diesel-import dependency under EV uncertainty — not total refining capacity.” Verdict: don’t restart; build for the diesel deficit.
Case 04 — the political read
“Good news on the economic thread is masking a narrowing political path.” Good news on one axis can be bad news for the decision.
Case 05 — honest sizing
Gross mode-optimisation looks like ~15–25%; net is ~2–5% after the cost of consolidation discipline. Quoting net not gross is a band-4 defensibility marker.
Teach, then do.
The half-day spine: areas of competence → the framework → typical cases → crack one live → how feedback works. The live clinic gives each group a case brief (answer key withheld): fill the Decision Brief, build the issue tree, name the binding constraint, one-line verdict.
“Your synthesis didn’t distil to a verdict — that’s the EM move, and it’s your unlock,” not “you’re not strategic enough.” A named, learnable move + a band = structured, growth-directed feedback. Coaching prompts map to the toolkit: “Is that a decision or a topic?” (Define) · “Is the first cut T×E×P?” (Structure) · “Which one decides it?” (Prioritise) · “Where’s that number from, and what would change it?” (Defensibility).