Client Organizational Readiness Classifier

A readiness classifier exists so you stop presenting a recommendation the client cannot execute this quarter as if they can. The score is a sequencing and support decision. It is not a veto, and it is not permission to water the recommendation down.

The point of this classifier is quality. You still say the right thing. You say when it starts, who has to be in the room, and what support has to sit beside it. A low score that silently deletes the idea from the pack is a different failure: the client never hears the argument, and you never find out whether they would have backed it with a different sequence.

Readiness as an excuse for timid advice is the same failure in different clothes. If the operating or financial case holds, keep the recommendation. Put it in a later wave, staff a capability gap, or name the political blocker in a restricted note. The circulating pack still carries the idea.

Ranking impact, effort, and urgency is a different job, done by an initiative prioritization matrix generator. Readiness tells you what can actually start. It does not tell you which idea is best.

Do not blend them into one number. The fix for each is different, and a blended "readiness" score hides which fix you owe.

Capacity is competing programs, calendar, management attention, freeze windows, and open roles that already absorb the same people. The fix is sequence, fewer concurrent workstreams, or waiting for a freeze to lift.

Capability is skills, systems, and process maturity for this specific recommendation. The fix is staffed support, a smaller first slice, paired delivery, or training that has a named owner. Capability is not culture.

Politics is who would lose budget, headcount, or identity, who can actually say yes, and who can stall without saying no. The fix is coalition, a different first site, or a wait. It is almost never "communicate more," and it is never "they lack skills."

If you collapse the three, every hard recommendation becomes "the organization is not ready," which is unactionable and easy to hide behind. Report politics apart from skills on every row. If a cell has no evidence, write unknown. Do not average unknown with a real observation.

Do not invent a composite change-readiness index, and do not dress the output as a named change-management score. Tools in the Notion, Excel, and Microsoft Copilot class will hold notes, tables, and a first-pass draft. They do not know this client's shop. Treat them as a worksheet, not as a diagnosis.

The model should score from evidence you already collected, not from a culture narrative.

Feed it three kinds of source, and forbid a fourth:

1. Interview notes and transcripts, including contradictions across roles from a multi-stakeholder interview transcript analyzer. Who agrees in the room and who contradicts in a 1:1 is often the political signal.
2. On-the-ground observation. Who attended, who cancelled, whose dashboard is actually open, whether last year's program still has an owner, whether the "single source of truth" is a spreadsheet on someone's desktop.
3. Observable facts. Concurrent programs on the same team, freeze dates, open roles, who can approve spend, what happened to the last similar recommendation.

The failure mode is the model guessing culture from a handful of interviews, usually people the sponsor chose. That sample is the sponsor's coalition. It is not the organization. If six people said "we are change-ready" and you never walked the floor, the capability column is still unknown.

Require a citation on every scored cell: a quote, a named observation, or a fact. If the model cannot point to one, the cell stays unknown. Unknown delays a wave. It does not authorize dropping the idea, and it does not authorize writing "low cultural readiness" into a document that will travel.

This is a worked example with made-up firms, written to show the cuts. It is not a case study and it has no results.

A partner at Cole & Vane is taking a shared-services finance hub to Harborline's steering committee. The operating case is to concentrate close, payables, and master data in one hub, starting with two regions whose close is still manual. An analyst runs a first-pass classifier over interview notes. The model returns a single "low organizational readiness" label and a paragraph about "change fatigue." The tempting edit is to replace the hub with "strengthen local controllers," which is timid advice wearing a readiness score.

The partner splits the row into three columns.

Capacity: finance already has two other programs this year, and the close calendar has no spare weeks until after Q1. That comes from the program list and from sitting in the close war room, not from a vibe.

Capability: two regions still close in Excel with undocumented adjustments. The hub recommendation is still right. Wave one is those two regions, with a named close lead from the hub sitting with each local team through two cycles.

Politics: regional CFOs hired the local controllers and treat that span of control as part of the job. Centralizing threatens that. This sentence does not belong in the circulating deck.

The hub stays in the pack. Wave one is the two manual regions plus a sponsor-backed mandate. Wave two is the rest, after Q1. The implementation business case generator still builds the case for the full hub, with wave one as the conservative case, not a different recommendation. Political notes stay in a restricted file. The partner tests the sequence with the sponsor privately: "we are not dropping this; we are not leading with the regions where you do not yet have cover." If the sponsor says the political read is wrong, the partner believes the sponsor and marks the model wrong.

Writing politics into a circulating deck is how a restricted observation becomes a forwarded slide. "Regional CFOs will resist" in a file that hits forty inboxes is a problem for someone, often not you. Translate politics into phasing and named support in anything that leaves the team.

Working order:

1. Score in a restricted worksheet. A locked Notion page or a separate Excel workbook that is not the client pack. Microsoft Copilot can draft the first pass from notes in that file. Do not ask it to "summarize for steering" from the same file.
2. The partner edits the political column. Analysts own capability and capacity evidence. Partners own names, motives, and what can be said.
3. Private sponsor test, before the pack is frozen. You are checking whether the sequence and support match what the sponsor sees, not asking permission to keep the idea.
4. Circulating pack: recommendation intact, waves and support visible, no villains. Rehearse remaining objections with a recommendation adversarial stress-tester, using the restricted political notes as input to the rehearsal, not as slides.
5. After go, a workstream delivery risk predictor watches whether the capacity assumptions were real. A readiness score is a hypothesis about the next quarter. Delivery data is how you find out.

Trial this on one recommendation set you are already taking to a steering committee. Keep using it if hard ideas stayed in the pack, political sentences never left the restricted file, and the sponsor's private read matched or corrected the political column before anyone else saw it. Stop if the team used a low score to delete a recommendation they did not want to defend. The classifier is for sequence and support. It is not a way to make the advice smaller.