Methodology ● Current edition · Published May 2026
The Levelset rubric is five pillars, twenty-two sub-criteria, five-point scoring with anchor descriptions per score per criterion. It is published in full here so any score you receive can be re-read against the criterion that produced it.
01 · The thesis
Why five and not seven
The rubric started from a single observation. In the post-mortems of programs that missed go-live, the explanation almost never traces to the platform. It traces to sponsorship that drifted, to a change-management workstream that was never elevated, to a workforce that was trained the month before cutover, to a communications cadence that left rumor to fill the silence, and to a steering committee that re-litigated the same design decisions every month. The technical work was usually fine. The organizational work was treated as scaffolding around the real program.
Five pillars is the smallest set that distinguishes those failure modes from each other. Fewer collapses the categories into a single "soft side" bucket that gives no leverage for action. More dilutes the rubric into adjacent disciplines (data architecture, vendor management, financial modeling) where the failure mode is different and the rubric we use here would mis-grade.
The pillars are weighted equally. We have tried weighting Sponsorship and Change higher, on the grounds that those failure modes are the most common, and found that the recommendation set gets worse, not better. An equal weighting forces the analyst to look at each pillar on its own evidence rather than nudging the score toward what the rubric already weighs.
02 · The rubric
5 pillars · 22 sub-criteria
Whether the program has a strategy that the operating groups understand, or an implementation plan that IT is executing on. 5 sub-criteria.
Grades
Does not grade
Sample probes
Whether the people doing the work have been brought into the change or had it sent to them. Where OCM reports. How resistance is being identified. 4 sub-criteria.
Grades
Does not grade
Sample probes
Whether the program tells a coherent story to the people whose job it changes. CEO message, cadence, and translation across operating groups. 4 sub-criteria.
Grades
Does not grade
Sample probes
Whether training is role-based and workflow-oriented or system-click. Whether SMEs and super users have time blocked to co-design. 4 sub-criteria.
Grades
Does not grade
Sample probes
Whether the program can detect drift and recover from setbacks. Leading indicators, course-correction history, and the rhythm of contentious decisions. 5 sub-criteria.
Grades
Does not grade
Sample probes
03 · The anchor scale
One sub-criterion, all five anchors
Every sub-criterion has anchor descriptions for each of the five score levels. This is what keeps the rubric defensible: there is a definition of what "3" looks like, not just a gestalt. Below is one example sub-criterion, drawn from Pillar 01.
Decision rights at the working committee
The question is whether the working committee has the authority and the rhythm to resolve contentious design decisions on its own, or whether everything routes to steering by default.
Critical risk.
No working committee charter exists. Every contentious design decision escalates to the CEO or CFO. Steering is in tactical conversations regularly. Decisions take 2 to 4 weeks. The program is governance-bottlenecked.
Significant risk.
A charter exists on paper but is not followed in practice. Steering routinely re-litigates working committee decisions. Decisions land in 1 to 2 weeks with rework. Operating-group reps escalate around the committee chair rather than through them.
Mixed health.
Charter exists and is mostly followed but lacks explicit tie-breaking authority. Most decisions land in 5 to 10 business days. Contentious decisions still escalate to steering. The committee chair is leaning on personal credibility, not structure.
Every other sub-criterion in the rubric is anchored the same way. The full document, all sub-criteria with their anchor scales, is available on request and shipped with every Enterprise tier engagement.
04 · How scoring works
From sub-criterion to overall health
A pillar score is the rounded mean of its sub-criterion scores. The overall program score is the rounded mean of the five pillar scores. There is no weighting hidden inside the formula.
A 3 means mixed health. It is not a passing grade. It is identifiable risk that will compound if not addressed, and the report's roadmap is sequenced to address it. In our calibration work, programs scoring 4 or above before go-live were consistently more likely to go live on time than those scoring 3 or below.
The diagnostic also reports a separate confidence score per pillar, 1 to 5, that captures how much signal the interview surfaced for that pillar. A pillar with a score of 4 and confidence of 2 deserves more interview time than its number suggests.
05 · Honest omissions
What this rubric won't tell you
Not graded
Technology choice and architecture.
We do not grade whether SAP, Oracle, or Workday was the right choice. We do not grade integration architecture, data model design, or technical debt. Those are real risks, and they need their own diagnostic. We are not that.
Not graded
Vendor and SI relationships.
We do not grade whether your SI is the right one, whether the contract is structured well, or whether the rate card is reasonable. We grade only how the SI relationship interacts with the human-side risks above.
Not graded
Business case and ROI math.
We grade whether you have agreed-upon KPIs and outcomes. We do not validate the numbers. A program with a great rubric score and a fictional business case is still going to be in trouble at year three.
Not graded
Personality and culture diagnostics.
We do not grade your CEO, your operating-group heads, or your team culture. The rubric describes structural conditions, not personalities. Those structural conditions are tractable. Cultural change is a different and longer engagement.
06 · Calibration
Where the rubric came from
The rubric is calibrated against post-mortems and live transformation engagements, weighted toward programs that missed go-live or went live and then quietly under-delivered. Most rubrics are written from success cases. We have found that success cases tell you less than failures: they share too many uncontrolled variables.
The mix is roughly 60% ERP (SAP, Oracle, Workday), 25% HCM and finance transformations, and 15% operational and supply-chain programs. Geographically, the set skews North America and Western Europe. Programs under $5M are not represented in the calibration set and are not within the rubric's scope.
Three rubric criteria were discarded in prior iterations because they correlated with seniority of the interview subject rather than with program outcomes. Two criteria were added in the current edition (the Recovery pillar's "leading indicators" and "decision velocity" sub-criteria) because they correlated more strongly with on-time go-live than any of the criteria we already had.
07 · The interview style
How the analyst behaves
The analyst's job is not to take your word for it. The job is to push until either the program has a real answer or the absence of one becomes the score. Below is the kind of exchange the analyst is calibrated to drive.
How is your steering committee structured?
CFO chairs it. CEO is on it. My COO peers are on it. We meet monthly.
When was the last time the steering committee decided something that the working committee had already taken a position on?
Last month. There was a question about how procurement requisitions would route in the new system.
Which way did steering decide?
Against the working committee's recommendation, actually.
How did the working committee chair learn about that?
I told them after the meeting. They were not happy.
And what is the next contentious decision the working committee is queued up to make?
Probably the role-based access model. It's contentious because finance wants tighter controls than operations does.
What does the working committee chair think will happen if they make a call this time?
Honestly, after last month, they're probably going to bring it to steering preemptively. Which is exactly the dynamic I was hoping to avoid.
ScoringThe subject walked in believing their governance was healthy. The probe surfaced that steering had overridden the working committee on a tactical decision, that the chair had no warning, and that the next contentious decision will preemptively escalate. The sub-criterion was scored 2 (significant risk) with high confidence. The score reflects the structural pattern, not the subject's effort.
The rubric is a tool
25 min · PDF by Friday