Value

What it is worth, domain by domain

Running each regulated object as a governed holon, one source of state, rules, history, and role-specific views, is not an architectural nicety. It removes the reconciliation, re-entry, and re-mapping work that quietly consumes most life-science operating budgets. Below, the added value is split by the same domains and holons you met on Practice, described qualitatively and quantitatively, with the calculation shown for every figure.

read this first

How every figure is calculated

The mechanism is always the same. Today, the same fact is captured, reconciled, and checked again in every system and for every use. In a holon, state, rules, history, and views are governed once and reused, so the repeated work disappears. Each figure therefore compares today's per-system, per-use effort against a single governed object:

the shared formula Saving = (manual effort per task × how often it repeats) − (governed effort, done once). Money converts saved hours at a loaded specialist rate of ≈ $80 / hour unless noted. The reuse effect is grounded in a real enterprise-scale unification of 20+ years of clinical data, where ~3,500 studies, ~900,000 subjects and 3M+ source variables collapsed onto ~4,000 governed target variables across 50 domains (a ~750× consolidation). All numbers below are illustrative planning estimates, not audited results, and should be re-based on your own volumes.
Domain: Clinical Science

Holon: Clinical Trial

One governed trial record replaces the protocol / EDC / CTMS / TMF patchwork. Every dashboard, monitoring view, and study report reads the same state, rules, and history, so milestones stop waiting on cross-system reconciliation and audit trails never diverge.

Effort −45% data-management effort per study

How: a milestone reconciliation across 4 systems ≈ 200 h; reading from one governed record ≈ 110 h → ~90 h saved × ~6 milestones ≈ 540 h saved / study.

Time ~40% faster to database-lock readiness

How: continuous boundary checks (visit windows, eligibility, query aging) replace the manual pre-lock sweep → 6 weeks → ~3.5 weeks.

Money ≈ $43k saved per study (labor)

How: 540 h × $80. Example: a 25-study portfolio ≈ $1.1M / year, before the value of earlier locks.

Domain: Regulatory

Holon: Dossier Submission

Submission content is generated from state that was already governed and validated, instead of being assembled module by module against the deadline. Late discovery of missing or inconsistent data, the classic filing-date risk, disappears because validation is always on.

Effort −50% on dossier assembly & QC

How: module assembly + cross-module consistency QC ≈ 320 h manual; generating from governed state + SHACL validation ≈ 160 h → ~160 h saved / submission.

Time ~3 weeks earlier filing readiness

How: continuous validation removes the pre-submission "consistency sweep" (~15 working days of manual checking against eCTD and controlled terminology).

Money ≈ $13k saved per submission (labor)

How: 160 h × $80. Example: plus a protected filing date, each avoided month of delayed approval on a mid-size product is worth far more and is not counted here.

Domain: Manufacturing

Holon: Batch / Lot

A release view is computed from linked, governed manufacturing state, materials, process version, test results, deviations, instead of being pieced together from LIMS, MES, and paper. Genealogy is traceable end to end in seconds, and the certificate of analysis is a projection, not a rebuild.

Effort −35% per batch release review

How: assembling and checking a batch record across systems ≈ 24 h; a computed release view ≈ 15 h → ~9 h saved / batch.

Time 5 → 2 days to disposition a lot

How: genealogy, CoA, and open deviations are already linked to the batch holon, so there is no manual trace-back across systems.

Money ≈ $720 saved per batch (labor)

How: 9 h × $80. Example: 600 batches / year ≈ $430k / year, before faster release frees inventory and working capital.

Domain: Quality Assessment

Holon: Quality Event

A deviation, its investigation, and its CAPA stay connected to every batch and product they touch, and those links never rot. Because events are linked rather than scattered across spreadsheets and the QMS, recurrence and systemic patterns become visible instead of being rediscovered each time.

Effort −40% per investigation

How: gathering evidence + scoping impact across QMS and spreadsheets ≈ 30 h; with linked context ≈ 18 h → ~12 h saved / event.

Time 45 → 30 days investigation-to-closure cycle

How: impacted batches and products are already linked, so the manual impact search that stalls most investigations is removed.

Money ≈ $960 saved per event (labor)

How: 12 h × $80. Example: 500 events / year ≈ $480k / year, before fewer repeat deviations reduce the event count itself.

Domain: Safety

Holon: Safety Case

A safety case carries its full timeline and its links to related events, and reporting timelines are governed by rule rather than by hand. Because cases stay connected, recurring patterns across them surface in time to act, rather than after the reporting window has closed.

Effort −30% per individual case (ICSR)

How: case compilation + consistency checks ≈ 8 h; with a governed case holon ≈ 5.5 h → ~2.5 h saved / case.

Time weeks → days to first signal detection

How: cross-case aggregation is continuous over linked cases, replacing the periodic manual pull that delays signals.

Money ≈ $200 saved per case (labor)

How: 2.5 h × $80. Example: 10,000 cases / year ≈ $2M / year, before the risk avoided by earlier signals.

Cross Domains

Holon: Medicinal Product

One product identity exposes the right regional view on demand, instead of product attributes being re-entered per market and per system. "What is the approved strength here?" stops taking days. This is the reuse backbone of the whole model: map once to a shared identity, project everywhere.

Effort −60% on product-data maintenance

How: attributes today are re-entered across ~8 systems and markets; a governed identity is updated once and projected → one update replaces eight.

Time days → seconds to answer a regional attribute query

How: a region/role projection reads one identity, so "approved strength here?" is a query, not a cross-system hunt.

Money ≈ $250k per year, mid-size portfolio (labor)

How: removing ~7 duplicate re-entries × ~450 h each × $80 ≈ $252k. This mirrors the ~750× source-to-target consolidation seen at enterprise scale.

Domain: Sample Management

Holon: Lab Result

A result carries its provenance and specification checks with it, so instrument provenance and review history never thin out as the value moves from instrument to LIMS to report. What is "official" is decided once and travels with the result.

Effort −50% on result review & transcription

How: manual transcription + spec check across instrument → LIMS → report ≈ 12 min / result; with checks that travel with the result ≈ 6 min → ~6 min saved / result.

Time hours → minutes to an official, released result

How: specification checks are attached to the result, removing the re-verification handoffs between systems.

Money ≈ $8 saved per result (labor)

How: 6 min ≈ 0.1 h × $80. Example: 200,000 results / year ≈ $1.6M / year.

the payoff, as a score

A Net Promoter Score the graph computes for itself

No survey needed. Because every holon's state, rules, history, and views are linked, shareable, and reusable, each object can in effect be asked “would you recommend this way of working?”, and its four graphs already hold the evidence to answer. We keep the classic scale, respondents score 9–10 as Promoters, 7–8 as Passives, 0–6 as Detractors, and NPS = %Promoters − %Detractors (−100 to +100), but replace the survey question with facts read straight off each holon's graphs.

the fact-based formula Recommendability (0–10) = the average of four graph-native signals, one per graph: how complete the state is, how well it conforms to its rules, how complete its provenance is, and how far its views reach across roles and other holons. Every instance is classified Promoter / Passive / Detractor on that score exactly as in a survey, and per-holon NPS = %Promoters − %Detractors. The master NPS pools every instance across all seven holons into one score, weighting each holon by how many instances it contributes.
scene graph

State completeness

Share of the object's mandatory fields that are populated and valid right now.

boundary graph

Rule conformance

Share of applicable SHACL constraints the instance satisfies without exception.

event graph

Provenance completeness

Share of the object's events with an unbroken, attributable lineage.

projection graph

Reuse reach

Role views and cross-holon links served from this one object, against target.

worked example · one clinical trial instance

A trial instance scores 9.4 on state completeness, 9.6 on rule conformance, 9.1 on provenance completeness, and 8.7 on reuse reach → recommendability ≈ 9.2, a Promoter. Repeat this across the whole population of instances: say 78% land as Promoters, 16% Passives, 6% Detractors → per-holon NPS = 78 − 6 = +72.

HolonFact-based NPS
Clinical Trial+72
Dossier Submission+64
Batch / Lot+69
Quality Event+61
Safety Case+58
Medicinal Product+81
Lab Result+66
Master score +70 fact-based NPS, all seven holons pooled

How: pool every instance across all seven holons (equivalent to weighting each per-holon NPS by its instance count), then apply the same %Promoters − %Detractors formula once, at the population level. All NPS figures on this page are illustrative planning estimates, to be re-based on your own graphs.

why the savings compound

Each holon saves work on its own, but the real return is shared: the governed product identity feeds the trial, the batch, the safety case, and the submission at once. Map a fact once, and every domain that reuses it stops paying to re-create it, which is why the per-holon figures above are conservative when a whole organization runs this way.

The point: value is not a slide, and it is not a survey either. Because every fact is linked, shareable, and reusable, each holon earns a fact-based NPS straight from its own four graphs, and those seven scores integrate into a master NPS of roughly +70. See it end to end on the Case page, then see how it is delivered on Method.