Multidimensional value metrics
Editorial note. This chapter preserves the earlier concise English edition. A full translation of the expanded Spanish edition published on July 21, 2026 is in progress.
IMIA asks whether the organization is ready. This chapter asks whether the intervention was worth it —and for whom. The thesis of the bridge insists on measuring value in the person, not in the model. That requires an explicit measurement system, agreed at the start and reviewed at closure, not a slogan invoked at the end when the budget must be justified.
Most projects defend themselves with vanity metrics: models deployed, chatbot queries, hold-out accuracy. Those figures can rise while the organization gains nothing. A multidimensional dashboard avoids post-hoc justification with whichever metric looked best and avoids ignoring externalities —well-being, equity, maturity— that determine whether the outcome is sustainable.
The “GenAI Divide”: the macro evidence
That a single number is not enough is not an aesthetic preference: between 2025 and 2026 the evidence at scale reached the same conclusion through the big figures. A preliminary MIT report —the one that popularized the term GenAI Divide— finds that the vast majority of organizations get no return from generative AI despite massive investment. It should be cited carefully, being preliminary; but what matters is not the number, it is its diagnosis: the failure is attributed not to the models or the budget, but to an organizational learning gap —deployed systems do not retain context or learn from operations—. It is the bridge thesis said by others: the bottleneck is sociotechnical, not technological. The divide separates those who treated the problem as one of engineering from those who treated it as one of organizational learning; the bridge lives on that second side.
And there is hard evidence of where it does pay off. A journal study (Brynjolfsson, Li and Raymond, 2025) measured 15% more cases resolved per hour in support, with a revealing distribution: the least experienced workers improve far more than the experts. It is the validation of “value in the person” and of the multiplier with a sign —AI well placed levels up those who knew least, rather than merely speeding up those who already knew—. The contrast that closes the argument: according to McKinsey’s late-2025 survey, about 6% of respondents qualified as AI high performers and attributed at least 5% of EBIT to AI; 39% reported some EBIT impact. The distance between “we deployed AI” and “we captured value” is exactly the one that separates the vanity metric from the value metric. The macro figure does not replace the five-axis dashboard: it justifies it. If at global scale most do not capture value, measuring well stops being tidiness and becomes the difference between activity and sustained value. (These are global evidences from other contexts: they illustrate mechanisms and give scale, they do not promise results to a local SME, whose anchor remains regional evidence.)
The analytical ladder
Parallel to the technology adoption ladder, data also climbs step by step. First descriptive —what happened—; then diagnostic —why—; higher up predictive —what will happen—; at the top prescriptive —what to do. Many organizations believe they are predictive because they bought AI, but operate on poor descriptive: each area calculates “sales” differently and no one audits definitions. That is metric anarchy —two dashboards, two truths— and it precedes any model.
The verdict states which analytical step the organization is really on. Bridge rule: do not promise predictive or prescriptive without reliable descriptive and agreement on what each number means.
Metric governance (before the dashboard)
Before choosing chart colors, you need governance of meaning:
| Practice | What it avoids |
|---|---|
| KPI owner (from the business, not only systems) | Indicators no one defends when they change |
| Semantic layer —one definition of “net revenue”, for example | Two dashboards showing different figures for the same question |
| Few reliable indicators (twelve or fewer on the executive view) | Dashboard fatigue where no one knows what to look at |
| Review ritual weekly or quarterly | Zombie KPIs still on screen though they no longer drive decisions |
| Question in the dashboard header | Pretty charts with no associated decision |
When the client asks for “a dashboard”, the first question in link 1 is: what Monday decision changes if this number goes up or down? If there is no answer, it is not yet time to build dashboards.
The five axes
A single number —“we saved X hours”— rarely tells the whole story. The dashboard is organized into five axes; not all apply to every client, but it helps to know them:
| Axis | What it measures | Example indicators |
|---|---|---|
| 1. Economic value | Hours freed, errors avoided, attributable revenue | Hours/week on task X; reprocessing cost |
| 2. Decision value | Quality, speed and traceability of the decision | Time to decision; % decisions with evidence |
| 3. Human value | Well-being, perceived autonomy, team learning | Brief survey; post-intervention interviews |
| 4. Organizational value | Data maturity, governance, absorption capacity | IMIA delta by dimension; policies adopted |
| 5. Social value | Gaps closed, service to the citizen, equity | Coverage; appeal rate; bias audit |
In ESG engagements, the social axis crosses honestly with environmental: infrastructure footprint, digital inclusion, closing the loop in smart projects. Greenwashing destroys trust faster than not measuring.
Not all axes apply to all clients. Two or three are prioritized at the start according to the sociotechnical diagnosis —in the public sector, the social axis is usually mandatory; in the SME, often economy and decision suffice.
Protocol
- Baseline in the first link, alongside the IMIA profile when appropriate. Without a baseline, any improvement is narrative.
- Objectives per axis agreed before building —avoids retroactive vanity metrics.
- Review at closure —and in sociotechnical hygiene if there is evolutionary accompaniment.
“We did not measure that before” is valid information: it sometimes prevents promising a percentage and forces designing a pilot measurement.
Honesty
Published figures only with verified backing or labeled as internal estimates. A pretty dashboard without a baseline is decoration —and decoration feeds the confusion between activity and outcome that the thesis of the bridge combats. IMIA measures the organization’s configuration —whether it is ready—; this dashboard measures intervention outcome across the five axes.
On Distribuidora Norte, second pass (SMEs), an honest dashboard might prioritize only two axes at the start: decision value —does the supervisor decide sooner with the unified dashboard?— and human value —does the driver feel the system helps or surveils them?—. Promising peso savings without a baseline would have been narrative; time-to-decision on complaints was measured and the delivery team was interviewed at closure. The organizational axis (IMIA delta on D2 and D6) closes the circle with the previous chapter.
See also: Concepts of our own (value metric) · IMIA — the maturity instrument · The bridge applied to SMEs