The Dashboard Problem — When Measurement Replaces Delivery

Pritiraj BrahmaFounding Partner, Pritiraj & Partners · 23 July 2026

India has more real-time government dashboards than any comparable state. Pendency, learning outcomes and delivery gaps have not moved with them. Visibility is not accountability.

ProvenancePublished 23 July 2026; revised 31 August 2026 to add two figures and a provenance line. The Comptroller and Auditor General is the source for the 61 per cent fund utilisation by urban local bodies, the 56 per cent property tax collection efficiency and the skilling programme beneficiary survey that returned 171 usable responses; these are three separate audits, grouped in Figure 2 by basis of measurement and not netted against one another. Judicial pendency past five crore cases is from the National Judicial Data Grid, a live series. Learning outcome measurement is PARAKH and ASER. The classification of metrics in Figure 1 into inputs, outputs and outcomes is this Review’s own reading of what scheme dashboards publish, not a published taxonomy. Data vintage note: dashboard figures are current by construction and audit findings are not, so the two are never strictly contemporaneous — which is part of why they diverge.

India solved the measurement problem and assumed it had solved the management problem. It had not. A number that nobody is required to act on is decoration.

Over the last decade India has built genuinely world-class public monitoring infrastructure. The National Judicial Data Grid publishes case pendency down to the individual court, updated daily. PM Gati Shakti maps infrastructure projects across ministries. PRAGATI reviews stalled projects at the highest level of government. Scheme dashboards report beneficiary counts, disbursement and physical progress in near real time. State chief ministers' offices run their own command centres.

This is a real achievement and other developing states study it. It has also, in a specific and important way, failed to convert.

Four findings anchor this analysis:

1. Visibility rose and the underlying indicators did not follow. Judicial pendency is more precisely known than at any point in Indian history and has continued to rise past five crore cases. Learning outcomes are measured in detail by PARAKH and ASER and remain far below grade level. Scheme dashboards report high disbursement while independent audits report weak outcomes. Measurement improved; the measured thing did not.
2. The metrics that get published are the ones that are easy to count. Funds released, beneficiaries enrolled, certificates issued, structures sanctioned. Each is an input or an output. The outcome — did the household's income rise, did the child learn to read, did the case get decided — is harder to count and is largely absent from the dashboard layer.
3. Self-reported data has no independent check. Almost all dashboard data is entered by the implementing agency being assessed. The CAG's finding that a skilling programme's beneficiary survey returned 171 usable responses, padded with duplicates and identical photographs, is the extreme case of a general condition: nobody verifies the number against the ground.
4. Where verification is independent, the picture changes. The CAG's finding of 61 per cent fund utilisation by urban local bodies, and of 56 per cent property tax collection efficiency, come from audit rather than self-report. Independent measurement consistently produces harsher and more useful numbers than dashboards do.

Why dashboards drift toward theatre

Three mechanisms, each rational at the level of the individual official.

What is displayed is what is optimised. If a district's standing depends on the percentage of funds disbursed, funds get disbursed — on time, in full, and not necessarily well. The dashboard does not cause bad delivery; it redirects effort toward whichever proxy it has chosen to display.

Entry is a task, not a consequence. The officer entering the data is usually the officer being judged by it, and entry competes for time with the delivery itself. A field officer spending a fifth of the week on reporting is spending a fifth less on the work being reported.

There is no consequence loop. This is the decisive failure. A dashboard is a control instrument only if a red cell triggers something — a review, a resource shift, a defined escalation with a deadline. Most Indian dashboards display without triggering. They are windows, not thermostats.

What independent verification found when it looked

The clearest available test of the gap between a dashboard and the ground is the Jal Jeevan Mission, because the government commissioned the check itself. The 2024 Functionality Assessment of Household Tap Connections covered 19,812 villages already certified as Har Ghar Jal, across 761 districts. It found that 98 per cent of households had a tap connection — and that only about three in four were receiving water that met the Mission's own standards of quantity, quality and regularity.

The component figures differ between published accounts of the assessment, and the difference is worth stating rather than smoothing over. One account reports 86.5 per cent of households with working connections, 83.6 per cent receiving water on schedule and 80.2 per cent receiving the prescribed quantity. Another reports 83 per cent receiving water at least once in the preceding seven days, 80 per cent meeting the quantity norm, and 76 per cent meeting basic safety standards on tests for E. coli, faecal coliform and pH. The headline conclusion is consistent across both: near-universal connection, roughly three-quarters actual service. Gujarat and Tripura, both reporting complete tap coverage, recorded functionality below half.

This is the dashboard problem in a single dataset. The reported indicator — a connection installed — is real, auditable and was delivered at extraordinary scale. It is simply not the same variable as water arriving in a house at a usable quantity and quality. A programme reporting 98 per cent and delivering 75 per cent is not falsifying anything. It is publishing precisely what it was asked to count.

Two things follow, and the second is more encouraging than this Review's usual finding. First, the corrective information came from a commissioned independent assessment rather than from the reporting system, which is the general rule: the harsher and more useful number almost always arrives from outside the implementing chain. Second, the response to that assessment has been to build the consequence loop — under the mission's next phase a gram panchayat cannot certify Har Ghar Jal until operation-and-maintenance arrangements and user charges are confirmed operational by the state. That is a red cell wired to a trigger, which is exactly what a dashboard needs and almost never has.

The same pattern appears in air quality spending, where the binding number surfaced through a parliamentary reply rather than a portal. The Union environment minister told Parliament that Delhi had utilised ₹14.1 crore of ₹99.77 crore released to it under the National Clean Air Programme — about 14 per cent — while Ghaziabad and Meerut had spent over 80 per cent of theirs. An RTI reply on the same subject reported a different denominator, ₹81.36 crore allocated with 17 per cent utilised. The figures are not reconciled publicly, which is its own comment: a programme with a national dashboard produced two official utilisation numbers for the same city.

The counter-case, honestly stated

The strongest defence of the dashboard decade is that transparency has value independent of immediate effect. The NJDG made judicial delay a public fact rather than an anecdote, and it is now impossible to argue about pendency without data. Direct benefit transfer dashboards made leakage visible and correctable in a way paper never allowed. Aadhaar-authenticated portability of food entitlements works precisely because the transaction is recorded centrally in real time.

That is true, and it establishes what dashboards are good for: they are superb at exposing a problem and poor at resolving one. The error is not building them. It is treating construction of the instrument as completion of the reform.

What we would do

1. Attach a consequence to every published indicator. For each metric, state in advance what happens when it breaches a threshold, who owns the response, and within how many days. An indicator with no defined trigger should be removed from the dashboard, not celebrated on it.
2. Verify a sample independently. Field verification of a small random sample — one or two per cent — by an agency that does not report to the implementer. This is inexpensive and it changes reporting behaviour immediately, because the possibility of being checked is what makes self-report honest.
3. Publish one outcome per scheme, even if it is late and imperfect. Employment retained at twelve months. Learning level at end of year. Case disposal rate rather than pendency count. One honest outcome measure is worth twenty input measures.
4. Reduce reporting burden at the field level in proportion to what you add centrally. Every new field a state adds to a reporting format is time taken from delivery. Publish the total reporting load on a field officer and treat reducing it as an objective.
5. Protect the officer who reports a bad number. A system that punishes accurate reporting will receive inaccurate reports and then govern from them. This is the cheapest reform on this list and the hardest to sustain.

India can now see its own government in more detail than almost any comparable state. Seeing is not governing. The distance between the dashboard and the household is the same distance this Review keeps measuring — and no amount of resolution on the screen shortens it.