Prepaid Was Never a Meter Mode: What Uttar Pradesh's Rollback Teaches the Rest of Us

Executive Summary

In May 2026, Uttar Pradesh instructed its distribution companies to move every smart prepaid meter to postpaid billing. Roughly 83 lakh consumers out of about 86.5 lakh smart meter connections were switched through the central backend, reportedly after sustained complaints about the consumer experience. Not one meter was uninstalled. The AMI stack kept recording interval data. What failed was everything wrapped around the meter: recharge channels, balance visibility, grievance response, and the sub-divisional capacity to absorb a new kind of customer conversation.

I have spent the last several years on the other side of this problem, in North Bihar, where prepaid is the default rather than the exception. The lesson I take from UP is not that prepaid is wrong. It is that prepaid is a behaviour contract between a utility and a household, and the meter is merely the instrument that enforces it. Sign the contract before you have the service capacity to honour it, and the rollback becomes a question of when, not whether.

This post sets out the five gates I now insist on before a DISCOM converts a feeder to live prepaid.

The number leaders are tracking is the wrong one

India crossed 7.24 crore smart meters — consumer, DT and feeder — as of 30 June 2026, of which 5.73 crore sit under RDSS against a sanctioned 20.33 crore and a March 2028 sunset. Every review meeting I attend opens with that installation curve.

It is the wrong headline metric. An installed meter is a capital event. A sustained prepaid relationship is an operating capability. The two are separated by a layer of work that no procurement contract fully captures: recharge liquidity in a village with patchy connectivity, a balance SMS that arrives before the disconnect and not after, a JE who can explain a load-based fixed charge without escalating, and a billing engine that reconciles a mid-month tariff revision against a token already consumed.

UP demonstrated what happens when the capital event outruns the operating capability at scale. The correction was not incremental. It was a single backend instruction across four DISCOMs and KESCO.

What actually broke — a diagnosis, not a verdict

I want to be careful here. I have no privileged view into UPPCL's internal data, and public reporting on the switch is thin and partly advertorial. So I will frame this as the failure modes I have personally seen break prepaid conversions, in the order they usually bite.

The first is recharge friction. Prepaid shifts the payment from a monthly event the consumer plans for to a continuous obligation they must monitor. If the recharge takes four taps and a failed UPI callback, the household does not become disciplined. It becomes angry.

The second is balance opacity. A consumer who cannot see their balance cannot manage it. Every disconnect then reads as arbitrary, and the SDO office absorbs that perception.

The third is tariff translation. Fixed charges, security deposit adjustments and arrear instalments behave differently in prepaid. UP is now reimposing security deposits that were adjusted during the prepaid period, with recovery from July 2026 bills — a reminder that the accounting tail of a mode change outlives the technical one.

The fourth is grievance throughput. UP ran grievance camps at Executive Engineer and SDO offices from 15 May to 30 June 2026. Camps are a signal. They mean the standing channel could not carry the load.

A five-gate prepaid readiness framework

This is the gate review I use. A feeder does not go live on prepaid until all five are green. It is deliberately boring, because the failures are boring.

Gate 5 is the one leaders skip, and it is the one UP needed. A rollback designed in advance is a controlled manoeuvre. A rollback improvised at 83 lakh connections is a political event.

North Bihar's quieter lesson

North Bihar took the slower road. NBPDCL has installed close to 60 lakh smart prepaid meters, with Bihar crossing 90 lakh smart meters statewide and over 1.03 lakh DT meters. The outcome that matters to me is not the count. It is that billing efficiency reached 89.58% in FY26 — roughly ten points above FY20 — alongside collection efficiency of 97.63%.

Those two numbers together are the real prepaid scorecard. Collection efficiency near 98% says the payment behaviour held. Billing efficiency below 90% says the metering-to-billing chain still leaks, and that is where the next tranche of work sits: DT-level energy audit, unmetered load, and the stubborn gap between a meter that reads and a bill that lands.

What made it hold, in my experience, was forward deployment. Engineers sitting in the sub-division rather than the state headquarters, watching the recharge funnel the way a product team watches a conversion funnel, and closing the loop in days rather than quarters. A ₹0.25 per unit rebate helps adoption. It does not substitute for someone in Muzaffarpur who can tell you by Thursday why recharges on one feeder dropped 18% on Tuesday.

What This Means for Leaders

1. Retire the installation curve as your primary KPI. Replace it with a prepaid cohort dashboard: recharge frequency, median balance at recharge, involuntary disconnect rate, and 30-day repeat-recharge rate. Report it monthly at board level.
2. Pilot on cohorts, not pincodes. Convert a small, deliberately mixed set of feeders — urban, rural, agricultural, high-arrear — and hold them for 90 days before scaling. Geographic pilots flatter you; mixed cohorts inform you.
3. Write Gate 5 before Gate 1. Mandate a documented, tested rollback and a balance-reconciliation rule as a precondition for go-live. Make it a contractual deliverable from the AMISP, not an internal note.
4. Fund the service layer explicitly. Carve out a line item for recharge channel density, alert delivery and SDO query capacity. If it is not budgeted, it will be improvised at the counter.
5. Deploy your engineers forward. Put two or three capable people in the sub-division for the first 120 days of every conversion wave, with authority to change configuration. The cost is trivial against a rollback.

Key Takeaways

• UP moved roughly 83 lakh smart prepaid meters to postpaid in May 2026 without removing hardware — an operating-model failure, not a metering one.
• Prepaid is a behaviour contract. The meter enforces it; the service layer earns it.
• India's 7.24 crore installed meters against a 20.33 crore RDSS target means most of this conversion risk is still ahead of us.
• North Bihar's 89.58% billing and 97.63% collection efficiency in FY26 show the model works when the field layer is staffed.
• No feeder should go live on prepaid without a tested rollback and a balance-reconciliation rule.

Written from the field in North Bihar, where the recharge funnel teaches more than the dashboard does.

Sources: India's smart meter population at 7.24 crore — Parliament (T&D India) · UP govt ends prepaid system for smart meters (EMF Radar, citing ET EnergyWorld) · Your UPPCL smart meter is now postpaid (The Tribune — syndicated release) · How smart meters are powering a new era for North Bihar (Indian Masterminds)