How we count

Everything on this site is derived from one government source by one arithmetic step. This page states the definition of every number, and the limits of each, in enough detail that you can check us.

The source

The registered-vendor list published by the PM Surya Ghar: Muft Bijli Yojana portal at pmsuryaghar.gov.in, operated by the Ministry of New and Renewable Energy, Government of India. We read it through the portal's own public interface, at the rate a person would, and record what it returns as a dated snapshot. We do not scrape anything the portal does not publish to any visitor, and we never request vendor documents.

The one arithmetic step

For each district the portal publishes two things about each empanelled installer: that they are empanelled there, and how many installations they have recorded — nationally, in the state, and in that district. We count how many installers listed for a district have a district installation count of zero.

In Bengaluru Urban, Karnataka, that is 266 of 573 installers, or 46.4%, as recorded on 7 September 2026. There is no model here, no estimate and no weighting. It is a count.

Why some names are withheld

336 of the 573 installers on the Bengaluru Urban register are listed here without a name. That is deliberate and it is not a data gap.

Many registered solar vendors in India are sole proprietors. When a vendor is a sole proprietor, the business name is a natural person's name, and publishing it — attached to a number a reader might read as unflattering — is processing personal data. India's Digital Personal Data Protection Act 2023 requires a lawful basis for that, and reproducing government data under a reuse licence is not one: a reuse licence says what we may republish, not what personal data we may process. Those are different regimes and we do not treat one as the other.

So we publish a name only where we can establish that the vendor is an incorporated entity — today, where the registered name carries a legally reserved corporate form such as Private Limited, Limited or LLP, which an unincorporated proprietor cannot lawfully use. Everything else is withheld. Industry words prove nothing: "Solar", "Energy" and "Power" are exactly the words a proprietor appends to their own given name.

Withheld rows keep their rank and their numbers. We do not drop them, hide them, or move them to the bottom. If eleven installers rank above the first one we can name, you can see all eleven and their counts. Hiding them would make the register look shorter and the named installers look better placed than they are.

This is reversible in the right direction. A withheld name that we later establish belongs to a company gets published. A published name that turns out to belong to a person cannot be unsent. We take the delay.

What we do not do

• We do not rate installers. No stars, no score, no "verified" badge, no ranking by quality. We have no basis for any of those and we are not going to invent one.
• We do not publish or hold installer contact details. The portal shows them; we do not store them.
• We do not assert that any installer has been de-empanelled on the strength of a name or contact match. The government publishes a deactivation list, and matching a name against it is a candidate, not a fact. Nothing of that kind is published without individual human verification and a link to the official record.
• We do not estimate. A district we have not read has no page. We never extrapolate one district's figures to another.

Known limits, stated before you find them

• We inherit the portal's errors. If the register records an installation against the wrong district, so do we. We are a mirror with arithmetic; we are not an auditor and we cannot independently verify an installation happened.
• A zero can be stale. An installer active in a district since the last snapshot shows a zero until the next one. That is why the date is on every page, and why corrections are free and require no account.
• Ratings on the portal are not used. The portal carries consumer ratings with a median of 4.9 and no meaningful spread. We record them and publish none of them, because a number that does not discriminate is decoration.
• Coverage is partial and says so. Each page states its own scope. There is no national total on this site until a national snapshot exists to support one.

For machines

Every data page has a JSON twin at the same URL plus index.json, with the same numbers, their definitions, the scope and the date. There is a /llms.txt summary of the whole site and a /llms-full.txt containing the full factual payload in one file. Crawling and quoting is welcome. Please carry the date with the number — these figures change every snapshot, and a figure quoted without its date will be wrong within a month.