City is the most requested filter in Indian B2B data and the only commonly advertised geographic filter that is not a field in the GST register. The state is structural — it is the first two digits of the GSTIN. The city is not. It is parsed out of the principal place of business address, a free-text field, by whoever built the file.
That single fact explains most of what goes wrong with metro lists: why two providers give different Bengaluru counts, why a "Mumbai database" contains Thane and Navi Mumbai in one file and not another, and why a Delhi list can quietly mean three different things.
There is no city field
The public Search Taxpayer record exposes the principal place of business as an address, not as structured components. To produce a city column, a provider must do one of three things:
| Method | How it works | Failure mode |
|---|---|---|
| String matching on the address | Look for "Mumbai", "Bengaluru", "Bangalore" in the text | Misses misspellings, alternate names and suburb-only addresses; over-matches street names containing the city |
| Pincode range mapping | Map the PIN to a city via a postal lookup | Clean and reproducible, but postal boundaries are not commercial ones — see below |
| District field mapping | Use the district and treat it as the city | Districts are larger than cities and share names with them inconsistently |
None is wrong. They produce measurably different lists from the same underlying registrations. A provider that cannot tell you which one they used has not thought about it, which is itself the answer.
The two-name test
Ask for counts on both spellings — Bengaluru and Bangalore, Mumbai and Bombay, Gurugram and Gurgaon, Vadodara and Baroda. If a provider's city matching is string-based and naive, one spelling will return a fraction of the other. Registration addresses were typed by thousands of different people over eight years and contain every variant.
Delhi NCR is three state codes
The clearest case where a city request and the register disagree. NCR spans Delhi (state code 07), Haryana (06) and Uttar Pradesh (09). Gurugram registrations carry a Haryana GSTIN; Noida and Ghaziabad carry Uttar Pradesh; only Delhi proper carries 07.
So "Delhi NCR leads" can legitimately mean any of:
- Code 07 only — Delhi proper, excluding Gurugram and Noida entirely
- 07 + the NCR districts of 06 and 09 — the commercially useful definition, and the hardest to build
- 07 + all of Haryana and all of Uttar Pradesh — technically defensible, and it will fill your list with Hisar and Gorakhpur
A team that buys the third expecting the second finds out during dialling. Ask which state codes are in the file before you buy it, and ask for the count broken down by code. The same problem in milder form affects any metro that has grown across a state line.
Pincode is a clustering tool, not an address
Pincode filtering is reproducible, which makes it the best of the three methods — but its precision is easy to overestimate. A pincode tells you where a registration was filed, and registration addresses are systematically not where businesses operate. That distortion, along with the virtual-office clustering problem, is worked through in state-wise GST registration data and applies with more force at city level, because city-level lists are the ones that get handed to field teams.
The practical consequence: use pincode to understand density and to allocate between branches, not to promise a rep that a business is at a location. For branch-level assignment specifically, the catchment logic is in branch catchment lead routing.
Metro versus tier-2, and why the mix matters
Metro lists are the ones vendors build first, so they are the most saturated. Every provider sells a Mumbai, Delhi, Bengaluru, Hyderabad, Chennai and Pune database, which means the businesses in them have been called by everyone. Registration volume in tier-2 and tier-3 cities is where the same list is likely to be commercially fresher, simply because fewer teams have worked it.
This cuts against the instinct to buy the biggest metro file available. If your product travels — software, insurance, banking, compliance services — the marginal newly registered business in a tier-2 city has usually received far fewer calls than one in a metro business district, and the registration data covers both equally well because the register does not care about city size.
Building a city list that survives contact
- Start from the state code, then narrow. It is the only structurally guaranteed geographic field.
- Choose pincode ranges over name matching where the provider supports it, and get the range list in writing so the boundary is reproducible next month.
- Ask for the NCR or metro-boundary definition explicitly if you are buying one of the cross-border metros.
- Run the two-name test on any string-matched file.
- Group by exact address and sort by count before distributing — the virtual-office clusters are concentrated in exactly the central pincodes city lists target.
- Layer the non-geographic filters early. Constitution of business, nature of business activity and filing status narrow a list far more usefully than tightening a radius. Which of those filters are real is covered in filtering GST leads by industry.
- Check the age of the records, because a city file assembled once and resold for a year is a different product from a dated feed — see what "daily" and "fresh" actually mean.
Why city counts disagree between providers
Beyond the parsing method, the same methodological differences that affect state counts affect city counts more sharply: deduplication key, whether cancelled registrations are included, whether TDS deductors and ISD registrations are counted as businesses, and whether additional places of business generate extra rows. At city level these differences compound, because the denominators are smaller.
Treat any city count as uninterpretable without the methodology attached. Two providers quoting "48,000 businesses in Pune" may be counting different things by a factor of two.
City and pincode filtering is offered by most GST-based providers. FinScreener — built by the team that publishes this site, see our disclosure — filters new registrations by state, city and pincode. GSTDataProvider and EMarket Zone sell prebuilt city databases, and directory-derived sources such as JustDial and IndiaMART have city coverage with a category bias rather than a registration bias. Compare them on the specific cities you sell into; general accuracy claims will not tell you which is stronger in Coimbatore.
Common questions
Can I get GST data filtered by pincode? Yes — pincode is derivable from the registration address and most providers support it. What it gives you is where the registration was filed, which is a reliable clustering signal and an unreliable statement about physical presence.
Is a city database better than a state database? Only if the city boundary was drawn in a way you can inspect. A state file you filter yourself by pincode is often more trustworthy than a city file whose boundary rule is undisclosed.
Why does a Bengaluru file contain businesses in Karnataka towns I have never heard of? Usually district-field mapping — Bengaluru Urban and Bengaluru Rural are districts, and Bengaluru Rural extends well beyond the city. Ask whether the filter was district-based.
Does the register update the address when a business moves? Only when the business files an amendment, which many do not do promptly. Address-derived fields age quietly, and nothing in the record flags that they have.