Open any GST data sales page in India and the filter list looks roughly the same: industry, turnover, company size, HSN or NIC code, state, registration date. It reads like a database schema. It is closer to a wish list — three of those six are not fields in the GST registration record at all, and of the two that are, one behaves nothing like buyers assume.
This is not an accusation of fraud. Providers genuinely have industry and turnover columns in their files. The question is where those columns came from, because a filter derived by a keyword rule over a trade name and a filter read from a government register are the same column heading attached to very different reliability.
What the registration record actually contains
The GST portal's Search Taxpayer result — the pre-login one, available to anyone without an account — returns a fixed field list:
GSTIN/UIN · legal name · trade name · effective date of registration · constitution of business · GSTIN status · taxpayer type · administrative office · other office · principal place of business · effective date of cancellation · nature of business activities · dealing in goods and services · details of return filed.
That is the universe. Everything a provider offers beyond those fields was either derived from them, joined in from another register, or inferred. All three are legitimate techniques and all three have error rates. None of them are the GST register saying so.
The provenance question
For every filter you plan to actually use, ask: "which field in which register does this come from, and if it is derived, what is the rule?" A provider running a real pipeline answers in one sentence. The answer also tells you the error mode you will be living with.
HSN and SAC codes: real, but only the top five
HSN codes are the strongest of the industry-filter claims, because they genuinely are collected at registration. The GST REG-01 application has a Goods and Services tab, and the portal's own guidance is explicit about the cap: "You can add maximum 5 goods and 5 services. In case, you have more than 5 goods or services, you must add the top 5 goods or services you are dealing with."
Those declarations surface publicly under "dealing in goods and services". So HSN filtering on registration data is possible. Four caveats decide whether it is useful to you:
- It is capped at five. A distributor carrying two hundred SKUs is represented by five codes. The long tail is invisible, so filtering for a niche HSN systematically misses businesses that genuinely trade in it.
- It is self-declared at application time, frequently by the accountant filing the form rather than by anyone in the business. Broad, safe, chapter-level codes are common because they are the least likely to cause a query.
- It describes intent, not activity. The codes are entered before the business has traded. A company that pivots in month four does not generate a new registration record.
- It can be amended without officer approval — HSN sits in the non-core amendment set. That cuts both ways: it can be corrected, and nothing obliges anyone to correct it.
The codes that reflect what a business actually sells are the ones reported in Table 12 of GSTR-1, which is return data, not registration data, and is not public. Any provider offering HSN filtering that tracks real trading activity is describing something they cannot have obtained from the register.
Turnover: not in the public record
Turnover appears as a filter on several GST database sales pages, sometimes as banded "revenue scale" segments. Aggregate turnover is declared in the registration application, but it is not among the fields the public search returns, and it is not published anywhere else as a per-GSTIN figure.
Where a turnover column exists in a purchased file, it came from one of:
| Likely source | What it actually is | Reliability |
|---|---|---|
| MCA financial filings | Filed annual accounts, companies and LLPs only | Good, but covers a small share of GST registrants and lags by a year or more |
| Udyam registration | Self-declared MSME classification | Self-reported, and only for businesses that registered on Udyam |
| Inference from taxpayer type | Composition scheme implies below the composition threshold | A ceiling, not a figure |
| Modelled estimate | A guess from employee count, sector or address | Presented as data, indistinguishable from data in a spreadsheet |
The composition-scheme inference is the only one derived from the GST record itself, and it tells you a business is under a threshold rather than what it earns. If turnover banding drives your qualification, verify which of these rows you bought before it drives anything.
NIC codes belong to a different register
NIC codes turn up in GST database listings, and they are a real classification system — they are just not collected by GST registration. NIC is used by Udyam and appears in MCA records. A GST file carrying NIC codes has been joined to one of those registers, which means the NIC coverage is bounded by the overlap, not by the GST universe. For proprietorships and partnerships that never appear in MCA, that overlap is thin. The boundary between the two registers is worked through in GST data vs MCA company data.
The two filters that do work
Strip out what is not there and two genuinely useful segmentation fields remain, both often overlooked because they sound less precise than an HSN code.
Nature of business activities. A small controlled vocabulary — retail business, wholesale business, factory or manufacturing, supplier of services, office or sale office, warehouse or depot, and similar — attached to the principal place of business. It is coarse, which is exactly why it is robust: there is little room to misdeclare "factory / manufacturing", and the categories map cleanly onto how most sales teams actually segment. If you sell to manufacturers, this field does more work than any HSN filter.
Details of return filed. The public record shows filing history, which makes filing frequency derivable rather than claimed. This is a genuinely strong signal and it is underused. A monthly filer is transacting at a scale that took it out of the quarterly QRMP option; a registrant with no returns filed since registration may never have commenced business at all. Filtering on filing behaviour separates trading entities from dormant GSTINs using nothing but public data — and it is a far better proxy for "is this worth a call" than any modelled turnover band.
Constitution of business is a third useful axis — proprietorship, partnership, private limited, LLP — and it drives document requirements as much as targeting, which is covered in current account KYC by entity type.
Why the 2025 rate slabs are not an industry filter
Since the GST Council's rationalisation took effect on 22 September 2025, the rate structure has centred on 5% and 18%, with the 12% and 28% slabs withdrawn and a 40% rate applied to luxury and sin goods. This occasionally gets repackaged as a targeting attribute — "filter businesses in the 18% slab".
It does not work as one. Rates attach to goods and services, not to registrants. A business selling across two slabs has no slab. The rate change is worth understanding because it altered what many businesses charge and therefore what their accountants had to reconfigure, which is a genuine conversation opener for software and advisory sellers. It is not a segmentation field.
What to ask before buying an industry-filtered file
- Which register did the industry column come from — GST HSN declaration, Udyam, MCA, or a derivation rule?
- If derived, what is the rule, and what is its measured accuracy on a sample you choose?
- What percentage of rows have the industry field populated at all? Filters are worthless on sparse columns, and fill rate is rarely on the sales page.
- Is the HSN the registration top-five, or something claimed to reflect current trading?
- Can you filter on nature of business activities and filing status? If the provider does not expose the two fields that are most reliable, ask why.
Sample-testing methodology for all of this is in the B2B data quality checklist, and the separate question of how old the records are is covered in what "daily" and "fresh" actually mean.
Providers in this space approach it differently. FinScreener (built by the team that publishes this site — see our disclosure) works from new GST registrations with geographic and date filtering. EMarket Zone and IndiaDatabase sell segmented GST databases advertising turnover and industry bands, Digitap approaches it from enrichment and KYC, and Probe42 and Tofler offer genuine industry classification because they work from MCA filings where it exists. The broader comparison is in where to buy B2B leads in India.
Common questions
Can I get a list of GST businesses by industry at all? Yes, at the granularity the register supports — nature of business activities, and chapter-level HSN from the top-five declaration. What you cannot get from GST data alone is precise vertical classification of the kind MCA-based or Udyam-based sources provide for the businesses they cover.
Why do two providers give different industry tags for the same GSTIN? Because they derived them differently — one from the HSN declaration, another from a keyword rule over the trade name, a third from an MCA join. All three can be internally consistent and mutually contradictory.
Is filing frequency really a better filter than turnover? For deciding whether a business is worth contacting, usually yes. It is observed rather than modelled, it updates continuously, and it reflects current activity. A modelled turnover band reflects an estimate made once.
Does a business with no returns filed mean the GSTIN is fake? No. It commonly means the business registered ahead of commencing operations, or registered to meet a customer or marketplace requirement. It is a strong signal to deprioritise, not evidence of anything improper.