Why cluster size is the wrong first question
First-time users of CorpIntel's address-cluster surface often react to a large cluster size — say, 400+ companies at one registered office — as a signal of suspicion. It is not. It's a prompt to investigate, which is a different thing.
Three very different patterns can produce a 400-company cluster. Two of them are entirely legitimate and fully expected. One is a signal worth taking seriously. The entire value of the cluster view on CorpIntel is in being able to distinguish them, not in reacting to the count itself.
Pattern 1 — Group headquarters
Large Indian business groups often register many of their operating subsidiaries at the same physical office — the group HQ. If a group has 60 operating subsidiaries (not unusual for a multi-sector conglomerate), they can all share a single registered address. Add the holding-company structure and various special-purpose vehicles, and the cluster size at a major group HQ can easily reach 100+ companies.
Signatures of this pattern: one or a small number of dominant directors (often family members with shared surnames), consistent industry focus aligned with the group's sectors, high-quality filings, ages spread across decades, and a clear paper-trail connecting the entities.
What you should not do: treat this as evidence of anything other than a legitimate group structure. The address cluster size here is simply the corporate equivalent of an office directory.
Pattern 2 — Virtual-office / company-secretary service
A large category of Indian companies use virtual-office services or company-secretary firms as their registered address — a legitimate convenience for small companies, solo founders, or companies with distributed operations that need a stable compliance address.
A popular company-secretary service in a Tier-1 city can have 500+ client companies registered at its office address. These clients are completely unrelated to each other — different promoters, different industries, different company ages, different activity levels. You'll see a mix of active companies, recently struck-off companies, dormant companies, and occasional Section 8 non-profits.
Signatures of this pattern: high variance in promoters, directors, and industries; many different surnames in the directors field across the cluster; a mix of company types (PTC, OPC, LLP); a spread of ages; and — critically — the directors do not overlap between companies except in very small numbers (typically the CS firm's partners or staff directors on a small number of entities).
The important thing to note: the cluster size itself is also not a signal here. The cluster size is a proxy for how popular this particular CS service is in the local market. What you want to look at is whether the specific company you care about matches the overall pattern of the cluster (in which case it's likely another legitimate client) or whether it stands out in a specific way.
Pattern 3 — Shell network
The third pattern is the one worth investigating. A shell-network pattern at an address cluster has a very specific signature.
First, high director overlap across supposedly unrelated companies. You'll see a handful of DINs appearing on 20+ of the cluster's companies, which are across ostensibly different industries. This is unlike pattern 2, where director overlap is minimal.
Second, the companies are frequently recent (last 2-3 years) and carry minimal paid-up capital. They rarely file their annual returns on time, and many are at the stage of being marked 'Under Process of Striking Off'.
Third, the industries are a mix that does not match any coherent business group — a tech company, a real-estate company, a trading company, a consulting company, all bearing the same directors and the same paid-up capital range. That's the signature.
A cluster matching this pattern warrants real investigation. It may be a legitimate (though unusual) serial-entrepreneur network, or it may be a shell-entity network used for money movement, tax arbitrage, or other patterns that sit outside the normal incorporation purpose.
How we classify clusters on CorpIntel
Our platform surfaces the cluster size on every company page and links to the full cluster page for each registered address. On the cluster page, we display the full list of companies at that address, the directors in common, and an inferred type tag — group / service / mixed / suspect — derived from the composition signatures described above.
The inferred type is a signal, not a verdict. It's based on the statistical pattern of directors, industries, and activity levels across the cluster. We mark it clearly as inferred and provide enough supporting context so you can form your own judgment.
Illustrative examples
We deliberately do not name specific addresses in this piece. The point is the analytical framework, not to brand any specific cluster. If you are investigating a specific address, the cluster view on CorpIntel gives you the data; the framework here gives you the lens.
If you find a specific cluster that appears to match the shell-network pattern and you have actionable information, the right place to escalate is the MCA complaints interface or, for suspected financial crime, FIU-IND. We do not mediate complaints — our role is to make the data readable.