Here's a mistake I watch founders make on repeat. They hear "trademark your brand," open the list of 45 classes, panic slightly, and file in six or seven of them "to be safe." (Our trademark class finder cuts that list down to the classes your business actually touches.) That's over ₹50,000 in government fees alone — for protection they mostly don't need, on businesses they don't run, in classes where they'll never trade.

Caution is good. Indiscriminate caution is just expensive.

The service-class map

Of the 45 NICE classes, eleven cover services — classes 35 through 45. The ones that matter to most startups, in plain terms:

Class 35 — advertising, business management, office functions. The "we run businesses" class. Agencies, consultancies, marketplaces, and anyone whose service is business itself lives here.

Class 36 — finance and insurance. Payments, lending, wealth management, insurance broking. If money moves through your product, start here.

Class 38 — telecommunications. Messaging, broadcasting, data transmission — the pipes, not what's in them.

Class 41 — education, training, entertainment. Edtech, coaching, content platforms, events.

Class 42 — scientific and technological services, including software. Software development, SaaS, cloud services, design and development. This is the default home of most tech startups, whether they realise it or not.

Class 45 — legal services, security services, and personal services — which, in classification logic, includes social networking and matchmaking. Yes, really.

The remaining five — 37 (construction and repair), 39 (transport and logistics), 40 (treatment of materials), 43 (food services), 44 (medical and beauty services) — matter enormously to the startups in those sectors and not at all to anyone else.

The picking framework: two questions

Forget "what might we do someday." Ask two questions instead.

One: where do customers pay you today? Not where the website says you operate — where the invoices go. That's your core class, and it's non-negotiable.

Two: where will they plausibly pay you in the next two to three years? The adjacent service you're already building toward. The expansion your roadmap actually commits to. File for the answers.

Everything beyond that is speculation you're paying ₹9,000 per class to insure. (₹4,500 if you're a recognised startup or small enterprise — and most early-stage companies qualify. Check before you pay the higher slab.)

The usual combos, by startup type

SaaS: 42 for the software, 35 for the business built around it. Two classes, done. (The Class 9 vs Class 42 question trips up SaaS founders constantly — read that before you file.)

Edtech: 41 for the teaching. Add 42 if the app platform is itself the product rather than just the delivery van.

Fintech: 36 for the money movement. Add 42 for the platform underneath it.

Agency or consultancy: 35. That's usually the entire story — I've talked founders out of 42 here more times than I can count. Building client websites doesn't make you a software company in the classification's eyes; your service is the business advice, not the code.

Notice the pattern: nobody needs all eleven. Most startups need two, maybe three.

One application, multiple classes

A procedural fact that changes the economics: you don't file eleven applications. A single application on Form TM-A can designate multiple classes — the government fee is per class, not per form. So the decision was never about paperwork. It's about money, and about honesty.

The honesty part

You file on the basis of use or a genuine intent to use. Filing in classes where you have no real plans isn't "being thorough." It's clutter on the register — and clutter can be challenged. A competitor can seek cancellation of a registration for non-use once five years have passed. Marks you don't use are marks you can't defend.

Examiners read specifications, too. "Software as a service" sails through in 42. Grand phrases like "all technology-related services" draw an objection for vagueness, and objections cost months. Write what you actually do, in the vocabulary the class headings already use. The classification was written by people who've seen your business model before.

When to add classes later

Here's the part that should calm you down: filing narrow now doesn't lock you out later. When the business genuinely expands into a new class — the SaaS company launches a lending product, the agency starts a training academy — you file a fresh application for the new class. Your original registration keeps its original priority date; the new one gets its own. That's normal portfolio growth, not a mistake you made at the start.

What you can't do is backdate. If a competitor files in 36 while you're still "thinking about fintech someday," their date beats your intention. So revisit the class map once a year — fifteen minutes, same two questions. The register rewards founders who file when the plan becomes real, not before and not after.

Do the cost math before you file

Two classes at the standard rate: ₹18,000. Six classes: ₹54,000. The ₹36,000 difference doesn't buy you extra safety — it buys you renewal fees on classes you'll never enforce, watch notices you'll never act on, and a portfolio that looks impressive and means little. Spend the difference on the product. The brand is only as valuable as the business behind it.

Picking classes is a fifteen-minute conversation with someone who's mapped a few hundred startups onto the classification — our class search guide is the written version. Our team will put your business in the right classes — and talk you out of the ones you don't need. That's the more valuable half of the advice.