A brand architecture framework isn't a design exercise. It's a capital-allocation decision that determines where your buyer's risk sits. There are five layers — masterbrand, sub-brand, endorsed, independent, invisible — and most Indian groups end up on layer three because it feels safe, not because it's correct.
I've sat through the meeting where this gets decided. It usually runs about ninety minutes, involves a slide with three logo lock-ups, and ends with someone senior saying "let's keep the parent name small in the corner for now." That corner logo then costs the business two years of muddled positioning and a media plan that has to fund two stories instead of one. Nobody ever writes that cost down anywhere.
The reason it goes wrong is that architecture gets treated as a branding question when it is really a question about accountability. Who does the customer blame when the product fails? Whoever they blame is the brand that has to be on the front of the box. Everything else is decoration.
Why most brand architecture decisions get made for the wrong reason
Three bad inputs dominate the room. The first is the org chart — if the businesses report separately, someone argues the brands should look separate. The second is history — an acquired brand has "equity we can't throw away," a claim almost never tested with actual data. The third is politics, which is the honest name for the first two.
None of those inputs is about the buyer. And the buyer's world has become measurably harder to navigate. McKinsey's B2B Pulse survey of nearly 4,000 decision makers across 13 countries found buyers now use an average of ten interaction channels before purchase, up from five in 2016. Every additional brand you put in front of that buyer has to earn its own recall across ten touchpoints. That's not a design cost. That's a media cost, repeated quarterly, forever.
Simplicity has a measurable price attached to it. Siegel+Gale's 2018 World's Simplest Brands study, covering more than 15,000 respondents across nine countries, found 55% of people would pay more for a simpler experience, 64% were more likely to recommend a brand that delivered one, and companies failing to deliver simplicity left an estimated $98 billion on the table. The same study reported that a stock portfolio of the world's simplest brands had outperformed the average of the major indexes by 679% since 2009.
I don't read that as proof that fewer brands is always better. I read it as proof that every brand you add has to pay rent.
What does a brand architecture framework actually decide?
A brand architecture framework decides three things, and only three. Which name the customer buys. Which name carries the warranty in their head. And which name gets the marketing budget when the year gets tight.
That third one is the tell. If you can't say, without hesitation, which brand gets funded first in a bad quarter, you don't have an architecture — you have a logo family. I've watched groups run four brands off one budget line and then act surprised when none of them registered. The question of how much a B2B company should spend on brand is unanswerable until the architecture question is settled, because the denominator keeps changing.
The stakes at group level in India are not small. Brand Finance's India 100 2025 ranking put Tata Group's brand value at USD 31.6 billion, up 10%, with the combined value of the top 100 Indian brands at USD 236.5 billion. Adani Group was the fastest riser at 82%. On a broader methodology, Kantar BrandZ valued India's top 100 brands at USD 523.5 billion in 2025 — roughly 13% of the country's GDP — with HDFC Bank at USD 44.99 billion and Mahindra up 53% to USD 5.5 billion. These are balance-sheet-scale numbers being shaped by a decision most companies make in one afternoon.
The five layers, and what each one costs
Here is the ladder. Layer one is maximum consolidation, layer five is maximum separation. The middle three are where the arguments happen.
| 1. Masterbrand | One name across everything. Descriptors, not brands: Tata Motors, Tata Steel, Tata Power | One promise, one warranty | Lowest per unit of awareness | Buyers across businesses overlap, and the parent already means something specific
| 2. Sub-brand | Parent plus a named product world with its own character: Birla Opus, Tata Neu | Parent's credibility, product's personality | Roughly 1.5x a masterbrand launch | Entering a category where the parent has permission but no product language
| 3. Endorsed | Independent brand carrying a parent signature: "a Mahindra company" | A name it must learn, plus a reassurance it may ignore | Highest. Two stories, one budget | Rarely. See below
| 4. Independent | House of brands. Parent invisible in market: HUL's portfolio | Only the product brand | Full freight per brand, no transfer | Buyer segments genuinely don't overlap, or parent equity would constrain pricing
| 5. Invisible parent | Parent deliberately absent, sometimes contractually | Nothing of the parent | Full freight, plus a governance overhead | Parent association is a liability, or the business is being groomed for sale
The cost column is the part that usually gets left off the slide. It is also the only column a CFO will remember.
Which layer should your business sit on?
Four tests, in order. Run them honestly and the answer usually falls out in twenty minutes.
- The blame test. When the product fails at 2am, whose name does the customer say out loud? If it's the parent, you are already on layer one or two whether your brand book says so or not.
- The overlap test. What share of your buyers could buy from two or more of your businesses? Above roughly a third, consolidation pays for itself in media efficiency. Below a tenth, you are subsidising a story nobody in either segment needs.
- The permission test. Does the parent name give you the right to be believed in the new category, or only the right to be noticed? Noticed is not enough. Aditya Birla had cement and distribution credibility, and it still spent to build a product brand rather than selling paint under a corporate name.
- The exit test. Is there a realistic chance this business gets sold in five years? If yes, layer four or five, and accept the cost. Untangling a masterbrand later is far more expensive than running separately from day one.
Notice what isn't on that list: the org chart, the founder's preference, and whether the acquired brand's team will be upset. Those are real political constraints, and you will have to manage them. They are not inputs to the decision.
The endorsement trap
Here's where I part company with the standard advice. David Aaker's brand relationship spectrum treats the endorsed brand as a sensible middle path, and almost every agency deck reaches for it. I think it's the weakest layer on the ladder, and the most expensive.
Endorsement is the most-chosen layer in brand architecture and the least defensible one. It's a hedge, not a strategy.
The logic against it is simple. An endorsement asks the buyer to store two names and one relationship between them. In a category where the Ehrenberg-Bass Institute's 95-5 rule applies — John Dawes' 2021 finding that up to 95% of business buyers aren't in market at any given time — you are trying to build three memory structures instead of one, among people who aren't paying attention. The maths does not work. It's the same reason tracking raw brand awareness in B2B tells you so little: recall of a name is cheap, and recall of a relationship between two names is not.
The honest defence of endorsement is that it buys internal peace. That's a legitimate thing to buy. Just price it, put it in the budget line, and revisit it in eighteen months instead of letting it become permanent by default.
What Tata Neu and Birla Opus got differently
Two of the largest architecture bets in recent Indian business went in opposite directions, and both are instructive.
Tata Neu is a layer-two sub-brand built to sit above an existing set of strong businesses — a masterbrand wrapper applied after the fact. The scale is real: Tata's chairman has cited GMV of ₹46,515 crore within four years of launch. So is the cost. Tata Digital reported a net loss of ₹4,974 crore in FY26 on revenue of ₹35,990 crore, with BigBasket's B2C arm accounting for roughly 64% of the losses. The architecture question that sits underneath is whether customers who already trusted Croma and BigBasket needed a third name in between them and the parent.
Birla Opus went the other way and, so far, more cleanly. The Aditya Birla Group put roughly ₹10,000 crore behind a named product brand rather than selling paint as a corporate line extension. By March 2025 it had reached the top three decorative paint brands in India by revenue share, with around 50,000 dealers onboarded, 137 depots and coverage of over 6,600 towns. Outlook Business reported in 2026 that the brand had taken roughly 7% market share, with Asian Paints' share moving from 51% to 47%. The parent name is present, but the product brand carries the promise. Layer two, executed as layer two, not as a hedge.
The counterpoint a smart peer would raise: Birla Opus bought that share with distribution and capital, not architecture. Fair. Grasim's net debt rose from ₹4,300 crore in FY22 to ₹35,402 crore in FY25. But distribution gets you tried once. The name is what gets you specified the second time.
What the Tata Motors demerger says about structure versus brand
The cleanest recent proof that architecture and corporate structure are different questions came in 2025. Tata Motors' commercial vehicle business demerged effective 1 October 2025 and listed separately on 12 November 2025, splitting a company into two listed entities with different customers, technology platforms and business models.
The brand didn't split. Both entities still lead with Tata Motors. A fleet operator buying trucks and a family buying an SUV are about as far apart as two buyer sets get, and the group still concluded that one name serves both better than two. That's the blame test and the permission test answering the same way, independently of who owns the P&L.
If you take one thing from this: legal separation is a tax and governance decision. Brand separation is a memory decision. They are allowed to disagree, and when they do, the memory decision should usually win.
What this means for you on Monday
A working sequence, in the order I'd actually run it inside an engineering-led business:
- Write down every name you currently spend money on. Product brands, sub-brands, programme names, internal initiative names that leaked outside. Most groups find twice what they expected.
- Put a rupee figure against each. Media, collateral, trade, events, digital. If a name has no budget, it isn't a brand; delete it from the architecture and stop protecting it.
- Run the four tests on each one. Blame, overlap, permission, exit. Record the answer in one sentence each, not a slide.
- Kill or fold anything that fails three of four. This is where the savings are, and it's the only part of a rebrand that pays for itself inside a year.
- Fix the naming rule before the logos. A one-page rule for how new products get named prevents the next ten arguments. Designing the lock-ups first guarantees you'll redo them.
- Sequence the rollout by revenue, not by visibility. Signage and stationery are the last 10% of the work and the first thing everyone wants to discuss.
Globally, the direction of travel favours clarity. Interbrand's Best Global Brands 2025 put the total value of the top 100 at USD 3.6 trillion, up 4.4%, with Nvidia rising 116% and Tesla falling 35% in the same year. The gainers are, almost without exception, single-name businesses where the corporate brand and the product promise are the same thing. That's not a coincidence, and it isn't only a tech story — the same pattern shows up in how the strongest supply chain and logistics brands have consolidated over the last decade.
None of this gets easier with better tools. AI has changed a lot about brand work, but it can't tell you who your customer blames at 2am. That answer only comes from asking them.
So the question I'd put to anyone sitting on layer three right now: if you had to remove one name from your architecture this quarter, and the budget stayed the same, which one would you remove — and what has stopped you so far?
