Google's brand is worth $433bn. Or $1.48tn.

On 28 October Interbrand publishes Best Global Brands 2026, and the headlines will do what they always do. They will take one number per brand and report it as a fact. Before they do, it is worth putting this year’s numbers side by side.

Kantar BrandZ, published in July, values Google’s brand at $1.48 trillion. Brand Finance’s Global 500, published in January, puts it at $433.1 billion. That is the same brand in the same year, about 3.4 times apart. Amazon is $1.02 trillion at one house and $369.9 billion at the other. Apple is $1.38 trillion at Kantar and $607.6 billion at Brand Finance. Interbrand’s 2025 edition puts it at $470.9 billion. That edition is the latest until the 28th, so it is a year older than the other two.

The usual reaction comes in two forms. One says that if the experts can’t agree, the number is worthless. The other comes from the valuers themselves. In 2015 Brand Finance’s chief executive, David Haigh, called a 42% spread on Apple across the three houses “hardly surprising” and said the variation showed the industry’s “health and vitality”. A 42% spread is one thing. A spread of more than three times is a different conversation, and neither reaction fits it.

The divergence is not noise. It is the most useful thing in the tables.

The methods answer different questions. Brand Finance uses royalty relief. Roughly, it asks what a company would pay to license the name if it did not own it. Interbrand only admits brands with a positive long-term economic profit and a brand strength score of 50 or more, so it starts from the profit the business actually makes. Kantar says AI pushed its Top 100 to a record $13.1 trillion, up 22% in a year. Brand Finance’s total rose 11%. Its Google rose 5%.

Read that way, the gap shows where the houses disagree, and they disagree most about the brands whose future is most tied to AI. A licensing-rate method tends to be cautious about expectations. A method that lets this year’s AI story into the number will move with it. So the 3.4 times on Google says less about Google’s brand than about how much of the future each method is willing to count today.

That matters to anyone who has to quote a brand value inside a company.

First, a brand value without its method beside it is a headline, not a number. If a board deck says the brand is worth X, the next line should say whose X it is and what question it answers.

Second, choose the question before you choose the house. If you want to know what the name would be worth in a sale or a licence, royalty relief is closer to that question. If you want to know how much of your profit depends on the brand, use a method built on economic profit. Either way, know which question a given table is answering before you repeat its number.

Third, read the spread as information about risk. When competent valuers land three times apart on your brand, the market hasn’t settled what your brand is for.

It is also worth remembering who publishes these tables. All three houses sell brand valuation or the research behind it, and a record year is good marketing for that product. That doesn’t make the numbers wrong. It means each number is a position, not a neutral fact.

So when the 28 October numbers arrive, the interesting question won’t be who is first. It will be how far apart the three houses stand on the same name, and which brands they can’t agree on.