Brand Is Worth Zero on Your Books. Until It's Worth Billions.

In 2021, Intuit agreed to pay about $12 billion for Mailchimp. Mailchimp had bootstrapped for twenty years. It spent its own money on a winking chimp, strange podcast spots and a voice people liked. For two decades that spending sat in the expense column. On closing day it became part of the price.

Under international accounting rules, a company can't put the brand it built on its own balance sheet. IAS 38 is blunt about it: internally generated brands are not recognized as intangible assets. Most CFOs know this. Most founders have never heard it.

So every dollar you spend building a brand lands as a cost. Then someone buys you, and the same brand gets valued and booked as an asset in the buyer's accounts.

Brand is a cost on the way in and an asset on the way out. Throughout my career, brand measurement has always been "mushy", but with the arrival of AI, we can be more deliberate about how we measure it beyond the "well duh, that worked" (see Ramp's Broadway show). It's important for brand people like myself because most new AI companies aren't really marketing and if they are, it's growth hacking, so brand has to sing for its supper. This note is about why the companies treating brand as a cost in 2026 may learn its value by going without it.

What a brand is

A brand is what the market already knows, remembers and believes about you before you walk into the room. It comes down to two questions. Memory: when a buyer's problem shows up, do they think of you? Meaning: once they do, do they trust you?

The logo, the color palette and the campaign are inputs. The brand is what they leave behind in someone's head, and most of that comes from places marketing doesn't own.

• Your people. Who works here, why they stayed, what they say about you at dinner.
• Your story/design. The product, the docs, the error messages, the invoices. Linear's job posts sum it up as "quality creates gravity."
• Your culture. How you treat a customer at 2 a.m. or a vendor who's late, and everyone who interviews at your company (pay attention to this companies!. If you ghost that candidate, they're not going to be a future vendor or buyer, likely)
• Your reputation. What people say when you're not there.

Why now: the 1,000 robot problem

Unless you're strictly running a PLG motion, when anyone can build the product, the product often stops being the difference. Anyone can copy features, or heck: reverse engineer your entire (Adobe) product.

Start with robots. Berg Insight counts more than 100 companies building full-size humanoids. In the first half of 2026, China registered 116,000 new companies in the humanoid robotics sector. Soon they'll all have a robot that walks across a stage and folds a shirt. The buyer will pick the one they remember and trust.

Software tells the same story. Scott Brinker's martech map counted 150 tools in 2011 and 15,384 in 2025, and that was before coding agents made a weekend MVP normal. When engineering is abundant, attention and trust are what's scarce.

Buyers' attention has moved, too. LinkedIn launched a Creator Marketplace in 2026, and says 82% of B2B marketers believe creators increase credibility with decision-makers. Creators are rented trust. But Kantar finds only 27% of creator content strongly ties to brand messaging. Rented voices with nothing to point at are just expensive noise. (Hot take).

Meanwhile, inside the company

Only 69% of CMOs say their CEO and CFO believe in long-term brand building, down from 80% in 2024.

Marketing has quietly been shrunk to two jobs, demand gen and product marketing. Both are necessary, and neither one creates the memory that demand gen depends on.

Demand gen is a harvest, and somebody has to plant. A company that treats marketing only as a cost center, or only as a pipeline machine, is eating its seed corn and calling it efficiency.

A feeling is a fact

In 1959, Volkswagen had a small, odd car with an awkward history in a market of tail fins and chrome. Doyle Dane Bernbach put a tiny Beetle in a sea of white space under two words: Think Small. Then came a second ad with one word: Lemon. In 1999, Ad Age named the campaign the best of the 20th century.

Nobody bought a Beetle for the horsepower. The ads made people feel smart for choosing the small car. The spec sheet told you what the car did; the ads told you what kind of person bought one.

B2B buyers like to think they're above this, and they aren't. The oldest line in enterprise sales, nobody ever got fired for buying IBM, says nothing about features. It sells safety to a buying committee.

The math behind the feeling

John Dawes of the Ehrenberg-Bass Institute gave us the 95:5 rule. Companies switch banks or law firms about every five years, so in any quarter roughly 5% of buyers are in market and 95% aren't.

When the 5% do show up, they've mostly decided. In a WSJ Intelligence / B2B International survey, 83% of business buyers said they start with only two to four vendors. 79% were very familiar with the eventual winner before the search began, versus 33% for a vendor that lost.

So the shortlist gets written by memory, months before the RFP, and demand gen can only compete for the names already on it.

Bernbach was half right

In a 1947 memo, Bill Bernbach wrote that persuasion "happens to be not a science, but an art." He was right about the making and wrong about the proving: the work is still art, but in 2026 its results can be measured.

Brand to demand, not brand vs. demand

Every company pays for brand, either up front as an investment or forever as a tax. The tax shows up as higher acquisition costs, deeper discounts, longer sales cycles and shortlists you never hear about.

Most B2B teams already know this. In Benchmarkit's 2026 survey of B2B tech companies, the median team puts 70% of budget into demand gen and 25% into brand, and says the ideal is closer to 50/40. Only 28% can link brand to pipeline.

Most teams believe in brand and can't show what it does. The missing piece is proof.

A better question for the CFO is what demand gen would cost without brand.

What replaces the funnel

In July 2025, Colin Fleming, then CMO of ServiceNow and now CMO of OpenAI's business unit, wrote a LinkedIn post called "Death of the Funnel?" His starting point is the buying group, which he puts at 22 people. Citing LinkedIn's own research, he notes that the hidden buyers vendors never meet carry nearly equal influence in the deal and are 70% more likely to reject brands they don't recognize. In 81% of winning deals, everyone or almost everyone on the buying team already knew the brand.

"Familiarity isn't a bonus," he wrote. "It's the entry fee." LinkedIn's CEO, Daniel Shapero, singled out another line in the comments: win the whole room, not just the lead.

Fleming put budget behind it. At ServiceNow he cut lead-gen spend by 63%, retired MQLs, reorganized around buying groups and moved to a 50/50 brand-to-demand split. The company reported conversion rates up 133%, without publishing a baseline, and Fleming himself admitted the new model is harder to measure.

That's the gap the scorecard below is built to close, and it's why its three jobs, be known, be chosen, be bought, are not a funnel with new labels. A funnel tracks one buyer moving through stages. In a 22-person buying group, all three jobs happen at once: the CFO who's never heard of you is still at known while the engineer who saw your talk is already at chosen. So the scorecard runs each job on its own clock and counts accounts, not leads.

The scorecard

I built a scorecard to answer the CFO's question in numbers finance already trusts. It follows the three jobs a brand has to do, in order: be known, be chosen, be bought. A fourth tier shows where that lands for the CFO. Each tier runs on its own clock:

The trick is one flag in the CRM. Split every linking and lagging metric by brand-exposed vs. cold accounts. If exposed accounts win more, discount less and cost less to convert, brand's effect is sitting in the CFO's own numbers.

The other half is time. As Matt Stevens put it in The Drum, agree with finance up front that brand gets judged on 18 months, not 18 days.

The case files, and the "well, duh" at the end of each

Four bets, from 2000 to this fall, plus one warning. None of them is a logo refresh. For each one: the bet, the scorecard line it moved, and the outcome that looks obvious in hindsight.

1. Salesforce: cross out the old word, then name the new one

The bet. In February 2000, Salesforce staged a mock protest at rival Siebel's user conference under a crossed-out "software" logo, then built Dreamforce into the place it launched almost everything. Twenty-six years later, on September 11, it introduced seven AI agents with names and job titles: Casey for customer service, Hunter for outbound sales, Piper for inbound pipeline and four more. Customers can rename them.

What moved. Shortlist rate, then and now. Naming the category put Salesforce at the top of every buyer's list for cloud software. The agent names are too new to score on their own, but they sit on a business that reported Agentforce ARR above $1.5 billion, up more than 240%.

Well, duh. Nobody remembers "agentic digital labor." People remember Hunter. The company that crossed out "software" is still turning abstractions into something you can picture.

2. Gong: the Super Bowl, but only where its buyers live

The bet. In 2019 Gong renamed its category "revenue intelligence" so a chief revenue officer couldn't delegate it. In February 2021 it ran its first Super Bowl ad, in just three markets: the Bay Area, New York and Seattle, for a reported $1.5 million.

What moved. Brand-led pipeline. Gong's CMO, Udi Ledergor, said Super Bowl week brought record web traffic that turned into significant sales within months, and the company credits the ad with its biggest-ever pipeline. The next year it tripled the ad budget.

Well, duh. A company named after the bell sales teams ring when they close bought the loudest 30 seconds in America.

3. Mailchimp: make people search for your name

The bet. A 2014 ad before the podcast Serial featured someone mispronouncing the brand as "MailKimp." In 2017, Mailchimp and Droga5 turned the joke into nine fake products and films, from MailShrimp to FailChips, each built so Google would ask: did you mean Mailchimp?

What moved. Share of search. The campaign sparked 3.8 million organic searches and drove 29% more visits than expected. It also won a Cyber Grand Prix at Cannes.

Well, duh. Twenty years of being the brand people liked saying, and mis-saying, ended in a $12 billion deal.

4. ElevenLabs: brand as a trust repair job

The bet. Voice AI has a trust problem, and ElevenLabs has lived it. In 2024, two audiobook narrators alleged its default voices were built from their recordings without consent; the company denied wrongdoing and later settled. Its answer has been to make licensed, famous voices its calling card: an Iconic Marketplace with voices like Sir Michael Caine's, The Odyssey narrated by his AI voice in June, a spot on the Formula 1 grid with the Audi Revolut F1 team, and a multi-year deal with Universal Music Group on September 10.

What moved. Brand value, for now. On September 30, ElevenLabs closed a $300 million employee tender at a $22 billion valuation, double its February mark. Enterprise is 55% of revenue. The open risk sits on the reputation line: in May, journalists and podcasters, including Peabody winner Robin Amer, filed class actions in Illinois alleging ElevenLabs and eight other tech giants trained voice models on their voiceprints without consent.

Well, duh, with a catch. Brand can earn trust faster than anything else, which is why famous voices work. But it can't outrun your training data. The story you tell and what you actually did have to match, or the reputation line on the scorecard eventually tells on you.

And one warning: the 101

The San Francisco Standard reported in August that the biggest billboards on the 101 skyway now run up to $200,000–$250,000 a month, about $3 million a year. Nearly all are leased by AI, tech and finance companies, with lines like "Auth for agents" and "That's so agentic."

Paying to be seen doesn't make you memorable. When every board says "agents," the scorecard's leading tier won't move for anyone. Think Small worked because nobody else could have said it.

The pattern

None of these were demand gen programs. Each is a bet that demand gen gets cheaper when buyers already know your name, and none of them tried to win on specs.

What to do Monday

This part is for every founder whose plan is "build it and the engineers will come." I've led brand through two IPOs and an acquisition, and this is where I'd start:

1. Write the sentence. What do you want buyers to repeat about you when you're not in the room? VW needed two words.
2. Pick one distinctive asset and keep it for years. A color, a sound, a shape, a name you can play with. Gong has a gong. Mailchimp has a chimp people can't pronounce.
3. Treat your interview loop like a launch. Every candidate should leave with a real answer and a good story, because they're next year's buyers and reviewers.
4. Build your own voice before you pay for someone else's. Creators can carry your story only if you have one worth retelling, so start with your founders and engineers.
5. Invest before you need the results. The 95% who'll buy next year are forming their shortlist this year.

Back to the balance sheet

Your accountant will keep booking brand as an expense, and that's fine. The rules aren't going to change. But in a market of a thousand robots and fifteen thousand marketing tools, the product can be copied by Friday.

Brand is worth zero on your books. Until it's the only thing on them a competitor can't copy.

What's the one brand metric your CFO actually believes? I'd like to hear it.

Sources are linked throughout.