Seven marketing lessons from Harry Crane, the least likable visionary in advertising history – and what he tells us about the AI lead who just started running your company.

He Opened the Wrong Paycheck

Every great career pivot has an origin story, and Harry Crane’s is petty. In the third episode of Mad Men’s second season, “The Benefactor,” a soft, bespectacled media man opens an envelope that is not his. It is Ken Cosgrove’s paycheck. Ken is making $100 more than he is.

Most people would sulk into a highball. Harry called a friend at CBS.

He had heard that an upcoming episode of the courtroom drama “The Defenders” dealt with abortion, and that sponsors were running from it like a grease fire. Harry’s insight was not moral. It was arithmetic: panic makes inventory cheap. He pitched the discounted slot to Belle Jolie lipstick. Belle Jolie passed – the subject was radioactive in 1962 – but Roger Sterling noticed the hustle.

He did not ask for a raise. He asked to run a department that did not exist.

Why Harry Matters More Than Don

We all want to be Don Draper – the carousel, the cigarette, the closed eyes and the perfect line. But Don sold the twentieth century. Harry priced it. And the numbers behind Harry’s little one-man department were the most violent shift in the history of American media.

Television went from a curiosity to an appliance in a single decade. Per the Television Bureau of Advertising, citing Nielsen: 3.88 million U.S. homes had a set in 1950, which was 9.0% of households. By 1955 it was 30.7 million homes, or 64.5%. By 1960: 45.75 million homes, 87.1%. That is a nine-fold jump in penetration in ten years.

The ad dollars followed with a lag, then a stampede. Television advertising revenue was roughly $41 million in 1951 and $336 million by 1953 – an eight-fold rise in twenty-four months. Media historians put early-1950s TV spend around $85 million; inside a decade it cleared $1 billion and buried radio. By the early 1960s, more advertising dollars were going to television than to print, and newspaper share of ad spend began a slide it has never once reversed.

Now run the same curve on your own desk. Salesforce’s State of Marketing survey of roughly 4,500 marketers tracks the share using generative AI in at least one recurring workflow: 51% in Q1 2024, 76% in Q1 2025, 87% in Q1 2026. That is 36 percentage points in twenty-four months – the fastest sustained technology adoption marketing has ever recorded. The Duke Fuqua CMO Survey puts it more soberly: generative AI now touches 15.1% of all marketing activities, up from 7.0% a year earlier.

Nine percent to eighty-seven percent. Fifty-one percent to eighty-seven percent. Same shape. Same decade-compressed panic. Same guy in the corner office nobody wants to sit next to at lunch.

Lesson 1: The Next Channel Always Shows Up Looking Like a Chore

In 1962, television was not glamorous inside an agency. It was logistics. It was screening rooms, ratings cards, and haggling with networks – work the creative floor considered beneath them. Harry took it precisely because nobody was guarding it.

That is the pattern, and it repeats with irritating reliability:

1. The thing that eats the industry always arrives disguised as unglamorous operations work.
2. The people with status refuse it because taking it looks like a demotion.
3. Whoever picks it up owns a monopoly for roughly three years, which is exactly how long it takes everyone else to panic.

Search was “webmaster stuff.” Social was “the intern’s job.” Programmatic was “ad ops.” Today it is the person quietly maintaining the prompt library and arguing with the legal team about model governance. Check who in your building is doing the boring version of the future. That is your Harry.

Lesson 2: Ask for the Department, Not the Raise

Harry’s genius move was structural, not financial. A raise is an expense line. A department is a budget, a headcount, a seat in the meeting, and a name on the org chart that nobody can quietly delete. He converted a favor into an institution.

The modern version is happening at absurd speed. IBM’s Institute for Business Value, surveying 2,000 CEOs across 33 countries, found that 76% of organizations reported having a Chief AI Officer in 2026, up from 26% in 2025 – roughly a threefold expansion in one year, which may be the fastest C-suite land grab in corporate history. ZoomInfo’s title data showed AI-related C-suite titles up 428% between Q2 2022 and Q2 2024 (81 to 428). Analyses of Fortune 500 disclosures indicate 54% of these AI chiefs report straight to the CEO.

The person explaining what a language model was two years ago now controls a budget.

Lesson 3: Do the Math Out Loud

Harry’s best trait was that he always knew his number. When he engineered the “Broadway Joe on Broadway” special to rehabilitate Dow Chemical’s napalm-scarred image, the partners handed him a commission check for $23,500 – more than his annual salary. He then stormed a partners’ meeting and reminded the room he had brought in $150,000 of incremental business, in daylight, on the record.

He was obnoxious about it. He was also right, and he was the only person in that room who could state his own contribution in dollars without checking a file.

Compare that with the current AI moment, which is drowning in adoption and starving for arithmetic. MIT’s NANDA initiative – 150 executive interviews, a survey of 350 employees, and 300 public deployments – found that roughly 95% of enterprise generative AI pilots delivered no measurable profit-and-loss impact, with only about 5% producing rapid revenue acceleration. The report’s quiet gut-punch for our field: budgets are concentrated in sales and marketing pilots, and that is where measured ROI is lowest.

It gets worse before it gets better. Industry surveys report 88% of marketers using AI tools daily while only 19% track AI-specific KPIs. Gartner’s 2026 CMO Spend Survey of 401 marketing leaders found just 9% rated their internal processes as fully optimized to implement and scale AI, with another 21% considered mature. Everyone is using it. Almost nobody is scoring it.

Be Harry. Pick one number – incremental revenue, cost per qualified lead, hours returned to senior staff, cycle time from brief to live – and defend it in public with a straight face.

Lesson 4: Fluency Is a Moat Until Everybody Is Fluent

Harry was not the best ad man at Sterling Cooper. He was not in the top five. What he had was fluency: he spoke network, he knew the buyers, he understood ratings the way Don understood longing. For roughly five years, that made him irreplaceable.

Then the fluency got automated. In 2025, U.S. programmatic digital display spending grew 16.6% to more than $187 billion, accounting for nearly 95% of all digital display ad spend, and 2026 is forecast to top $220 billion. The skill Harry built a career on – knowing which human to call to buy which slot – is now an auction that resolves in under 100 milliseconds without a phone.

The lesson is not “don’t learn the tool.” The lesson is that tool fluency is a ticket, not a seat. It buys you a few years of leverage, and you must convert that leverage into judgment, relationships, and ownership before the software catches up. Harry converted his into a Los Angeles office and a rolodex of network executives. That is why he survived past the point where anyone could buy time by dialing a phone.

Lesson 5: Competence Without Charm Has a Hard Ceiling

And now the tragedy, because this is where Harry stops being a role model and starts being a warning.

He was right about everything and liked by nobody. He asked for a partnership for the better part of a decade and was refused every time. When he compared himself to Bert Cooper, Bert replied that they were different in every way. When the partners bought him off with that check, Roger wondered aloud whether they should fire him before he cashed it. The man who invented the agency’s most valuable department could not get a vote.

Being indispensable is not the same as being promotable. Organizations do not hand power to people who make meetings unpleasant, however correct those people are. And the numbers say the modern equivalents are living on a similarly short clock: Spencer Stuart data put average CMO tenure at S&P 500 companies at 4.1 years in 2025, down from 4.3 in 2024, versus about 5 years across the C-suite and 7.6 for CEOs. Reported average Chief AI Officer tenure at Fortune 500 firms sits near 2.1 years, against roughly 4.2 for CTOs and 5.1 for CFOs.

Your technical edge is the entry fee. Calming a nervous CFO is what buys you tenure.

Lesson 6: Leverage Rots. Take the Deal.

Here is the detail that should keep every specialist awake. The partners finally decided to offer Harry the partnership he had begged for. He took too long to answer. While he deliberated, the firm sold to McCann Erickson – and the payout that would have made him rich closed without him in it.

He spent ten years being underestimated, won the argument, and then missed the window by savoring it.

Windows close fast right now. WPP ended December 2025 with 98,655 employees, down from 108,044 a year earlier, while its shares fell roughly 60% in twelve months and its market value shrank to about £3 billion from roughly £25 billion nine years prior; the company has since announced a plan targeting £500 million in annualized savings by 2028 as it converts from a holding company into a single AI-enabled operator. Omnicom and IPG cut about 8,200 roles before their merger – a 6.4% reduction in combined headcount – and signaled roughly 4,000 more after. Forrester projects that 15% of agency roles will be automated, and 91% of senior U.S. agency leaders expect AI to reduce headcount.

When someone offers you the seat, sit down.

Lesson 7: The Machine Eventually Comes for the Machine Guy

In season seven, Sterling Cooper installs an IBM System/360 and it swallows the creative lounge – the couch where the copywriters used to loaf becomes a refrigerated room full of tape drives. It is the show’s bluntest metaphor and it lands: the technology that promises leverage first takes your space, then your process, then your certainty about what you are for.

That is also the honest read of this AI moment, and pretending otherwise makes you sound like a vendor. Basis Technologies reports that 54% of advertisers believe generative AI has contributed to a decline in overall media quality. McKinsey research reported via MarTech found 62% of organizations still experimenting with agentic AI and only 23% actually scaling it. Gartner’s 2026 survey shows the martech share of marketing budgets at a five-year low of 19.4%, down from 26.6% in 2021, even as 62% of CMOs plan to invest more – a contradiction that means one thing: buyers are getting skeptical and specific.

Harry’s department was built on human relationships with network sales chiefs. Ninety-five percent of display is now machine-to-machine. He would have had to become something else. So will you.

What To Actually Do On Monday

1. Find the chore. Identify the unglamorous operational work in your company that touches the new channel, and volunteer for it before it has a nice title.
2. Name it. Do not ask for a bonus, ask for a function – a budget line, a headcount, and a place on the agenda. Institutions outlive favors.
3. Pick your number in week one. Baseline it before you deploy anything. Ninety-five percent of pilots show no P&L impact largely because nobody wrote down the “before.”
4. Publish a monthly scoreboard. One page. Wins, losses, dollars. Harry never had to be asked what he was worth.
5. Spend 20% of your effort on being pleasant. Not fake – pleasant. Bring the CFO in early. Explain things without condescending. This is not soft-skills fluff; it is the difference between 2.1 years and a partnership.
6. Convert fluency into ownership within eighteen months. Relationships, proprietary data, and P&L accountability survive automation. Knowing the keyboard shortcuts does not.
7. When the offer comes, take it that day.

The Bottom Line

Harry Crane is the most quietly prophetic character in Mad Men, and the show punishes him for it with seven seasons of mild contempt. He saw the channel shift first, built the department nobody wanted, priced the future by the minute, and still could not get a vote out of a room full of men who owed him millions in billings.

Your AI lead is Harry. Maybe your AI lead is you. The technical read on the future is the easy half. The hard half is making people want to hand you the keys – and then, when they finally offer, not being too pleased with yourself to reach out and take them.

Sources

Television Bureau of Advertising / Nielsen, “National TV Household Penetration Trends” (1950: 9.0%; 1955: 64.5%; 1960: 87.1%).

Britannica, “Television in the United States: The Late Golden Age”; Library of Congress research guides (1950 vs. 1959–60 household penetration).

Houston Chronicle Small Business, “What Caused the Advertising Industry Boom in the 1950s?” (TV ad revenue $41M in 1951; $336M in 1953).

Media Culture, “Expert Interview: The Evolution of TV Advertising” (early-1950s TV spend ~$85M to $1B+ within a decade; TV passes print in early 1960s).

Benedict Evans, “News by the Ton: 75 Years of US Advertising” (newspaper share of ad spend declining since the 1950s).

Salesforce, State of Marketing 2026 (~4,500 marketers; generative AI in at least one recurring workflow: 51% Q1 2024, 76% Q1 2025, 87% Q1 2026).

The CMO Survey, Duke Fuqua / Deloitte / AMA, Spring 2025 (generative AI applied to 15.1% of marketing activities, up from 7.0%).

IBM Institute for Business Value, 2026 CEO Study (2,000 CEOs, 33 countries; 76% of organizations report a Chief AI Officer, up from 26% in 2025).

ZoomInfo via Business Wire, August 2024 (AI C-suite titles up 428%, Q2 2022 to Q2 2024).

MIT NANDA initiative, “The GenAI Divide: State of AI in Business 2025” (150 executive interviews, 350-employee survey, 300 deployments; ~95% of pilots with no measurable P&L impact).

Gartner CMO Spend Survey 2025 and 2026 (marketing budgets at 7.7% of revenue; martech share at a five-year low of 19.4%; 9% of CMOs fully optimized to scale AI; 401–402 respondents).

Spencer Stuart CMO Tenure data, reported by Adweek, January 2026 (S&P 500 CMO tenure 4.1 years in 2025 vs. 4.3 in 2024; C-suite average ~5 years; CEO 7.6).

eMarketer / Basis Technologies programmatic forecasts, 2025–2026 (U.S. programmatic display up 16.6% to $187B in 2025, ~95% of digital display; $220B+ forecast for 2026; 54% of advertisers cite AI-driven decline in media quality).

VideoWeek and Storyboard18, February 2026 (WPP headcount 98,655 vs. 108,044; share price down ~60%; £500M savings plan by 2028).

eMarketer, “Omnicom’s Latest Report Signals an Agency Shift from Human Talent to AI” (8,200 roles cut pre-merger, ~4,000 planned after; 91% of senior U.S. agency leaders expect AI to reduce headcount).

Forbes, “10,000 Marketing Jobs Are Gone And AI Agents Took Them,” August 2026 (Forrester projection: 15% of agency roles automated).

Mad Men, AMC. Episodes referenced: “The Benefactor” (S2E3), “Shut the Door. Have a Seat” (S3E13), “To Have and to Hold” (S6E4), “The Monolith” (S7E4). Plot details via IMDb episode summaries and the Mad Men Wiki.