Seven marketing lessons from the smoothest man on Madison Avenue – and what ignoring them cost real companies between 1965 and last Tuesday.

There is a moment in season four of Mad Men when Bert Cooper – a man who is roughly nine hundred years old and has not worn shoes at the office in a decade – delivers the most surgical line in seven seasons of television. Sterling Cooper Draper Pryce has just lost Lucky Strike. Roger Sterling, partner, rainmaker, wit, the man whose name is on the door, comes looking for sympathy. Bert declines to provide any: “Lee Garner, Jr. never took you seriously because you never took yourself seriously.”

Roger is not the hero of the show, or even the best ad man in the room. What he is – and this is why he matters more than the conference-keynote crowd gives him credit for – is the only major character whose failures are strictly operational. Don Draper’s catastrophes are existential and largely unhelpful unless you also have a stolen identity. Roger’s catastrophes are things that happen inside a CRM. They fit on a whiteboard. So let’s put them on a whiteboard.

The Setup: One Client, Seventy-One Percent, Zero Leverage

In the season four premiere, Lane Pryce – the British money man, the only person at the agency who reads a balance sheet voluntarily – informs the partners that Lucky Strike now accounts for 71 percent of the firm’s billings. Seventy-one percent. One client. Serviced almost entirely by one man, who inherited the relationship from his father and maintained it, as far as anyone can tell, through lunch.

Then, in “Chinese Wall,” Lee Garner Jr. tells Roger over lunch that American Tobacco is consolidating its accounts at BBDO. Roger negotiates a thirty-day stay of execution to break the news internally and build a plan. He spends those thirty days publishing a memoir and telling nobody. The partners find out because Ken Cosgrove runs into a guy from BBDO at dinner with his fiancée’s parents. Cornered, Roger fakes a call to Garner in front of four partners by pressing down the receiver hook, then fakes a rescue trip to North Carolina while hiding in a Manhattan hotel.

Don’s verdict, delivered at a volume you can feel through the screen: “One damn account, and you ignored it!”

Everything below flows from that lunch.

Lesson 1: A Client at 71% Isn’t a Client. It’s a Landlord.

Modern acquirers have done the math Roger never bothered with. The consensus red flag in M&A diligence is any single customer above 10 to 15 percent of revenue, with the top five under 40 percent. Cross 20 percent and buyers knock 1 to 2 full turns of EBITDA off the price. Cross 30 percent and the discount runs 20 to 35 percent versus a diversified peer – on a business throwing off $5 million in EBITDA, that is $5 to $10 million of purchase price gone because of how your revenue is shaped, not how much of it there is.

By that yardstick Roger wasn’t running an agency. He was running a subsidiary of American Tobacco with its own letterhead. The tell showed up seasons earlier: when Lee Garner Jr. demanded that art director Sal Romano be fired, Sal was fired. A client that large doesn’t just buy your services. It buys a vote on your personnel decisions.

Concentration doesn’t only put your revenue at risk. It quietly transfers governance of your company to somebody who doesn’t work there.

Lesson 2: You Have a Lee Garner Jr. Too. It Just Doesn’t Take You to Lunch.

Here is the update Matthew Weiner didn’t write: your Lucky Strike is a distribution channel, and it will not call you first.

LittleThings was a feel-good publisher pulling 58 million unique visitors at its peak. In 2016 its CEO told the Wall Street Journal that “Facebook loves publishers” and that publishers had nothing to fear. In January 2018, Facebook retuned the News Feed to favor friends and family. LittleThings’ organic and influencer traffic – its highest-margin traffic – fell more than 75 percent. The shutdown memo landed on February 27, roughly five weeks later, and said that no prior algorithm update had come close to that level of destruction. Acquirers walked. The company was gone.

Chegg is the same story with a bigger number attached. Non-subscriber traffic – students arriving from Google – fell from down 8 percent in Q2 2024 to down 49 percent in January 2025 after AI Overviews rolled out. Chegg cut 248 jobs in May 2025 (22 percent of staff), then 388 more in October (45 percent of what remained), citing the new realities of AI and reduced Google traffic. A company once valued near $12 billion traded at $0.96 in July 2026. Nobody at Google took its CEO to lunch first.

And the tide is still going out. Pew found users click a traditional search result 8 percent of the time when an AI Overview is present, versus 15 percent without one. Ahrefs measured a 34.5 percent click-through decline for the top-ranking page in April 2025; by December 2025 it had widened to 58 percent. Chartbeat data across 2,500-plus news sites showed Google referrals down 33 percent over 2025. Roughly 60 percent of searches now end without a click at all – 77 percent on mobile.

Even the giants aren’t immune. When Apple shipped App Tracking Transparency in iOS 14.5, Meta’s CFO estimated the 2022 revenue headwind at about $10 billion – and, to his credit, said plainly in the same breath that the company couldn’t be precise and that it was an estimate. One operating-system update, one dialog box, ten billion dollars of somebody else’s revenue.

Lee Garner Jr. didn’t hate Roger. He consolidated. The algorithm doesn’t hate you either – which is exactly why charm won’t save you from it.

Lesson 3: “The Day You Sign a Client Is the Day You Start Losing Them”

That line is Roger’s, from his memoir, and it is the best sentence about retention ever written by a fictional alcoholic. It is also one he wrote and then comprehensively failed to follow, which is the most human thing about him.

The underlying economics were established in a 1990 Harvard Business Review study by Frederick Reichheld and W. Earl Sasser, which found that cutting the customer defection rate by 5 percent raised profits by 25 to 85 percent across the companies studied – with the top result, an 85 percent gain, coming out of a bank branch system. HBR revisited the figure in 2014 as the now-famous 25-to-95-percent range. The companion numbers are just as blunt:

• Acquiring a new customer costs 5 to 25 times more than keeping one you already have (Harvard Business Review, 2014).
• The probability of selling to an existing customer runs 60 to 70 percent. To a new prospect, 5 to 20 percent (Marketing Metrics).

Roger treated Lucky Strike as won. Won accounts don’t exist. There are only accounts that haven’t left yet, and the ones you’re actively re-earning.

Lesson 4: Charm Is a Distribution Channel. It Is Not a Product.

Roger’s most quotable professional observation is that half the time this business comes down to “I don’t like that guy.” He is right, and every B2B seller reading this knows he’s right. The problem is what Roger did with the insight: he treated the relationship as the moat rather than as the drawbridge.

Relationship capital lives inside a person, not a company, and it depreciates the moment that person coasts. M&A advisors flag the diligence version of this constantly: revenue where the relationship lives almost entirely in one founder’s personal rapport. Buyers hate it, because rapport is not transferable. It walks out with the founder.

The fix is unglamorous: multithread. If your largest client’s loyalty runs through exactly one steak dinner a quarter, you do not have an account. You have a friendship with a revenue line attached.

Lesson 5: The Fake Phone Call Always Gets Found Out

Roger had thirty days of advance warning – a genuine strategic asset most companies never get. He converted it into zero days of preparation and one humiliating scene involving a receiver hook. By the time the partners knew, they had no runway, no counter-pitch, and no story.

Compare Don’s move two episodes later: he buys a full-page letter in the New York Times announcing the agency will no longer take tobacco business. Was it reckless? Extremely. Was it good crisis marketing? Yes, and for a specific reason – it converted “we were dumped” into “we walked,” and it did so on a channel Don controlled, in front of an audience nobody could throttle. He couldn’t win back the client, so he bought the megaphone.

The window between “you know” and “everyone knows” is the most valuable inventory a marketer will ever be handed. Roger spent his on a book tour.

Lesson 6: Your Personal Baggage Is a Line Item on the P&L

In “The Chrysanthemum and the Sword,” the agency chases Honda. Roger, a Pacific theater veteran, nearly torches the meeting with a wartime grudge he has nursed for fifteen years. He is entitled to the feeling. He is not entitled to expense it to the partnership.

The modern translation is not about war. It is the founder who won’t touch a channel because they find it undignified, the CMO who kills a direction that isn’t to their taste, the executive who still refuses to go on camera in 2026. Each one is a personal preference being charged to the income statement, and the invoice arrives quarterly.

Lesson 7: Build the Owned Channel Before You Need It

Everyone on the show learns the same thing eventually: you want an asset nobody can cancel. In 1965 that was a client list in a locked drawer. In 2026 it is email, and the numbers are not close – $36 to $42 returned per dollar spent across industries, $45 in retail and consumer goods. Paid search returns roughly $2, paid social about $2.80, display $1.35.

Email wins partly because it’s cheap, but mostly because nobody else holds the dial. Every dollar spent building a list buys insurance against the next News Feed update, core update, privacy prompt, or AI Overview. Roger never had to think this way; he inherited his distribution. You did not.

The Sterling Correction: Seven Things Worth Doing This Quarter

1. Calculate your real concentration number. Top client and top five as a share of revenue, tracked monthly. Above 20 percent is not a milestone, it is an exposure.
2. Do the same for channels. If more than half your traffic, leads, or sales arrive through one platform, you are Roger in 1964 and you don’t know it yet.
3. Multithread every account above 10 percent. Three named contacts, different functions, introduced to three people on your side.
4. Build the email list on purpose, not as a byproduct. Put a real budget behind it and measure it against the $36-to-$42 benchmark, not against zero.
5. Run a loss drill. Assume your biggest client or channel disappears Monday and write the ninety-day plan on one page. Costs an afternoon. Not having it cost SCDP the firm.
6. Move the retention budget before the acquisition budget. A 5 percent retention improvement is worth 25 to 95 percent more profit. New logos are worth applause.
7. When bad news arrives, spend the window, don’t hoard it. The thirty days between the lunch and the leak is the whole game.

The Last Word

Roger’s tragedy is not that he lost Lucky Strike. Lee Garner Jr. was always going to consolidate; the decision had nothing to do with the quality of Roger’s martinis. His tragedy is that he built a career, a partnership, and an identity on top of a single relationship he did not control, treated maintaining it as beneath him, and then hid the loss until hiding it made the loss worse. That is not a 1960s problem. That is a Tuesday. The only real difference between Roger and an operator running 70 percent of revenue through one platform today is that Roger got a lunch, a warning, and thirty days. You get a changelog entry.

Sources

Retention economics: Reichheld, F. & Sasser, W.E., “Zero Defections: Quality Comes to Services,” Harvard Business Review, Sept-Oct 1990 (5% defection cut = 25-85% profit gain; 85% in one bank branch system). Gallo, A., “The Value of Keeping the Right Customers,” HBR, October 2014 (25-95% range; 5-25x acquisition cost). Marketing Metrics (60-70% vs 5-20% close probability).

LittleThings: Digiday, “LittleThings shuts down, a casualty of Facebook news feed change,” Feb. 2018 (58M peak uniques; comScore decline to 40M). Business Insider and TechCrunch coverage of the Feb. 27, 2018 staff memo (75%+ organic and influencer traffic loss). Wall Street Journal, June 2016 (CEO Joe Speiser on publisher-platform reliance). Nieman Journalism Lab, October 2018.

Chegg: Fox Business and Higher Ed Dive, Oct. 2025 (388 layoffs, 45% of workforce; up to $110M targeted savings). Forbes, “Chegg Stock Down 99%,” Oct. 2025 (non-subscriber traffic -8% in Q2 2024 to -49% in Jan. 2025, per Washington Post). EBC Financial Group analysis of Chegg Q1 2026 results and Form 10-Q, July 2026 ($0.96 share price; ~99% off the Feb. 2021 peak).

Search and platform shift: Pew Research Center, 2025 study of ~68,000 queries (8% click rate with AI Overviews vs 15% without). Ahrefs, April 2025 and February 2026 studies across 300,000 keywords (34.5% to 58% CTR decline at position one). Chartbeat / Press Gazette, 2,500+ news sites (Google referrals down ~33% across 2025). CNBC, Feb. 2, 2022 (Meta CFO David Wehner, Q4 2021 earnings call, $10B iOS headwind estimate and his caveat that it was an estimate).

Benchmarks: Wall Street Prep and CT Acquisitions, 2026 concentration thresholds and valuation discounts (10-15% red flag; 1-2 turns of EBITDA above 20%; 20-35% discounts above 30%). Litmus / DemandSage / Omnisend 2026 email benchmarks ($36-$42 per dollar; $45 retail; paid search ~$2, paid social ~$2.80, display ~$1.35). Mad Men, AMC, season four, episodes 4.01, 4.05, 4.11 and 4.12; Sterling’s Gold, Grove/Atlantic, 2010.