Marketing BS Briefing: At Long Last Edition
Branding Hackers, Gymnastic Difficulty, Amazon Advertising, Star Trek Redshirts and more!
It’s been a while since the last Briefing so lots of good stuff to share. If you get a chance let me know your relative preference for essays vs briefings. Now that I am only doing one per week, I can adjust relatively easily between the two. Just reply to this message.
Sponsor’s Message
Get paid for your feedback, join Wynter’s research panel
Want to give back to the community while having a low-key side hustle to fund your habit? Provide feedback on product messaging for industry-leading B2B companies, be compensated every time you do it ($15-$50 per survey).
Takes ~10-15 min to take one survey, low-key time commitment.
Follow-up
Degrees of Difficulty: Last week’s essay explored how difficult it is to determine the degree of difficulty when judging performance. I was remiss in not including the recent gymnastics news. Simone Biles was the first female gymnast to attempt the “Yurchenko Double Pike” in competition. She argues that the judges have given the move a far too low difficulty rating. One potential explanation is that because the move is so dangerous the “raters” do not want to give it a score that would tempt other, less skilled, gymnasts to attempt it. A reminder that “difficulty ratings” are not really about ranking difficulty, they are about shifting incentives.
Merchant Advertising: In “In Praise of Advertising” and “Platform Economics” I wrote about how advertising on platforms allows those platforms to collect more of the value created by making suppliers reveal their true willingness to pay. Last week Instacart revealed that it will hit $1B in revenue next year from it’s in-ap ads.
Stories: In “Everything becomes the same thing” I argued that platforms need to start out different to get to scale, but then begin to become more and more similar to monetize their customer base in every way possible. Case in Point: UberEats has added “stories”
Auction Economics: The morning I published “Platform Economics” Byrne Hobart at The Diff quoted from the piece and expanded on the idea. The post is behind his paywall (Do subscribe. It is excellent!), but here is the gist:
The Expedia model generated cash flow upfront, while Booking.com took in cash on a delay. While some customers didn't care about this, many of them did—but switching models would mean a huge cash flow hit, which would be hard to justify.
Expedia negotiated rankings one-one-one with hotels; Booking just let hotels choose how much of their margin to bid, and ranked them accordingly.
The first concern is the kind of strategic limitation that turns a temporary constraint into an indefinite one. If a growing company has negative working capital, any switch to a positive working capital model has a cost that __grows__ over time. If the switch doesn't happen, it gets harder every year. The second concern is also one embodying temporary tradeoffs. Every auction-based ranking mechanism initially leaves money on the table; it takes time for ad buyers to structure their business model around arbitrage opportunities. So auction-based systems are easy to start with, or easy to switch to early on, but hard to justify later.
Marketing
Instagram: Adam Mosseri (the head of Instagram) has a detailed post explaining how the social network’s algorithm works. If you are trying to optimize for organic reach it is an important read.
Hacker Branding: Why would you ever pay a criminal who hacked into your system? How can you trust that if you pay them they will do what they say? The way around this problem (if you are a hacker) is to build a brand. Hack small places, keep your promises, and scale up. Brands are about trust - even for criminals.
Amazon ads: The company has opened up their ad platform to third party auditing. Amazon ads are mostly focused on driving on-platform purchases, but Amazon owns or could own a lot more ad inventory than that (Fire tablets, IMDB, Prime Video, Kindles, etc). Third party auditing will allow Amazon to use it’s targeting capabilities to go after a much bigger audience (everyone who DOESN’T sell on Amazon). If you haven’t tested Amazon as an advetising platform yet, you should! Related: Amazon ad-supported television has reached 120MM viewers/month. And: Why Amazon will do very well when third party cookies go away.
Corporate Philanthropy: HBR explains how to get the most out of corporate giving - the key is to spend as much or more marketing the donations you give as the amount you give away. This is presented as a terrible thing to do, but it is standard in marketing. If you sponsor the NFL you better have at least as big a budget to advertise that you are a sponsor, otherwise you just wasted those sponsorship dollars.
Reviews: The trouble with reviews is you can’t trust the reviewers.
Marketing to Employees
Northface: The company refused to make jackets for an oil company. So the fossil fuel industry created a marketing campaign “thanking” Northface for using so much petroleum in their products. None of this matters to consumers one way or the other - both companies are appealing to their own employee base.
Middle East: Darren Grimes compiles examples of companies that publicly support LBTGQ rights in America, but choose not to wear the flag in the Middle East. “Those are my principles, and if you don't like them... well, I have others.”
Privacy
ProPublica: The public media company obtained tax records of a dozen billionaires and proceeded to show that said billionaires do not pay taxes on their unrealized capital gains. They made this out to be a big deal by dividing taxes paid by estimated increase in wealth (from a separate source) and called it “effective tax rate”. While there was much wrong with the analysis, what I think is most interesting is that the mainstream media is not talking about the privacy implications. Another example about how concerns about privacy is not really about privacy - it is about using the privacy as a stick to attack specific targets.
Fake Encryption: The FBI pretended to sell encrypted phones to criminals, but then tracked everything that was done on those phones. Technology only takes you so far… Related: The attackers in the Electronics Arts breach bought access to their internal Slack channel, and then asked engineers (through Slack) to turn off 2FA.
China: A good article on the employee surveillance tools being used in China.
Analytics and Prediction
Citations: Papers that don’t replicate do get MORE citations than papers that dp - even after the papers have failed to replicate. The rich get richer - even if they are proven to be frauds. This is the power of awareness.
Unforgiven: Notes from Clint Eastwood’s script reader. No one knows anything.
AI/GPT-3/Machine Learning
General AI: Andy Jones at LessWrong argues that we may have the technology right now to make an AI that is “generally intelligent”. He argues that taking GPT-3 and powering it up by ~two orders of magnitude may create an AI that could do just about anything (i.e., “Please write a novel in the style of Charles Dickens”). He goes into detail about how the two-orders-of-magnitude may be obtainable within months - it is just a matter of a few billion dollars into the project.
Compose.io is a browser extension that uses GPT-3 to make predictive text suggestions. It is free. The premium (paid) version will learn your specific style - so you could just start paragraphs and allow the tool to complete them for you. If you use my referral link we both move closer to the front of the waitlist.
Terminator: Cool visual on how effects artists added details in an otherwise mundane scene in the latest Terminator movie (including reflections in glasses)
Music: OpenAI (makers of GPT-3) now have tools that can create music in the style of any selected artist singing about any selected topic. Here are some new songs in the style of Elvis Presley.
COVID and the New World Order
Start-up CEOs: Andreesen Horowitz surveyed the CEOs of the companies they have funded to understand what the post-covid back to work plans looked like. 2/3rds said “hybrid”. More details at the link.
Project Starline: Google has built a technology that allows for remote conversations that feel like you are in the same room. It requires massive investment in hardware, but the effect is to create “magic windows” to anywhere else with a unit. They are being built into Google locations now.
Careers
Career Inheritance: The NYTs ranks careers by how likely one f your parents had the same occupation. Unsurprisingly children of doctors and lawyers are 19x and 27x more likely to have the same job as adults, but the most over-indexing occupations tend to be blue-collar work: Children of Textile Machine Operators are 415x more likely to grow up to be operating their own textile machines. Other jobs that are likely to be passed down through the generations include Boilermakers (275x), fichers (275x), drywall installers (136x), cabinet makers (127x). Another example of the power of awareness.
Marketing roles: I have had a few CMO roles come across my plate in the last couple of weeks and at least one Marketing Director role. The market is hot. If you are looking right now ping me with what you are looking for and I am happy to make introductions.
Fun
Star Trek: Merkel is in trouble…
Flying Cars: If you have not read it already, I highly recommend “Where is my Flying Car”, an excellent book that talks about (among other things) why we don’t have flying cars today (it was not due to it being impossible technology). Now it looks like we might actually have flying cars in the near future. The future comes at you fast!
Video as Photography: The new Canon R3 records video at 30 frames/second, “in effect, you can now record video at photo quality and pick any frame you want”.
Social Media and Mental Health: Oxford study finds no link between teens, technology and mental health.
Fertility: It decreased by income until $200K/year, then starts increasing again. The most fertile demographic in America are those with income >$1MMyear. Children are the ultimate luxury good.
Keep it simple,
Edward