Rabbbits Weeekly: Same Names, Different Day

Meta's Plan for the (Near) Future™ and TikTok's game of limbo. Plus, headlines and things about the æconomy, algorithms, ads, audiences, analytics, and attack vectors.

Ed note: the combination of being deep in Google Analytics 4 last week (we are officially in the last year of Universal Analytics (also why this is late)) and the expectation that this week will be light on news, there may not be an issue on Friday. Unless you have questions about GA4. In which case, reply back to this and ask away. Let’s see if I can do a mailbag issue, further masquerading as a media outlet.

Zuck’s Crystal Ball

Meta has outlined a three step approach to becoming a “discovery engine”:

1. Become TikTok (Reels everywhere!)

2. Become TikTok (rebuild the recommendation algorithm !)

3. Become iMessage (don’t text that funny video to your friends, Meta Message it (aka we’re putting Messenger back in the apps))

This is from the “leaked” memo that has been making headlines lately. Except it basically just says everything Zuck did in a public Facebook post back in April (like the date on this memo). 

Historically, Facebook has taken an entity-centric approach to discovery. We help you connect with the friends, groups, and pages you care about most.

…

Today this is changing. Social media products - including our own - are delivering value by investing more in discovery engines that help people find and enjoy interesting content regardless of whether it was produced by someone you’re connected to or not. 

Why these priorities? Young Adults (capitalization theirs)

Q: Are we still trying to build lots of different products and services and grow tabs?

A: The short answer is that we’re not trying to build a lot of new tabs right now - we’re focused on making our core products better and more integrated. We want our tabs to complement the content people discover, enjoy, and share on our Home screen. Our personalized tabs system is still very valuable, but we’re changing our tab strategy to put more focus on tabs that complement Feed by providing unique experiences that help people find or create content, and that are incremental to our overall app.

Zuck+co’s efforts to make discrimination in hosing (and credit and employment) ads harder / impossible are still a work in progress. The company announced a new settlement with HUD (Dept of Housing & Urban Development) that will "make sure the audience that ends up seeing a housing ad more closely reflects the eligible targeted audience for that ad." Which smells a lot like "we got caught not doing what we said we were going to do, again, so this time we'll do it for real." But that may be the Facebook cynic in me.

The part that really grabbed my attention was the announcement that Special Ads Audiences will be sunset. These were supposed to replace Lookalike Audiences for these special ad groups, by removing the sensitive data points (age, gender, ethnicity) from consideration. This change is a signal that, when it comes to machine learning algorithms, it's not that easy. With the amount of data Meta has (whether or not they know where it all is or what they're doing with it), who knows how many implicit data points there are for every explicit one. Just because a human can't spot the pattern doesn't mean a machine won't.

This is, after all, what machine learning and AI systems really are: turbocharged pattern matching systems at a scale of, roughly, 100 (1,000? 10,000?) interns. That don't need coffee. Or sleep. Or bathroom breaks. (They do, however, need a very large room full of very expensive GPUs and other equipment.)

I was also reminded of a question I fielded years ago about AI fairness. It boiled down to whether or not the models could be inherently biased. At the time I answered that it was all in the data. The systems found patterns and scaled them up. If your company has a discriminatory hiring history and turns to AI to be more fair, loading in the data of that discrimination to a system to train it is not going to solve the problem. You'll get more of the same, you'll just get it quicker. And you'll get to point to a computer as some sort of deniability.

While I still think it ultimately comes back to the data, these models won't always operate the way we anticipate based on how they are designed. (Based on my experience, there is not a lot of explicit instruction given in the coding. It's more trial-and-error until you get the results you expect from the test data, then turn it loose.) Algorithms ruthlessly optimize for their programmed goal, and they don't have an unspoken moral or ethical code guiding these optimizations. 

So you may remove a few data points from the feed, but that won't rewire the system automagically. It will just turn to the next set of data points that mirror the results of the now absent ones and continue right along. 

Their new plan to remain relevant and attract creators? Bribe them! I mean, help them monetize their content.

Not one to let someone else's idea go to waste, Instagram is testing a new feature called Notes. You know, like the one Twitter just announced. The concept is different though. It's like a text-only, self-destructing story that accounts can share with their close friends or accounts that follow back. 60 characters max, so start practicing your Hemingway impersonation. 

The subscriber-only groups feature makes more sense now, they're going after Discord. 

FacePlace updated their privacy policy to assure users they are not and will not sell their data. Which is a nifty rhetorical device that Congress gave them with inept lines of questioning. Meta doesn't sell data, they rent access to it in the form of targeted advertising.

And Facebook Pay is now Meta Pay. That is all. (Seriously, this is what they release announcements for now. The official announcement has "takeaways" that takes 34 words to say what I said before I started this parenthetical that increased my word count.)

TikTok Takedown?

TikTok remains the darling of social media. Except, it’s not a social network, according to themselves. And this is why they aren't worried about Meta. The argument does make sense based on what the platforms have prioritized to date and how difficult it has been for Zuck to remain top dog.

The Trend Machine™ is forecast to triple revenue this year (to $12b, we’re not talking peanuts). And that’s why Meta is terrified of it.

It’ll now play nice with EU rules geared towards consumer protection. Some seem pretty basic, some seem like major headaches in the making ("If a user has more than 10,000 followers, their videos are reviewed by TikTok against its Branded Content Policy and Community Guidelines to ensure that the content is appropriate;").

And it’s moved all US user data to Oracle servers stateside to address privacy concerns (what with sending US data to China and all). Except maybe it won't actually make a difference. Whoops. 

But it’s not all roses and unicorns, the FCC wants TikTok out of app stores. I’m not so sure about the “surveillance tool” part of the claim, but I’m also not eager to hand over data to the Chinese government (assuming that’s actually possible). Could be some Friday fireworks to reprise the holiday. I’m sure this will all be riveting legal theater that likely ends with nothing much happening. For the conspiratorially minded, the timing is interesting.

Æconomy

Fireworks exploded over the weekend, but maybe fewer than years past as people planned on spending less on the holiday. Lots of numbers and generational breakdowns in this report, but this is the most interesting to me:

Gen Z is more than twice as likely to favor Costco.com and millennials are 33% more likely to shop at Target.com.

US GDP dropped for the first time since the initial shutdown way back in 2020, which is, honestly, kind of amazing. Consumer spending, especially on large items, scaled back, unsurprisingly. When people are shopping, they are spending more time looking at reviews and price hunting.

Ad spending will keep on growing, but at a slower rate (only single digits) with all the ::waves hands:: stuff happening. I’ll let Vincent from MAGNA (the peeps behind the report) take over:

Most of the headwinds facing the advertising market this year were expected: economic landing following a red hot 2021, continued supply issues generating inflation, and mounting privacy restrictions slowing down the growth of digital ad formats. On top of that, the war In Ukraine now exacerbates inflation and economic uncertainty. 

Sales are slowing for the home goods sector, which isn't surprising. Summer could be slow with the travel boom and real estate slow down. But I'm interested to see if people who were previously looking for a new home might revisit their current home (a la early pandemic) and ramp this sector back up.

Algorithms

It’s been a weird week (or so) for Pinterest. Just anecdotally, it feels like performance is declining in a way that suggests the Apple privacy crows might be coming home to roost. Factually, they’re getting sued. A (former) friend of Ben Silbermann, one of the founders, is claiming they helped create some of the functionality and a marketing plan to grow the user base but never got paid despite a promise otherwise.

And now Ben is stepping down as CEO and they created the role of Executive Chairman for him on the board. His replacement has previous experience at Google, PayPal, and Venmo. Why was he chosen? Shopping and payments. The future of Pinterest is commerce.

The shopping revolution will be live-streamed. Apparently. App download growth in this segment knocked on the Door of Doubling last month. It's QVC for a new era as entertainment fractures further. Cut the cord and open an app to fill the time.

Shop around the clock with the new TikTok Shop. The Trend Machine is going Insta with the addition of a dedicated Shop tab. One original idea begets a thousand stolen ones in the platform wars.

Google: it's smarter than us. Or so it continues to imply / try to prove. The internet engine may or may be injecting words from searches into product titles in that shopping feed you sometimes see at the top of your results. I don't know. I guess this is just another reminder that you can tell it whatever you want but Google is gonna Google.

Google may have a new search feature that is actually smart, smarter than any of their previous ones at least. There is some proof The Company That Owns The Alphabet™ is testing showing pros and cons in search results, without those terms being used on the page. More questions than answers at this point, but I think this is our first example of MUM in action.

Walmart has announced a duet in AR. First, virtual try on for furniture in your space (like IKEA did a really long time ago). Second, shopping personalization in-store via your phone camera. Point it at the shelves and get a unique experience based on your preferences.

Sounds like we should anticipate Apple's first XR (hybrid of augmented and virtual reality) headset in January. Perhaps with pre-orders starting sometime in November? Just a wild guess, no real reason for that. This timing also comes as Meta is having to rein in their dreams to focus on their ad business and actually make money to support everything else. Maybe the Apple tracking privacy stuff was all just a ploy to hobble the XR field for their entry.

Deepfakes, coming soon to Alexa. The devices may soon allow voiceprints to customize the assistant to sound like, well, whoever you want. What could go wrong? 

The tractor trailer use case has long made the most sense as one of the first successful places for self-driving technology. Highways, infrequent lane changes, consistent speed (when possible), and a lot of terribly incentivized human behaviors amongst drivers. Now we just need fully electric rigs.

Ads

We regularly talk about delivering the right message at the right time to the right person in marketing. But the psychology goes deeper than the obvious face value of a statement like that. People associate “knowledge” (in quotes so we know it’s not the pretentious kind) with the place and/or mood they obtained it. So it’s not just about having your message match the medium, it’s about having your marketing match the mood and location to drive future recall.

What do you do when people have unsubscribed from your email list and have fallen out of your retargeting timeline? You could try sending them a letter. Like, an actual, physical piece of mail. One Facebook Ads power user reported much success with this tactic.

The good news for marketers staring down the post-cookie landscape, contextual advertising is better. At least display ads do better when they match the context of the page they're on. I don't think this is a surprise, but current tools from the major players make really good contextual advertising harder than user targeting.

Pay for people to like you on Twitter with Branded Likes, their new ad feature that is really just a custom animation they make for you? If you give them a million dollars, that is.

The Former Fail Whale has also released Campaign Manager in an effort to make their ad platform more useful for advertisers with things like data and save buttons.

The future of influencer marketing and/or TikTok advertising might just be paying people for the stuff they were already doing: talking about what they bought from you. This isn’t really new, but the combination of platforms designed to facilitate this and the Trend Machine explosion (coupled with the currently reality that typical paid advertising really doesn’t work great on it) could take it mainstream. 

TikTok is where the hype is, but YouTube is where the ad spend is. Spending on video ads as a category grew 30% since Q4, mostly thanks to the web video OG.

And if you do a lot on YouTube (or, maybe, just run video ads) here is the King of YouTube™, MrBeast, explaining the algorithm. TL;DR: make stuff people actually want to watch. Wait, what?!

The race to become Netflix’s advertising partner is on. I wonder what this means for the rumored Roku purchase. Or what would happen to this race if the purchase happens.

Walmart is opening a new store near you: inside your TV. A partnership with Roku will allow viewers to click on an ad and buy the product via Roku Pay.

This piece from Stratechery is a look at how ATT works as a glimpse into how Apple thinks of privacy. And doubles as a handy primer on how to inerpret legal disclaimers in tech.

tracking is only when data Apple collects is linked with data from third parties for targeted advertising or measurement, or when data is shared/sold to data brokers. In other words, data that Apple collects and uses for advertising is, according to Apple, not tracking;

Wrapping cars is all the rage these days. Tesla owners do it as an alternative to a new paint job (which apparently damages the value of the car). Rideshare drivers do it to actually make some money, or at least pay for gas. Well, some gas.

Honestly, this is more legalese than I want to read right now, but if you do SMS (that means text message) marketing it might be just the right amount for you to read right now. 

Audiences

I recently walked a client through a trigger event exercise as part of our marketing strategy process and I'm convinced of the concept's value. I mentioned them last time, but Stacked Marketer circled back to the topic and shared some examples that might help illustrate the idea further.

I think I finally understand zero-party data. First-party is data you receive in a transactional way. The user gives it to you to get something from you (like an email address to create an account). Zero-party is data they willingly give that is not required for use. The example given in this podcast episode is the thumbs up in Netflix. It's not a requirement to binge more, but it is very valuable data to Netflix.

You may have the most highly curated and lovingly polished list of keywords on the face of the earth, but are any of them actually being searched by people? Here's a fun new (free!) tool to see what people are actually searching related to your keywords.

A good read on user research and what it can uncover in systems. My favorite example:

We found that, for police staff, an incorrectly positioned button on a system’s manual input form isn’t just a problem once a day. It’s something they have to work around between 20 and 30 times a day, or 150 times a week. It’s a design issue that affects the user’s full-time job. The system, and how it makes them feel, is part of their lived experience.

People want to travel. 80%+ of American adults want to travel this summer, that is. Material consumption is giving way to escapist consumption as we collectively emerge from our covid cocoon.

Of course, airlines might have something different to say about that desire.

JD (the Chinese Amazon) is getting into the brick-and-mortar game. As Ben Evans says: “There is no binary distinction between ‘online’ and ‘offline’, only a spectrum of logistics and experience.”

Analytics

Some Connected TV (any content delivery mechanism to a TV that isn't a cable box) impression metrics got the Facebook treatment, counting an impression when the TV was actually off.

If you're running TikTok ads, you can now tailor your attribution window to better match your business. 

Attack Vectors

Get rid of those browser notification pop-ups for good.

First Apple, now the UK government. Everyone is coming after Buy Now, Pay Later (BNPL) services. The Crown™ (is that a nickname for the British government?) is enacting regulations to protect consumers from taking on too much BNPL debt, because they're really just loans with fancy branding.

First Brexit, now GDPR-xit. The UK is going their own way with digital privacy regulation. Calling out the burdens it causes for businesses (especially small businesses) in the process. This is what happens when you write a regulation targeting a certain group of businesses (tech giants in this case) but enact it to apply to everyone.

The shine is fading on GDPR, leaving more of a tarnish behind. What’s that mean? Lots more privacy regulation fun ahead as no one can agree on what that should look like or how it should be enacted. Can’t wait…

Why wait for government regulation when you can voluntarily not be sketchy with location and other sensitive data?

Mozilla continues its sprint into the dark forest of privacy. The new version of Firefox includes a setting that automatically strips URL tracking parameters for platforms like Drip, Facebook, Hubspot, and more. If this push expands out to more parameters we could soon find ourselves in measurement hell.

Late last year a security vulnerability in Apache was uncovered and patched. It was known as Log4j or Log4Shell. Now, 2 US gov't cybersecurity groups have said exploits of it are still happening. Update your shit!

There's a new piece of government-grade spyware for Android. Victims are texted a bad link that tricks them into downloading the malware. So, first, don't click links you don't know the source of. Second, read this piece for a view into what's possible in spyware.

Alternates & Assorteds

Your favorite wine may soon come from a robo-vineyard thanks to Beagle, a company that retrofits common farm equipment with an AI sensor package to automate things.

The hungry caterpillar might be the answer to our plastic problem. At least when it comes to styrofoam. Turns out darkling beetle larvae love eating the stuff.

You can now recycle one concrete building into another through a carbon neutral process (with the right energy source) thanks to some engineers from Cambridge University.

For now, self-checkout is an option at some stores. Soon it may be the only option, at least at Dollar Generals. (What advertising options present themselves via a point of sale screen that customers have to pay attention to to complete their purchase?)