Meta

    Turns out Meta has a hidden quality score for ads.

    According to my Meta rep (who can’t see the actual score), the algorithm analyzes over 500 aspects of each creative to determine how performant it is / will be and that drives placement opportunity.

    Right now, Meta wants Reels.

    Bloomberg: Baidu Readies AI Smart Glasses to Rival Meta’s Ray-Bans

    Meta’s are becoming computers (thanks AI!). And the true vision is in prototype.

    Apple is joining in.

    The VR players are pivoting. AR glasses are the path. Ambient computing.

    Think Snap wants some credit? And maybe some commission?

    Now you can ignore everything I put in the big election spending post from yesterday.

    We will pause ads relating to US elections from serving in the US after the last polls close on election day, November 5th, 2024. This will include US Election Ads as well as ads that refer to US elections, their processes or outcomes.
    The restriction period for ads about social issues, elections or politics is being extended until later this week.

    Pressure released.

    Apple Inc. is exploring a push into smart glasses with an internal study of products currently on the market, setting the stage for the company to follow Meta Platforms Inc. into an increasingly popular category.

    I’ve long felt that glasses make the most sense as the next popular computing form factor. Augmented by headphones.

    Both form factors are well established by now. They don’t require new behavior like VR.

    Computational dispersal Ambient computing

    Election Campaign Spending Crowds Out Your Messaging

    I’ve been harping about election campaign spending and its impact on marketing expectations this year for a bit now. So here’s an Election Day post running through it all again. Go vote!

    2024 is projected to be a record-breaking year, surpassing $12 billion in political ad spend.

    Seventy percent of that windfall will be spent after Labor Day

    The Harris campaign has instead focused its digital spending on larger platforms like Google and Meta, where it has spent over $280mn this year. The vast amount of political spending on those platforms — already more than $1.5bn
    For context, in the second half of October, political parties spent well over $2.5 million per day on YouTube. For Meta, it’s closer to $1 million per day.

    Why won’t I shut up about this?

    Because this level of spend puts intense pressure on the ad auctions and inventories of the utilized platforms. These are mostly net-new dollars flowing in at a rapid clip. For most brands this means higher ad costs and lower return reducing the efficiency of their spend.

    Emarketer curated some reporting that includes:

    • brands reporting ad costs increasing 2-3x in the final weeks of the election
    • surging TV ad costs, especially in swing states

    Of course, the impact on marketing budgets is just the beginning.

    This messaging overload overwhelms consumers, raises anxiety levels, and increases ad fatigue (who else is sick of all the election messaging you get bombarded with?). Which can all lead to a “pre-election slump” in consumer spending. Of course, this theoretical slump is not universal—I bet alcohol sales and the like are doing just fine.

    But the bigger the price tag, the bigger the impact.

    In times of uncertainty, people sit on their cash. They purchase discount or off-brand. They hold off on large luxury purchases made in celebration or confidence.

    Election Day itself is no exception, more the final culmination of the building gravitational pull. A black hole that will collapse into itself tomorrow (right? please?!).

    the election is fraying the nerves of the electorate, with nearly 70% of US adults calling it a significant source of stress

    According to a recent Ipsos poll of roughly 1,000 US adults, 47% said election stress is causing them to spend less and save more

    Meanwhile, the scale of an Election Day productivity slowdown could be staggering. An analysis by Challenger, Gray & Christmas estimates that productivity losses could reach $3.5 billion per hour
    Your marketing and messaging is like a fire, it needs a steady source of fuel to keep burning—to act as a beacon to the people it's for. The election is a vacuum, it sucks all the oxygen (and attention) out, making it all but impossible for that fire to keep burning without any changes.

    My team is probably tired of hearing me talk about this stuff, but context is important. Especially when talking to clients.

    For context, in the second half of October, political parties spent well over $2.5 million per day on YouTube. For Meta, it’s closer to $1 million per day.

    November 6 will be a pressure release. In more ways than one.

    Some of your ads feel more expensive lately?

    This might be why:

    The Harris campaign has instead focused its digital spending on larger platforms like Google and Meta, where it has spent over $280mn this year. The vast amount of political spending on those platforms — already more than $1.5bn

    I’m not sure how they’re defining “this year” since the Harris campaign technically started in July when Biden suspended his campaign.

    $280 mil in 3 months?

    Either way, a lot of money putting pressure on the auction algorithms.

    via Financial Times

    Meta is developing its AI search engine (and apparently has the “most-used AI assistant in the world”).

    Now OpenAI is fully launching ChatGPT search

    ChatGPT can now search the web in a much better way than before. You can get fast, timely answers with links to relevant web sources, which you would have previously needed to go to a search engine for. This blends the benefits of a natural language interface with the value of up-to-date sports scores, news, stock quotes, and more.

    A conversation about the weather forecast for Positano, Italy, on November 2-3, 2024, showing mild temperatures and rain. The user then asks for dinner recommendations in Positano on Friday night, with responses listing local restaurants.

    Search is now about your preferred interface, experience, and engine.

    Meta is developing an AI search engine, to be embedded in its Meta AI chatbot. The company has reportedly been indexing the web for at least eight months. 

    Meta AI currently uses Google and Bing’s search engines when it fetches users answers about current events, financial markets, and sports.

    This highlights a few things:

    • Search is no longer a platform, it’s a feature.
    • Users want answers, not necessarily options.
    • Google will not be replaced by a copycat. It will be chipped away at by new, novel alternatives and user dispersion across other channels. (and maybe by antitrust)
    • Meta continues to reduce reliance on other companies wherever it can. The true realization of this will be a hardware platform, likely in the XR space.
    • For brands / creators, having your content on the web—preferably on a platform you own—where it can be crawled is once again the path to relevance. (see: Gwern)

    Meta’s plan for growth among younger demos:

    • Discovery
    • Utility

    Surface content like an entertainment app (social media is basically TV now).

    Do the classic “Meta experience upgrade” on things like marketplaces, forums (Groups), and events.

    Network effect + massive infrastructure = advantage

    Ben Thompson on Meta:

    Market = time & attention

    Differentiation = horizontal services that capture more time and attention than anyone

    This is what makes the Zuck’s Bucks machine run.

    It also means whatever captures attention and increases time spent gets rewarded.

    Imagine this but in a pair of glasses like Meta’s Orion:

    Sensors will only proliferate (ambient computing!) creating even richer augmented experiences.

    Meta is working on “New Brand [Safety &] Suitability Controls”.

    The big one I know some clients will love:

    We’re testing the ability for businesses to turn off comments on ads

    There is research that suggests ads with comments perform better (social proof?), but some industries and brands attract a lot of negative comments and/or don’t have the bandwidth to actively moderate and engage. Control on this capability would be nice.

    B2B advertisers are increasingly turning to social to spread the word.

    Unsurprisingly, LinkedIn is #1. Followed by Meta.

    YouTube is riding the growing B2B video wave.

    via EMARKETER

    a VR headset is about withdrawing from interactions, whereas augmented reality is obviously about adding that layer onto existing reality. And so in that way, I think it's a lot more similar to a smartphone. So if you're trying to figure out what does the next platform look like, VR looks very different from that and AR looks much closer to the existing interactions we have while using our smartphones.

    Why I’ve been on the AR bandwagon for years.

    Meta just casually announced an AI-powered AR glasses prototype.

    Ambient computing accelerates.

    Meta reportedly rolled out an ad algorithm update in August. It caused a lot of volatility, but includes some nice sounding features.

    • Conversion Value Rules

    You can assign certain audience profiles OR conversion types higher values

    • Incremental Conversions Attribution

    measure which conversions would not have happened w/o the ad being shown (true impact) AND optimize for these types of conversions

    • Third-party Analytics Integration
    • Cross-Channel User Journeys

    more emphasis on cross-channel journeys to drive incremental conversions.

    Social Savannah’s top…

    Black Friday strategies / hooks for Meta:

    1. Unboxing reaction
    2. Price / discounts & deals

    KPI to measure creative performance:

    • Click-to-purchase ratio
    • Hold rate (watch length)
    • Hook rate
    • Spend (aka the algorithm’s love language)

    Meta wants your holiday ad dollars, so it’s rolled out:

    • More prominent promo callouts
    • Expanded reminder ad capabilities
    • Site links for all (these do get used based on my tests)
    • More features to get people taking action at a physical location
    Posts with a lot of comments look to be more heavily weighted when it comes to what shows up in your Threads feed. That’s over re-shares and Likes, with the experiment seemingly suggesting that the Threads algorithm is geared towards incentivizing discussion over everything else.

    This makes sense at first glance because commenting is the highest form of public engagement. And engagement is what Meta’s other apps struggle with.

    So Meta wants engagement and the algorithms are primed to reward the actions the platform wants.

    But this article raises a great point I hadn’t thought about—about what Meta really wants.

    ROBOT FOOD!

    What better input to train your social platform’s LLM on than public user conversations?

Older Posts →