From The Trenches Part 9: Wherein Lies The Power – Losing Control

By /

Algorithms are making more of our daily choices than we think.

While academic studies quantifying the impact of microtargeting have come mostly from the political ad space and have mixed results, the growth of businesses based almost entirely on microtargeting, namely Netflix and Spotify, demonstrate the power of this mechanism.

Microtargeting means that through extremely nuanced data gathered on our daily behaviour, businesses can deduce our preferences and tailor the content they send to us to match what they think we would like. The more data they collect, the more accurately they can predict when we might need a given product or service and therefore send us recommendations.

Doesn’t This Require a Ton of Compute Power?

That’s the point, isn’t it? Think of deep learning as data running through a maze which spits out insights. In microtargeting, what comes out are the definitions of groups that behave alike.  Deep learning algorithms can identify combinations of characteristics that would take a very long time for humans to think of because they can analyze multiple characteristics at once, mixing and matching them in different ways.

But the bigger and more complex this maze gets, the more compute businesses need to support it. And that means, yep, bigger data budgets. The question is, for companies whose revenue runs in the single digit billions or even less than a billion per annum, how is it possible to afford all this infrastructure? Is there even enough data, if your business covers purchases that people make only a couple of times a year (like travel) or a few times in a lifetime (like housing) to predict purchase intent accurately?

That’s where paid ads come in as a component of growth and acquisition marketing strategy. Essentially, we’re plugging into platforms that get more signals than we could about what consumers are thinking about, in real-time.

Advertisers’ Ad Spend Optimization Is Platforms Revenue Management

Advertising makes up a huge proportion of platforms’ revenue. Meta is almost entirely dependent on ads, with ad revenue making up 98% of its total revenue in 2024. Alphabet is not far behind, with over 75% of 2024 revenue coming from its Google Search and YouTube ad products. And for ByteDance, ad revenue took up about 60% of total 2024 revenue.

What does this mean? If it benefits companies to use AI for yield and revenue maximization, platforms have the incentive to create “black boxes” for bidding, because advertisers’ ad spend is their revenue.

And when that happens, it shifts power away from advertisers, who get less bidding and targeting levers because platforms want to forecast and maximize the predicted ad revenue per consumer (across multiple product and service providers) using their own data signals.

The result? Power goes to whomever owns as much of the consumer’s data as possible, by driving interactions on their platform. Google, Meta, TikTok, and now ChatGPT are in a race to own as much of the consumer’s engagement as possible, so they can own a bigger share of the consumer’s purchase moments.

Therein, The Rich Get Richer

“Power and Progress” by Professors Daron Acemoglu and Simon Johnson of MIT (2024 Nobel Memorial Prize winners for Economics, and “Reshuffle” by Sangeet Paul Choudary (2025 Thinkers50 Strategy Award winner) touch on the principle that in history, gains from technological advancement don’t necessarily spread fairly into society. In fact, without specific actions to prevent it, the natural outcome would be that technology makes the rich richer.

We’re seeing and living this in the paid ad industry right now. When channel strategy, consumer visibility, and channel practitioners’ career viability depend intensely on which platform owns more of the consumer’s time and data, as well as the platform’s own ad product strategy, these are volatile times.

In a November 2025 podcast with Tyler Cowen, Sam Altman said that ads are “something [Open AI will] try at some point”, but “[he does] not think it is [their] biggest revenue opportunity”. He admitted that he “[has] no idea” what such ads would look like, but indicated that paid bidding would not supersede relevance, in the interest of the customer experience. More tellingly, he highlighted that he expects “margins are going to go dramatically down on most goods and services, including things like hotel bookings”.  That’s a rather direct statement about the level of power – and share of wealth – that he expects e-commerce advertisers to have, in the world of agentic commerce.

How Do Brands Bring Back Leverage?

With the paid space becoming increasingly squeezed by margins and leaving advertisers at the mercy of platforms, brands’ areas of leverage are narrowing.

So experiential marketing, social marketing and the creator economy have become trendy (data in reference section). To own the consumer, brands must get to them before they go to Google or ChatGPT and be memorable enough to lock them in before they turn to AI. Not every brand has this luxury, as it depends on how differentiated the product can be.

In Part 10, I’ll talk about how all marketing is becoming performance marketing, and how the process of media buying might come full circle (sort of). Stay tuned!

References:

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Scroll to Top