LLMs Could Slow Down the Attention Economy and Reduce Ad Inventory

Access to information got partly hindered through the attention economy, here are 10 reasons why!!! The last one will blow your mind!!!

Oups … sorry, let me try again.

Access to information got partly hindered through the attention economy, slowing down the user to access content or even luring him to irrelevant content. The name of the game is to get as many eyeballs as possible to print ads on. LLMs seem to be shifting this game as they are becoming the entry points for much of non-entertainment information access.

Everyone knows the sensation of googling for something that should take two taps and ending up on a ten minutes YouTube videos or accepting 14 pop-ups before hitting a paywall. Part of the speed improvements offered by a system like ChatGPT are simply about cutting the revenue-milking layer that has been added on top of relevant information, part of it are linked to its superior capacity to skim through content.

If usage shifts to using LLMs based systems for information access, it means that the current information holders will see there userbase shrink, and so will opportunities to display ads.

LLMs and ads

The current LLMs model seems to be a freemium / subscription based approach. It might be a trend set by investors addicted recurring revenues, or it might be a lasting thing.

The rise of “small” efficient models like DeepSeek shifts a lot the entry barriers and remove the edge that big actors were hoping to gain. It could also leave the door open for competitive open-source solutions, which could shift the created value from corporations to the actual end-users, and probably remove the window for inserting ads in the end product.

It seems however kind of unlikely that everyone will train their own model. I am sure that some people are thinking very hard about how nice it would be to have ChatGPT being trained to “think” favorably about their brand.

A world with less ad inventory

Ads are often seen as the evil side of the internet. This is an arguable position, however I would like to defend the opposite for a bit. Every starting business needs to reach an audience, Google, Facebook Meta, Reddit, Microsoft and their subsidiaries provided the economy with comparatively cheap ways to get new consumers.

With the years, all those companies are displaying a booming ARPU (average revenu per user) over the years. So either people are using those platform more and more, either despite the coating of a magical algorithm and bidding system those companies are just hiking their prices per view to be sure to meet their targets. Being a buyer of ad space, I would pick the latter option.

This also means that the advertisers are always getting less and less return on their investment. To be convinced of that, just check the average product price that you see advertised online on social platforms. I tend to think they are very expensive, the reason being that it is likely that every new customer costs between 10 and 100 €.

In a world with a strongly reduced ad inventory:

1. Ads unit price might increase, because all those players need growth to sustain their stock value and this is the easiest (and only?) move for them.
2. But at the same time, people will see fewer ads. They will have the inverse effect of ad fatigue, this could increase the efficacy of a single ads.
3. It is unlikely that 2 will compensate for 1, so the value of investment in advertising will decrease.
4. If 3 is correct, the low margins actors will become unprofitable and the moats protecting incumbents will grow.

If that was an interesting read, I can recommend: Who Will Teach the Teachers