Business Decline in 2026, at the Start of the AI Era

2026 will probably be a year when quite a few small and medium-sized businesses, especially small businesses that rely on online marketing, run into some major problems with marketing and sales.

 

This essentially comes from two directions: one indirect and one direct.

 

This article is about observations. I’m not saying this is definitely how things work. But from what I can see right now, things seem to be leaning this way.

 

First, when we talk about organic social traffic in Taiwan, we’re generally talking about Meta’s platforms, whether that’s Facebook, Instagram, or Threads.

 

Otherwise, we’re talking about Google’s platforms: Google Search and YouTube.

 

All of these platforms are facing challenges from AI. What exactly is the problem?

 

First, Facebook itself is dealing with huge amounts of AI junk and bots. So right now, Facebook’s crackdown on automated AI posting is pretty obvious.

 

If you try posting the same content, the first thing you’ll find is that to automate your posts through Facebook’s API, without using some other add-on, you have to use a Facebook Page.

 

The problem is that Pages have another characteristic: basically, Facebook wants to make money. And to make money, Facebook wants you to run ads.

 

So the reason Pages can publish through the API is that Facebook wants you to use it to publish posts and run ads, making money for Facebook.

 

Facebook makes it convenient for business owners to post text and images, then pay to promote them, because it wants to make money. That’s essentially how Facebook works.

 

So if you use the API without making Facebook any money, post piles of junk that create problems for Facebook, and still dream of getting lots of exposure from it, that’s unrealistic right now.

 

Things used to be a little less clear-cut. Of course, posting through the API used to mean more or less the same thing, too.

 

At this point, though, even if you don’t post through the API, posts with sales-oriented content usually won’t do very well.

 

The way posts were judged used to be more mechanical. Now AI can read your posts. Just like when you use AI to write, it all happens very quickly.

 

Facebook develops its own AI. So if Facebook uses AI to read what your post is saying as soon as you publish it, wouldn’t it be very easy for Facebook to tell that you’re selling something?

 

If you’re selling something, you’re using up Facebook’s traffic. Is Facebook going to let you do that for free? That’s pretty unlikely.

 

Sure, you can point to plenty of exceptions. But look at it from Facebook’s perspective: after all the work of bringing people together and gathering that traffic, wouldn’t Facebook want to package it up and sell it, instead of letting other people freeload?

 

So whichever Meta platform you’re on, the platforms under Meta all share this characteristic.

 

Google works more or less the same way. The problems it faces fall into two parts: Google Search and YouTube.

 

The problem Google Search faces is that websites of any kind may no longer serve much purpose, especially those built around knowledge.

 

Generally, we use Google Search to find information, though we might also be looking for products. But AI has already crawled it all for you. What AI itself does isn’t actually all that different from what Google Search does.

 

And if you’re looking for information, AI has already consumed huge amounts of it and built its own database. History, business strategies, whatever it is—AI has already consumed it.

 

So basically, if that route is faster, users are more likely to head toward AI.

 

You end up in a situation where a lot of the traffic you get through Google Search is actually just bots crawling your site.

 

YouTube is a video platform, so naturally, the problem it faces is automated video content.

 

What’s becoming fairly clear is that video platforms tend to favor content featuring real people. You need to show your face and let people know who you are.

 

But that doesn’t mean your content will definitely get recommended just because you’re a real person. The content itself matters, too.

 

So until AI can completely replace human performances, those performances still have something of a moat. This will lead a lot of people to sell this way.

 

But YouTube has its own characteristics. Sales on YouTube aren’t very direct. Even 100,000 page views might not bring in even one or two orders.

 

The thing about YouTube is that people get to know you, and they can keep knowing you for a long time. Especially if you publish long videos: if people are willing to watch, that shows they have a certain level of trust in you, because they’re listening to you talk for so long.

 

That’s a little different from short videos. With short videos, it’s basically impulse buying. Buying is mostly impulsive anyway.

 

After all, we live in a time when, broadly speaking, people aren’t going to starve. That doesn’t mean nobody actually starves. It means throwing food away is more common now, while having no food is less common.

 

That’s the kind of society we’ve grown into, especially as social welfare keeps improving.

 

Take 7-Eleven’s i珍食, its near-expiry food discount program. Everyone here is familiar with it.

 

For example, you could wear UNIQLO, NET, a Taiwanese clothing chain, or clothes from a night market stall. So why buy something more expensive?

 

A more expensive product isn’t necessarily a better product. It might simply cost more to make. And that higher cost might come from the company making it being smaller.

 

So buying these days is mostly impulsive. The reason is that big companies can drive prices down and produce goods in huge quantities.

 

We actually have a surplus of products already—more than society as a whole needs.

 

Who actually doesn’t have enough clothes at home? You just don’t like them anymore, right? You’re always after something new and tired of the old. You have to eat wagyu; frozen pork slices just won’t do.

 

Do you really have to travel abroad? Can’t you get some sun at a riverside park?

 

So when you’re promoting purchases—or, simply put, running ads—you’re really trying to keep reaching customers until they feel like they might need the thing you’re selling.

 

The problem is that these two big companies, Meta and Google, are both facing some fundamental problems of their own right now.

 

Simply put, AI has become another entry point to the internet.

 

For many people, the first thing they do online isn’t go looking for something. If they have a question, they ask AI. If they don’t know what to do, they let big data and the system’s algorithms recommend something. This new AI entry point hasn’t yet made the shift to advertising. The companies running it are hesitant to recommend ads, because there still isn’t a winner.

 

So no company dares to start aggressively stuffing it with ads. That would push users over to its competitors.

 

You end up wanting to keep reaching users, while those users have gone somewhere your ads can’t reach them and are spending a lot of time there.

 

The platforms you used to rely on already have fewer users, but more content. That’s the problem right now.

 

So if you want to reach real users at this point, ads may currently be the only solution.

 

The reason is that platforms have gained a lot of publishers who don’t pay, while losing users. What the platforms want most is ad revenue.

 

Of course, this is all an assessment of where things might go. But right now, it does look quite a lot like things are heading in this direction.

 

Caught up in the AI arms race, the major platforms don’t dare stop. Everyone is afraid of becoming someone watched by Skynet, instead of becoming Skynet themselves.

 

So in this era, if you don’t have enough profit to pay for ads, you could find yourself at a serious disadvantage.