First you convince the filter. The human comes second.

I barely open the first page of Google anymore, let alone the second.

There is obscenely too much content now

You can curse the models or bless them — it changes nothing, the market has already shifted and you can't un-grind that meat. The cost of producing content fell to cents, and everything that previously died from laziness or lack of budget is now in the feed.

Attention doesn't come out of a button

Meanwhile the cost of a touchpoint keeps rising everywhere. Attention, unlike content, can't be manufactured on demand: a person still has the same 24 hours, the same fatigue, and the same urge to close the tab and go live in the woods. That well was pumped dry long before all this — and now many times more people are crowding around it.

The industry's answer is, of course, more volume

Touchpoints get pricier, so let's generate five times more content, put content factories on a conveyor and stuff every crack with identical AI slop. It's already bad manners. And the slop is getting hard to tell from the real thing — plenty of people are using that very skilfully.

But the interesting part is one floor up

A new layer has settled on top of all that content: models that filter it on the user's behalf. People scroll results and piece opinions together less and less. They ask, and get a finished answer with the sources already picked for them. I catch myself doing it: googling less every month, asking more. Recommendation algorithms were the first bell, years ago. Now there's an intermediary between you and your audience, and it decides whether you're worth showing at all.

The filter plays by different rules

So the fight isn't for a person to click your headline. It's for a machine to consider you a sane source and retell you in its own words. And this intermediary could not care less about clickbait or how cleverly you tease in the first line. It looks at whether there is structure under the words, whether there are facts and specifics, and whether you're actually present in the places it pulls data from — reviews, forums, Reddit, ratings, live discussions.

This is what people now call answer engine optimization or LLM SEO: the work isn't chasing a blue link, it's AI search optimization — making sure the machine can parse you, trust you and quote you.

Funny turn, by the way: the industry spent years perfecting the art of manipulating a living human, and the main reader turned out to be a thing those tricks don't work on at all.

What it changes in practice

If the decision about what a person sees is increasingly made by a machine, then touchpoints have to be systematic and continuous — closer to marketing automation than to a campaign — not a heroic manual push once a quarter when somebody finally gets to it.

That's why the product moved past research into automating touchpoints and traffic: SEO Traffic Growth for being findable in search and in AI answers, and Social Traffic Growth for distribution that isn't rented from anyone. It's in beta, being built live with one corporate client, on their real problems and their real rakes. I won't pretend all the answers are in hand — they aren't — but the direction feels right.

Bottom line

The substance hasn't changed. The addressee has. First you convince the filter, and only then the human.