Over the last few years, the way website content supports business goals has changed. Blog posts, glossaries, guides and other top-of-funnel content no longer earn clicks or bring visitors to the page. Buyers do not use thought leadership pieces or old blog posts to get answers any more. AI Overviews and AI platforms are where the journey starts now.
Meanwhile, rankings slip. Outdated tactics and hundreds of now-harmful articles compete with each other for the same keywords and queries. They end up feeding the generated answer instead of the page you wanted people to land on.
So why publish? What is the point? Is there a strategy here, or are blogs a thing of the past?
The answer may sound too simple. The purpose of a blog in 2026 is to publish when you have something worth publishing. It does not need to be a daily activity for its own sake.
It needs to be:
Every piece has a function. Every piece plays a role in turning a lead into a conversion.
Find what people search for, write about it, rank, get the click, repeat. It worked for over a decade. For most content teams, running the Google playbook was the whole job.
One assumption held it together. Google had to send the searcher somewhere to get the answer. Someone wanted to know how a thing worked, or which option was better, and Google pointed them at a page. The page got the visit. Some visits became leads. Most did not. But the mechanism held.
The model was built on that click. The assumption is gone.
Pew Research tracked what people actually did in their browsers. When Google showed an AI summary, 8% of visits ended in a click on a normal search result. Without an AI summary, 15% did, almost twice as many. When the AI summary listed its sources, 1% of visits clicked one (Pew Research Center, July 2025).
Ahrefs measured what that does to the page sitting at the top of the results. On informational searches, the first result now gets 58% fewer clicks when an AI Overview appears above it. The same test eight months earlier put that at 34.5% (Ahrefs, February 2026). The number is getting worse, not settling.
Both studies measure Google, because Google is the only one of these you can measure from the outside. ChatGPT and Perplexity do not publish clickthrough data. But an assistant that answers inside the chat window has no reason to send anyone anywhere.
"We should publish four posts a month to grow organic traffic" is no longer a strategy. It is a habit.
When traffic falls, most teams reach for one of two answers.
The first is to publish more. AI made writing nearly free, so the output doubles and the quality stays where it was. The trouble is that everyone's output doubled at the same time. Everyone got faster and almost nobody moved.
The second is to rename the problem. Call it AEO, add some FAQ schema, and carry on writing the same articles.
Neither one helps. The first one hurts.
Every post you publish competes with your own website.
A product page is built to rank for a term that matters to the business. Then a blog post covers the same ground, because the calendar said Tuesday. Now two of your own pages argue for the same meaning. Google and the AI platforms have to pick one, and they do not always pick the page you sell from.
You did not add reach. You split it.
We nearly did this to ourselves in July, and the decision that stopped it is one most companies never make.
Our pricing page could have gone after "seo agency pricing". Roughly 2,400 searches a month, difficulty 40. Tempting. Then we looked at who was actually winning that search, and it was benchmark articles and buying guides. Vendor pages only showed up there when they read like explainers, and that is not what a pricing page is for. A pricing page exists to convert.
So we gave the term to a separate article and left the pricing page to the people searching for our pricing by name. One term, one page, chosen deliberately. That is the whole job of a keyword universe.
That choice is only possible if someone is making it. In most companies nobody is. The calendar decides, and the calendar cannot see what is already ranking.
Wasted effort would be survivable. This is worse. It is self-inflicted damage that compounds quietly for years before anyone traces it back.
A blog is a publishing system. It always was. We stopped treating it like one.
There are five good reasons to publish:
"We haven't posted this month" is not one of them.
When one of those is true, publish. You are adding something that did not exist before. Nothing of your own competes with it, and an AI platform has something real to cite.
When none of them is true, do not.
What you do, who you serve, what problems you solve, how the product works, what it costs. Those are permanent questions. They belong on permanent pages that get updated when the business changes, and they work best when you are writing to one reader instead of to a search volume.
Here is the split:
|
Website |
Blog |
|---|---|
|
Explains what the business does |
Publishes what the business has to say |
|
Represents the services, products and capabilities |
Shares ideas, lessons, evidence and arguments |
|
Answers the lasting questions every buyer asks |
Responds when there is something worth publishing |
|
Changes when the business changes |
Publishes when there is a reason to publish |
The website is not a static asset you rebuild every three years. It is the thing that has to stay true.
Instead of filling next month's calendar, go around the room and ask what anyone has learned since the last one.
Some months you will have three things worth publishing. Some months you will have none, and you publish nothing. That is a real answer, and it costs nothing. Publishing anyway does.
The blog still has a purpose in 2026. We just need to let it be a blog again.
Curious what your AI search visibility looks like? We can show you, as part of our search visibility work.