Yes, but not the kind of blog most companies run. Publishing weekly summaries of things anyone can already ask an AI model is now a waste of money. Publishing proof, original numbers, and real opinions still works, and it works better than it did five years ago.
We get asked this in almost every scoping call. A founder has a blog with sixty posts, traffic is flat or falling, and they want to know whether to keep going or shut it down. Our answer is usually the same. Keep the blog. Kill the content calendar.
This piece lays out the honest case for and against blogging in 2026, with real data behind both sides. We will take a clear position at the end, because hedging is not useful to anyone making a budget decision.
Because the web has far more content than search demand. Ahrefs studied roughly 14 billion pages in its Content Explorer index and published in December 2023 that 96.55% of them get zero traffic from Google. Another 1.94% get between one and ten visits a month.
That number gets quoted as doom, but we read it differently. It tells you the median blog post was never competitive in the first place. It was written to fill a slot in a calendar, not to answer a question better than the existing results.
The uncomfortable follow-up is that most of those pages were probably fine as writing. They were just unnecessary. Nobody needed a seventh article explaining what a meta description is, and Google could tell.
The easy tier disappeared. Simple informational questions now get answered inside the search page or inside a chatbot, so the traffic that used to flow to basic explainer posts has been absorbed. SparkToro found that 68.01% of United States Google searches ended without any click in the first four months of 2026.
Pew Research Center measured the click effect directly. Studying 900 United States adults across 68,879 Google searches in March 2025, it found people clicked a traditional result in 8% of visits when an AI summary appeared, compared with 15% when one did not.
So the blog post that existed to capture a definition query has lost most of its reason to exist. We covered the strategic response to this shift in our piece on zero-click search, and the short version is that you compete for the answer rather than the click.
What did not change is the value of being the source. Somebody has to be the thing the model quotes. That role pays better than it used to, because there is less competition for it than there was for generic traffic.
Longer than anyone wants to hear. Ahrefs analyzed 1 million URLs and 1.3 million keywords in a 2025 study and found only 1.74% of newly published pages reached the top 10 within a year. It also found 72.9% of pages sitting in Google's top 10 are more than three years old.
The average top-ranking page in that study was five years old, up from two years old when Ahrefs ran the same analysis in 2017. Search results have aged, and new pages have to displace incumbents that have been accumulating signals for years.
This is the single most important expectation to set with a client before they fund a blog. If someone promises search results in six weeks, they are either targeting keywords with no competition or they are lying. Our typical build runs four to eight weeks, and we say plainly that the content compounding happens on a much longer clock than the build.
Yes, and the data on this is clearer than the discourse. Ahrefs analyzed 600,000 pages from the top 20 results across 100,000 keywords in July 2025. It found 86.5% of top-ranking pages contain some AI-generated content, and the correlation between AI content share and ranking position was 0.011, effectively zero.
Ahrefs is honest about the limitation, noting that no AI detector is perfect because detectors are statistical models dealing in probabilities. We treat the finding as directional rather than precise, but the direction is unambiguous. Google is not scoring pages on whether a machine helped write them.
Google's own documentation supports this reading. Its guidance on helpful content says ranking systems prioritize content "created to benefit people, and not content that's created to manipulate search engine rankings". The question is who the content serves, not what tool typed it.
Our position is that AI is a fine drafting and research assistant and a terrible source of authority. If the substance came from a model, the page has nothing a model cannot already produce, and that is the actual failure mode.
Four kinds, in our experience. Original data that nobody else has collected. Real project results with real numbers attached. Strong opinions from people who have actually done the work. And genuinely hard how-to content where the small details matter and getting them wrong costs real money.
Google's helpful content documentation asks whether a page "provide[s] original information, reporting, research, or analysis" and whether it is "written or reviewed by an expert or enthusiast who demonstrably knows the topic well". Those two questions are a decent filter to run before you commission anything.
Trust is the part people underrate. Google names experience, expertise, authoritativeness, and trustworthiness as its quality framework, and says plainly that trust is the most important of the four. We broke that framework down in our guide to what E-E-A-T actually means, including the parts that are within your control.
Google also encourages clear authorship, asking whether it is self-evident to visitors who wrote the content and recommending accurate bylines. An anonymous content farm reads as exactly what it is.
By giving models something specific to cite. AI systems synthesize from sources, and a page with concrete numbers, named tools, and clear structure is easier to quote than a vague overview. Blog content is still the cheapest way to create citable surface area for a business.
The traffic that comes back is small but real and growing. Ahrefs studied 3,000 websites in February 2025 and found 63% of them received some AI referral traffic, while AI accounted for about 0.17% of the average site's visits. ChatGPT sent 50% of that traffic, Perplexity 30.7%, and Gemini 17.6%.
Those percentages look tiny until you read the caveat, which is that the real number is almost certainly higher. Ahrefs notes that platforms like Copilot, Mistral, and Jasper do not pass referrer information, so their visits land in analytics as direct traffic.
We also think referral traffic is the wrong scoreboard. When a model tells a founder that a particular studio is worth talking to, that person often arrives by typing the brand name into Google. It shows up as branded search, not as an AI referral.
Almost certainly yes. If we had to give one piece of advice to a company publishing weekly, it would be to publish monthly and put the same total effort into four times fewer pages. Depth is what survives summarization, and volume is what gets summarized.
There is a caveat, and it is worth being precise about. Coverage still matters for establishing that you own a subject, because search and AI systems both respond to breadth across a topic. The distinction is between breadth built from real substance and breadth built from padding.
The way we square it is to plan a topic properly rather than a calendar. Decide what your business genuinely knows, map the real questions inside that area, and write only where you have something to add. That gives you breadth as a byproduct of substance instead of as a target to hit.
On your main domain, in a proper CMS, rendered as static HTML wherever possible. A blog on a separate subdomain or a third-party platform sends its authority somewhere else, and a blog that only renders after JavaScript runs risks not being read by some crawlers at all.
This sounds like a technical footnote and it is not. If the article text is not in the HTML that arrives from the server, you are depending on every crawler executing your JavaScript correctly. Some do. Some do not, and AI crawlers are less predictable here than Googlebot.
Webflow CMS handles this well for most business blogs, which is why we reach for it on client builds. We wrote up the practical setup, including collection structure and SEO fields, in our guide to building a blog with Webflow CMS.
Blog, but change what you publish. Stop writing definitions. Start publishing the things only your business can say, which means real results, real numbers, and real opinions. Publish less often and let each piece carry weight it can defend for years.
The practical next step is an audit rather than a new post. Pull your existing pages, find the ones with impressions but no clicks, and rewrite those to answer faster and harder. Then look for anything in your business that could become original data, because that is the asset AI cannot replicate.
We have delivered over 150 projects, and the pattern we see is consistent. Companies that treat their blog as a proof archive keep getting value from it. Companies that treat it as a publishing quota eventually stop, and they are usually right to. If you want a second opinion on which one you are running, we are happy to walk through it with you at phoenix.studio.
Tell us where you want to go. We'll tell you how we'd get you there.