Over 92% of content marketers now use AI. Only 14% say their blogs deliver strong results.
Those two numbers should not coexist. If a tool is being used this widely, the results should follow. They have not. In Episode 147 of SEOTalk Spaces, five practitioners sat with that gap and tried to answer one uncomfortable question: is AI the problem, or are we?
- AI adoption is near-universal in content marketing. Strong results are not. The gap between those two facts is the real story.
- Most practitioners in this debate agreed: the tool is not the problem. The intent behind using it is.
- Measuring content by sessions is outdated. Citations, brand demand, and assisted conversions are what matter now.
- Volume is not the enemy of value. But volume without a brief, a brand voice, or a clear strategy is.
- If your brand stopped publishing tomorrow, would your audience notice? The honest answer to that question reveals more than any analytics report.
The Numbers That Opened the Debate
The 14% figure comes from Orbit Media’s annual blogger survey, the lowest level of blog effectiveness recorded in over a decade. It landed in the same period when AI writing tools became standard issue for content teams everywhere.
That parallel is hard to dismiss. Either AI has made content worse, or the relationship between publishing effort and audience results broke down for other reasons, and AI arrived at exactly the wrong moment to take the blame.
Parth pointed to measurement as a sharper diagnosis. Many SEO teams are still counting sessions as the primary indicator of content success. Sessions tell you whether someone arrived. They say nothing about trust, brand recall, assisted conversions, or citations in AI-generated answers. The KPI stayed the same while the value chain moved on.
Is AI the Villain, or Is It Our Excuse?
This was the fault line the whole debate ran along.
The case against AI: it dropped the barrier to publishing so far that volume replaced thinking. When generating a 1,500-word draft takes three minutes, the pressure to spend two hours on a brief disappears. The output looks complete. The strategy behind it is hollow.
The case against that case: Malhar put it plainly. “Have we started blaming AI for poor content because it is easier than admitting that marketers were always producing mediocre content?” AI did not create the impulse to publish without purpose. It just made that impulse cheaper to act on.
Isha came at it from a leadership angle. “As a leader I would not ask why AI produced poor content. I would ask why the content is poor if you have AI.” The tool is not making the decision. Someone is directing it badly.
Parth drew a comparison that stuck: when everyone on a hundred-brand category is feeding the same prompt to the same model trained on the same data, the outputs converge. It is content spinning with better grammar. The differentiation has to come from the inputs: the original research, the proprietary data, the brand perspective. Not from the generation step.
The Volume Trap
Tanay offered a partial defence of volume that the room largely accepted, with conditions.
He argued that volume is not inherently bad; it is how you learn what works. Run ten creatives, find the one that outperforms the other nine, and scale it. That is a legitimate testing strategy. The problem is when volume becomes the goal rather than the diagnostic.
His real-world example made the point concretely. Using Opus 5.5, he created a 90-second motion graphics video for his organisation’s booth at BrightEU. Someone at the event asked for the contact details of the editor. The video was entirely AI-generated. The question is not whether the volume was high or low. The question is whether intention and craft went into directing the tool.
Isha’s framing cut through the debate neatly: “The whole objective of AI is to add value to your volume. You are using the most sophisticated lab in the world to open your Amazon packages.”
The Research Hour Problem
One of the quieter costs of AI-accelerated content came up when Malhar asked whether content teams are undervaluing the five or six hours of deep research that used to precede every article.
Suchi’s analogy landed cleanly. The old process was hand-washing: slow, precise, fully in your control. The new process is a fully automatic machine. Faster, yes. But if one coloured cloth gets in with the whites, everything comes out stained. Speed without attention to what goes in creates its own category of damage.
Proma made the same point through her own career. She began writing about rugby, perfumes, and cars, topics she knew nothing about. Each piece took hours of research. That depth became transferable understanding. AI can compress research to minutes. What it cannot replicate is what happens to your thinking when you sit with a difficult subject long enough to find the edges of it.
Her current approach: treat AI output the way a good editor treats a junior writer’s draft. By publication, the voice, perspective, and accountability are entirely hers.
Would Your Audience Miss You?
Malhar asked the group to imagine a brand stopping all content for three months. Would anyone notice?
Proma’s answer was revealing: she would miss specific personal brands: writers whose work feels like a weekly conversation. She would not miss most consumer brands. The difference is not production quality. It is whether the content is built around a genuine point of view.
Suchi named Avinash Kaushik’s newsletter as something she would go looking for if it stopped arriving. The reason is not that it is well-formatted or SEO-optimised. It is that no one else can write it.
Isha pointed to Zomato and Amul as the rare corporate brands with that quality: a distinctive voice that would leave a gap. Most brands, she said honestly, would not leave one. Another brand producing similar content would fill the space within a week.
That observation carries a strategic implication. If your content can be replaced by a competitor’s content without your audience noticing, the problem is not your production process. It is your distinctiveness.
If You Were Rebuilding From Scratch
The closing question produced four distinct answers, each pointing at a different root cause.
Parth: The fundamentals have not changed. Define what success looks like before you generate a single piece. Measure citations, brand search, and assisted conversions, not sessions. Then scale with AI once the strategy is clear.
Suchi: Spend more time on planning. AI has made it faster to build a content calendar. That speed is a trap if the calendar is built around what AI thinks is relevant rather than what your audience actually needs.
Isha: Stop hiring generalists. Find people who can think about your brand, your customer, and your message before they open a tool. Keep AI out of the room until the human team knows what it wants to say.
Proma: Write the content calendar by hand first. Take a printout. Use a pencil. Work out what you want to say to your audience. Then feed that human intention to AI and ask it to build on what you have already decided.
The common thread: AI is most valuable at the execution layer. The strategic layer (what to say, to whom, and why it matters) still requires a human who is willing to think before they generate.
What the Community Did Not Resolve
The debate landed on some clear agreements. AI is a tool. Tools reflect the intentions of the people using them. Measuring sessions in 2025 is a category error. Brand voice is a human job.
But two questions stayed genuinely open.
The first is the over-reliance question Malhar closed with: at what point does using AI for decisions cross into letting AI make decisions for you? That line is blurry, and the community acknowledged it.
The second is the talent question. Isha’s instinct to protect creatives from AI until they have developed their own thinking runs counter to how most organisations onboard people now. Whether that approach scales is unresolved.
Both are worth carrying into the next conversation.
Frequently Asked Questions
Did AI actually cause content marketing results to decline?
Not directly. The data shows a correlation between rising AI adoption and falling content effectiveness, but the practitioners in this debate largely attributed the decline to strategy and measurement failures that AI made easier to hide. The tool lowered the cost of publishing without raising the cost of publishing badly.
Is measuring content by sessions still worth doing?
Sessions tell you whether people arrived. They do not tell you whether the content built trust, drove a branded search, or contributed to a purchase weeks later. For most content programmes in 2025, citations, brand demand, and assisted conversions are more meaningful indicators than session counts alone.
How can more AI use lead to worse content performance?
When a large number of competing brands use the same AI tools trained on the same data to cover the same topics, the outputs converge. The content becomes difficult to distinguish, which reduces its ability to build authority, earn links, or be cited by AI search systems. Differentiation requires original inputs: proprietary research, real experience, and a clear brand perspective that the model cannot generate on its own.
What is context engineering and why does it matter?
Context engineering refers to how much deliberate information and direction you provide to an AI model before asking it to produce content. A vague prompt produces generic output. A prompt that includes your brand voice, your audience’s specific needs, your internal data, and a clear framework for the piece produces something far more useful. The quality of AI-assisted content is largely a function of the quality of the context you give it.
Is high-volume AI content ever a valid strategy?
Yes, in specific circumstances. If the volume is used as a testing mechanism to understand what resonates with a specific audience, and if the best-performing content is then developed with more depth and intention, volume serves a purpose. The problem is when volume becomes the goal itself rather than a step in a learning process.
What should you change first if your content programme is not working?
Fix the measurement before you fix the content. If you are optimising for sessions, you will make the wrong decisions. Identify what conversion, trust signal, or demand indicator your content should be contributing to, and work backwards from there. Only once the success metric is correct does it make sense to audit the content itself.
Catch the episode replay here:
