Is it logical to produce content with AI? My honest answer as a content agency

One question: when you say "brings traffic," what exactly did you measure?

The reason I'm asking is this: over the past year, click-through rates for informational content have dropped on many sites because the answer is shown directly on the search results page. You might not get clicks even if your ranking stays the same.

So the statement "the first type brings traffic" might not be as true as it was three years ago. Have you had a chance to look at how your own click-through rate has changed over time?

Great question, I'll be honest: we looked into it and saw a trend in the direction you mentioned.

Our click-through rate dropped noticeably for searches asking for short definitions. On the other hand, we didn't see a drop in decision-stage searches like "how to," "which is better," or "when is it needed."

This changed our content plan: we no longer make definition content separate pages, we put them as sections within decision content. This way we answer the question and create a page that deserves the click.

So your objection is valid and we hit the same wall. I should have mentioned this distinction when writing the post.

i'm a content creator and honestly this topic stresses me out because suddenly everyone's an expert and everywhere is flooded with the same posts!

but i noticed something, do you know which content my followers react to the most? the ones where i talk about my mistakes. videos where i say i didn't like a product get more views than ones where i praise it.

anyway, what i mean is: a machine can't make mistakes for me. so i'll keep talking about my mistakes.

As a translator, let me look at this from my field, we went through a similar process.

When machine translation arrived, the same debate happened in our industry. The result was this: translation of high-volume and standard texts largely moved to machines, with humans remaining as editors. Rates dropped in that segment.

But these areas stayed with humans: legal texts, marketing texts, anything requiring cultural adaptation. Because the cost of error is high here and the correct version requires context.

I think the same will happen for content. If the cost of error is low, a machine is enough. If the cost of error is high, you need a human to take responsibility.

Let me add a warning too: the most expensive jobs in translation turned out to be fixing machine translations. It takes longer than writing from scratch because hunting for errors is more exhausting than writing.

Let me share a distinction I teach my students as an educator, it's useful.

Can you verify the accuracy of the generated text? If yes, use it, because you'll catch errors. If no, don't use it, because you'll publish errors.

a beginner can't spot errors in text generated on a topic they don't know. so the expert benefits the most while the novice gets hurt the most. same tool, different outcome.

My questions are cleared up, thanks.

thanks, this was very helpful.

You're right. If it's your first time, start small; scaling comes later.

If you post the result here, it will help others too.

Timely topic.

Good call starting this thread.

Thanks for writing this, thats the right way. honestly if you get three different answers on a topic the question was asked wrong.

If you have questions write them; I'll answer as best I can.

great work.

Same here. The harder it is to reverse a decision the slower you should make it.

The discussion got scattered, let me summarize. The harder it is to reverse a decision, the slower you should make it.

The harder it is to reverse a decision, the slower you should make it.

Thanks, that was the answer I was looking for.

I agree with this. Internal linking is the cheapest and most neglected task.

Good luck with that.

This is exactly what we experienced. A competitor's ranking might come from years of accumulation, not just their content.

Map listings bring more business than blog posts for most local businesses.

I have no experience with content, so I'm asking. Trying to do this alone is the most expensive way.

Processes without records never improve, because you don't know what to fix.