How Content Marketing Changed, and What Still Works

Content marketing changed more in the last two years than in the ten before, and most of the change runs in one direction: producing text stopped being difficult.

This page deliberately has no year in its title. Trend pieces are written to be replaced, and a page nobody maintains becomes wrong on a schedule. What follows is what changed and what did not, which stays true longer.

What actually changed

• Producing text became nearly free. Anyone can generate a competent article in minutes. Whatever advantage came from being able to write quickly is gone.

• Volume stopped working. When everyone can publish fifty pages a month, publishing fifty pages a month distinguishes nobody, and search engines have become correspondingly less impressed by quantity.

• Simple questions stopped sending traffic. Answers now often appear at the top of the results, assembled from pages, and the person never clicks. Content whose value was a short factual answer lost its audience.

• Sameness became the default failure. The easy way to produce content now produces the average of everything already published, which by definition cannot be better than what exists.

What stopped working

• Publishing on a schedule with nothing to say. The schedule was always a proxy for commitment and it is now just output.

• The guide-to-everything, assembled by summarising the pages already ranking. This is precisely what an assistant produces for free.

• Keyword-first writing, where a phrase is chosen and an article reverse-engineered around it. It reads like it, and readers can tell.

• Thin pages built to catch a search. There are far more of these than there is demand for them.

What still works, and it is a short list

1. First-hand experience. Something you did, saw, measured or got wrong. This is the one thing that cannot be generated, because it did not happen to anyone else.

2. A real position. Saying something that could be disagreed with. Content designed to be agreeable to everyone is filtered out by everyone.

3. Specificity. A named situation, a named trade, a named town, an actual number from your own work. Generic copy is ignored precisely because it fits nobody in particular.

4. Being genuinely useful at the moment of a decision. Somebody choosing between options and looking for help deciding will read a great deal, and will remember who helped.

Where AI fits, said plainly

We use these tools throughout our own work and there is a note at the bottom of every page here saying so.

They are good at drafting, restructuring, research and catching your own errors. They are not good at knowing what is worth saying, and they will not tell you when the output is fluent and wrong. The failure mode moved from publishing too little to publishing a great deal of the wrong thing quickly, which is harder to notice because it looks productive.

The practical rule we work to: a machine can produce the draft, and a person has to have had the experience the draft is about. If nobody has, there is nothing there worth publishing, however well it reads.

What to do about it

• Publish less and mean it more. Four pages a year that could only have come from you beat forty that could have come from anyone.

• Write down what you learned this month, before you decide whether it is interesting. It usually is, and it is the raw material nobody else has.

• Answer the questions customers actually ask you out loud. Those are the searches, phrased the way real people phrase them.

• Stop measuring output. Publishing volume is an activity metric and it does not correlate with anything you care about.

If you want to work out what only you could write about, book a free SEO consultation. You will speak to the person who does the work.

Written by Spencer Capron at Capron Media, in collaboration with our AI agents. The same way we do the work for clients. Spencer reviews everything published here.