Testing the Editorial Waters (A/B test)

Testing the Editorial Waters

Can You A/B Test a Writing Style?

I accidentally conducted an experiment.
Well... “conducted” may be overstating things.
No laboratory. No control group. No clipboards. Not even a respectable spreadsheet.

Just a rabbit, two slightly different writing styles, and a collection of readers who had no idea they were participating.

🐇

For quite some time, my AI Wonderland Weekly followed roughly the same formula: a handful of interesting AI stories, a little context, a raised eyebrow, perhaps a rabbit hole... then I moved on before explaining everything to death.

The reader was left to connect some of the dots.
And people commented.

Then I wondered: What happens if I explain more?
Dangerous question.... 😝

Version A: The Nudge

The older Wonderland posts were essentially roundups.

Story.
Observation.
Wink.
Next story.

If something deserved interpretation, I usually nudged rather than explained.

Something like:
AI agents are beginning to browse websites for humans.
Interesting. For those of us building websites, perhaps very interesting.

And off we go down the next tunnel.

The implication remains sitting there.
The reader can pick it up. Or disagree with it. Or tell me I have completely lost the plot.

All excellent outcomes.

Version B: The Editorial

Then I tried something different.

Instead of simply putting the stories next to one another,
I started explaining why I thought they belonged together.

Agents. AGI. Local AI. Huge data centers. Robots. Ownership. AI visiting websites.

Not merely: Here is what happened.
Yet: Here is what I think may be happening underneath it.

The result was a more complete editorial. More context. More connections. More interpretation.
And, I thought, a stronger article.

Then something odd happened.

Fewer Reactions

There seemed to be fewer likes.
Fewer comments.

Hmm.

Had I improved the article...
...and made it worse?

Welcome to publishing.

😏

My first thought was that perhaps clarity wins.
Readers don't necessarily want the author doing all the thinking for them.
The older style left little gaps. And gaps are interesting things.
Because when the writer doesn't finish the thought, the reader sometimes does.

That unfinished thought may become: “I hadn't considered that.”
Or: “Nope. I don't agree.”
Or: “Wait until you hear what happened to me...”

And suddenly there is a comment.

Except Something Else Happened

I had also shared the new Weekly on Facebook.

And when I looked at the traffic...
there was the familiar spike.

Except this time it was bigger.

Now the experiment became considerably more interesting.
Fewer visible reactions inside the platform.
But more people arriving.

So perhaps the new article wasn't performing worse at all.
Perhaps it was doing a different job.

That sent me down another rabbit hole.
Naturally....

A Click Is Not a Comment

We tend to throw all our numbers into one basket: Traffic. Likes. Comments. Shares. Conversions. Engagement.
Yet... they don't tell us the same thing.

A click says: You made me curious enough to enter.
A like says: I enjoyed this enough to leave a signal.
A comment says something quite different: You left something unfinished enough that I wanted to join in.

That last one fascinates me.

Because perhaps good community writing shouldn't always provide the conclusion.
Sometimes it should provide the starting point.

And Then There Is AI

Which brings me to the elephant in practically every writer's room.
Or perhaps the robot.

🤖

AI-generated content now haunts online publishing.

People see polished prose and wonder: Did AI write this?
Sometimes before they've even wondered whether the article was useful.

And we have created some rather funny assumptions around that.

Too polished? AI.
Too structured? AI.
Too many headings? Definitely AI.
Em dash? Call the authorities.

😂

But I think we're asking the wrong question.

The interesting question isn't: Was AI involved?
Increasingly, it probably was somewhere.
Research. Brainstorming. Editing. Grammar. Images. Ideas. Outlines. Fact-checking. Whatever.

The better question may be: Did the writer leave anything of themselves in it?
A point of view. Selection. Restraint. Humor. Curiosity. An odd connection. A question nobody quite answered.

Because AI can explain almost anything.
And that may turn out to be precisely why not explaining everything becomes valuable.

Two Audiences, Two Styles?

My little non-scientific A/B-ish experiment has now left me wondering whether I actually discovered two different formats.

Here on WA, perhaps the lighter version works better.

The nudge.
The unfinished thought.
The open rabbit hole.

Because this is a community. There is somewhere for the reader to answer.

For a broader public audience, the more explanatory editorial may work better.

A reader arriving from Facebook, Google, or elsewhere may never comment.
They may simply want the whole journey.

Same research.
Same writer.
Different editorial job.

And perhaps that matters more than declaring one style “better.”

✨ Fleeky

So Now I'm Testing You

Which style do you prefer?
(And yes, I am using the em dash, it fits perfectly here, and does its job).

A — The Nudge

Give me the facts, add a little personality, point toward the rabbit hole...
...and let me fall into it myself.

Or:

B — The Editorial

Give me the facts and tell me what connections you see.
Take the idea further. Explain why it may matter.

Or perhaps...

C — Somewhere in Between

Enough explanation that I don't get lost.
Enough empty space that I still get to think.

I have my suspicion.

But this time...
I'm not going to explain it.

🐇

Over to you.

Thank you for reading, likes and shares. And your comments? Always appreciated!

And the results? Watch out for the next edition, ☺️☺️...