Busted Boost?

What a simple Facebook boost taught me about advertising on a platform that refuses to stand still

A boost is a boost is a boost.

Well… apparently not.

Before every book launch, I usually do a quick, inexpensive reconnaissance mission on Facebook. I post a few potential images on my I. C. Cosmos page, boost them, and watch what happens.

This isn’t sophisticated advertising research. It’s more like dipping a toe in the water before jumping in.

Which images make readers stop? Which ones get a reaction? Which ones vanish into the Facebook void?

Before launching my latest thriller, Nemesis, I started with a photograph of Baltimore at night that I was considering for the cover. I boosted the post, and likes began coming in as expected.

A couple of days later, I boosted another potential cover using the same basic settings. This one featured two of Nemesis’ central symbols: a sword and the scales of justice.

And… crickets.

There were fewer than ten visible likes.

Now, different images obviously produce different reactions. That was the entire reason I was testing them. But I had never seen one of these little boosts produce almost no visible response.

My immediate conclusion? Wow. I really misjudged that cover.

Then Meta sent me a message congratulating me on the boost. Apparently, it had generated more than 200 likes.

Excuse me? I could see fewer than ten.

Where did my 200 likes go?

That sent me investigating.

What I discovered was that the boosted post had been shown across multiple Meta placements, and the engagement generated in those placements didn’t accumulate visibly on the original Facebook post.

In other words, Meta and I were both looking at the same promotion—but not necessarily at the same version of reality.

There are ways to exert more control. For example, you can build an engagement campaign in Ads Manager around an existing post and set the ad to display all engagement on that post.

I tried that with a couple of posts, but it led to more expensive clicks and decidedly unexciting results.

So rather than spend my launch trying to force a simple boost to behave the way I remembered it, I moved on to more systematic testing.

That proved to be the right decision.

But the boost story wasn’t done with me yet.

Same Meta. Different page. Different rules.

Somewhere in the middle of the Nemesis launch, I realized I had forgotten to introduce the book on my second Facebook page, which is aimed more at writers than thriller readers.

I put together a quick post and boosted it.

This time, Meta offered me a placement choice: Facebook and/or the Meta Audience Network.

I chose Facebook, and almost immediately, visible likes and shares began streaming onto the post.

Interesting.

So, naturally, I went back to my I. C. Cosmos reader page and tried another boost there.

Now Meta gave me placement choices on that page too—which it had not offered a couple of weeks earlier.

Except these choices were different. This time I was choosing between Facebook and Instagram.

Then, while writing this article, I checked the available placement options on both pages again.

Different again. Facebook. Instagram. Messenger. WhatsApp. Different combinations depending on where I was looking.

At this point, I half expected Meta to offer placement on my refrigerator.

So… are Facebook boosts broken?

Not necessarily.

And that’s actually the point.

Boosts are a relatively small corner of Meta advertising. I wouldn’t build an entire book-launch strategy around them.

But this innocent little experiment captures something much larger about advertising on Meta in 2026:

The platform moves. Constantly.

Interfaces change.

Defaults change.

Available settings change.

Automation changes.

Two accounts—or even two pages belonging to the same person—may not be offered the same choices at the same time.

And something that behaved predictably six months ago may behave quite differently today.

That creates a problem for anyone trying to learn Facebook advertising from a rigid recipe.

Click this.
Select that.
Turn this off.
Always use this objective.
Never use that setting.

Those instructions may be perfectly accurate when they’re written. The trouble is that Meta may rearrange the furniture before you finish reading them.

What authors can do instead

This doesn’t mean advertising has become hopeless. It means we need a slightly different mindset.

Instead of memorizing Meta, learn to observe it.

Before spending significant money, ask yourself a few basic questions:

What exactly am I trying to accomplish?
Do I want engagement? Traffic? Sales? Information about which creative resonates?

What is Meta actually doing with my ad today?
Don’t assume the settings are identical to those you used last month—or even on another page.

What would count as a useful result?
A cheap click isn’t automatically a good click. A pile of likes isn’t automatically evidence that people want the book.

What am I testing?
If you’re comparing images, try not to change the audience, copy, objective, and placement all at once.

What can I learn before spending more?
Small tests are cheaper than large assumptions.

And perhaps most important:

Check the actual ad after you publish it.

Don’t assume Meta delivered exactly what you thought you created.

That lesson becomes considerably more important once AI enters the picture.

Build the plane—but watch the instruments

Advertising on Meta increasingly reminds me of the old expression about building the plane while flying it.

That sounds terrifying.

It doesn’t have to be.

The answer isn’t to know every button in advance. Nobody can—not when the buttons keep moving.

The answer is to know where you’re trying to go, watch what the platform is doing, and adjust when reality disagrees with your plan.

For authors, that means having a clear advertising goal and getting comfortable with testing.

Not testing everything randomly. Not throwing a dozen ads at the wall and celebrating whichever one sticks.

Testing with a purpose.

Because when the platform keeps changing, your real-world results become the most useful manual you have.

And sometimes, even those results can be contaminated without your realizing it.

In my next post, AI, AI Everywhere – How to Keep Your Tests Pure, I’ll show you exactly how I learned that lesson.