I used to type the same three words into every AI tool when a draft felt off: “Make this better.”

The results were almost always polite, slightly longer, and somehow less useful than what I started with. After enough wasted revisions I stopped using the phrase. Here is why it fails on ordinary work and what I do instead.

What “Make This Better” Actually Asks For

The instruction is empty

“Better” does not specify better for whom, better at what job, or better according to which standard. The tool is left to invent its own definition—usually “more polished,” “more professional,” or “more detailed.” None of those automatically match the real goal.

The tool optimizes for the wrong target

Most models interpret “better” as smoother prose, richer vocabulary, or fuller explanations. That produces text that sounds improved in a vacuum and still fails the actual test: Can a busy colleague understand the point in thirty seconds? Does the tone match an internal update? Is every claim still accurate?

It hides the real problem

When I write “make this better,” I am usually reacting to something specific—vagueness, wrong emphasis, weak structure, or a tone that feels off. The vague prompt never names the problem, so the tool cannot fix it.

A Side-by-Side Example

Original rough paragraph (from a weekly report)

“Project A moved forward this week. We got some feedback and are making changes. Still waiting on a couple things before we can finish.”

Result after “Make this better”

“Significant progress was achieved on Project A throughout the week. Valuable feedback was received from stakeholders and is currently being incorporated into the ongoing revisions. A few remaining items are pending resolution prior to final completion.”

The second version is smoother. It is also longer, less specific, and harder to act on. No owner, no timeline, no clear status. “Better” produced worse information.

What Works Instead

I replaced the empty request with three short, concrete instructions. I use them in combination depending on the problem.

1. Name the single biggest problem

Examples:
“Make the status clearer and more specific.”
“Cut every sentence that does not contain a concrete fact.”
“Rewrite so a manager can see the risk in the first two lines.”

2. State the audience and the job the text must do

Examples:
“This is an internal update for my manager. It should take under a minute to skim.”
“These are comment replies. Keep them polite, short, and human—no marketing language.”

3. Give a hard constraint

Examples:
“Stay under 120 words.”
“Do not add new claims I did not supply.”
“Use plain language, no words longer than necessary.”

When I feed the same rough paragraph the three instructions above, the result usually comes back tighter, more specific, and closer to something I would actually send.

Why Specific Beats Polished

Ordinary work does not need writing that sounds impressive. It needs writing that survives contact with a real reader who is short on time and already has context. Specificity forces the tool to stay close to the facts and the purpose. “Make this better” lets it drift toward generic quality.

The difference shows up most clearly in the editing time that remains. Vague prompts leave me rewriting substance. Specific prompts leave me making small adjustments.

A Simple Replacement Rule

Before I ask an AI to improve anything, I force myself to finish this sentence:

“The main problem with this draft is _______________.”

If I cannot finish the sentence, I am not ready to prompt yet. If I can finish it, I put that problem in the prompt and leave “make this better” out entirely.

I tried it so you don’t have to waste your afternoon.