How to Get Better Output from ChatGPT

By now, most people have at least dipped a toe into the ChatGPT waters. Some people dove straight into the deep end and are using it every day (aka me). Others are still dabbling here and there. But for the most part in my experience, nearly everyone is using it in some form or fashion.

And as we’re all using it more, something interesting is happening.

We’re starting to learn what actually works … and what doesn’t.

Here’s what I mean … 

A year or two ago, ChatGPT felt a little like the wild west. Everyone was testing things, swapping prompts, posting screenshots of what it could do like we’d all just discovered fire. And to be fair, some of it was (and still is) insanely impressive.

But most of the focus back then was on the prompt.

How do I word this?

What’s the best format?

What’s the “right” way to ask?

And that made sense at the time. Prompt quality does matter. (If you know me, you know I’ll talk about that until I’m blue in the face.)

But as the technology has evolved, the way we use it has to evolve too. Because what I’m seeing now — both in my own work and in the questions I get from clients — is that better prompts alone aren’t solving the problem. 

That’s because using ChatGPT well isn’t just about what you tell it to do anymore.

It’s about how you interact with it.

How you guide it.

How you respond to what it gives you.

How willing you are to push it, question it, refine it, and call it out when it’s missing the mark.

In other words … it’s less about writing the perfect prompt, and more about being an active participant in the process.

So if your output has been feeling a little generic lately, it’s probably not because you’re doing it wrong. But it might be because you’re still treating it like a one-and-done tool instead of something you work with.

So today I want to share a few tips that have made the biggest difference for me. Maybe they’ll help you too. 

Why ChatGPT Responses Sound Generic (and How to Fix It)

1. Stop Letting It Agree With You

If ChatGPT is always agreeing with you … it’s not helping you. Yep, you read that right. 

But here’s the tricky part: ChatGPT is designed to affirm you. 

It’s built to be soft in its critique and give you something that sounds supportive and polished. Which, on the surface, feels productive. But in reality, it creates a false sense of clarity.

Because you walk away thinking: “Okay, this sounds good.”

When what you actually needed was: “Wait… does this hold up?”

This has probably been my biggest shift in how I use ChatGPT lately. I got frustrated when I was using it one afternoon and, without thinking, just typed “Stop just agreeing with me and actually be helpful! I don’t need you to affirm me. I need you to be an objective, strategic thinking partner.” 

And that right there — those three frustration-induced sentences unlocked another level of ChatGPT productivity. 

It finally stopped acting like my personal hype crew and started acting like a useful second brain. It pushed back a little more and started noticing gaps that I was too in the weeds to pick up on. It got better at being a true strategic brainstorming partner, and started helping me identify what felt weak, or where I needed to think harder. 

Game. Changer. 

You see, most of the time, the problem isn’t that we need more ideas. It’s that we need better perspective on the ideas we already have.

We need someone (or something) to help us sort through:

• what’s actually strong
• what still feels fuzzy
• what sounds good on the surface but doesn’t fully hold up yet
• and what might need another round before we run with it

And if ChatGPT is just sitting there clapping for everything you say, it can’t do that.

Now, I will say, I still have to remind it sometimes not to affirm me. It’s not like I told it once and now we ride off into the sunset together as perfect strategic partners forever. But making that shift changed the way I prompt it overall, because it forced me to stop looking for approval and start asking for actual insight.

So now, I’ll tell it directly:

• “Remember not to affirm me on this — give me your real feedback”
• ”Challenge my thinking on this one”
• ”Act like a strategist, not a cheerleader”

And just as importantly, I change the types of questions I ask.

Instead of:

“What do you think of this?”

I’ll ask things like:

• “What feels off about this?”
• “Where would you push back?”
• “What am I missing or overlooking?”
• “What would someone misunderstand here?”

That shift changes everything.

Because now you’re not looking for validation. You’re looking for clarity. And clarity is what actually moves your work forward.

It helps you get out of the “this sounds nice” zone and into the “this is actually stronger now” zone. It helps you pressure test your thinking before you waste time running with something half-baked. And it helps ChatGPT become a much more valuable tool, because now it’s not just polishing your first thought … it’s helping you refine it.

That, to me, is where things start to get really useful.

2. Give It Something to React To

Like asking it to stop affirming you, changing how you bring ideas to ChatGPT is a game changer too. Because it’s not just about how you ask for feedback. It’s about what you’re actually giving it to work with in the first place.

One of the biggest shifts I’ve noticed in my own use is realizing that ChatGPT is significantly better at reacting than it is at starting from scratch. And once you see that, a lot of the “generic output” problem starts to make more sense.

Let’s look at it this way: 

How many times have you asked ChatGPT something like: 

“Give me ideas for…”

“Write me a caption about…”

“Create a marketing plan for…”

And then been disappointed when the response feels surface-level or a little interchangeable? 

Yep, I’ve been there too. 

But if you think about it, ChatGPT gave you exactly what you asked for.

You gave it a broad prompt with no real context, no point of view, no direction, and it gave you a broad answer right back. There’s nothing specific for it to latch onto, so it defaults to what it knows will generally work, which is also what everyone else is getting.

That’s where the generic feeling comes from.

What I’ve found works much better is bringing something to the table—even if it’s messy. Actually, especially if it’s messy. A rough outline that’s not fully thought through, a half-formed idea you’re trying to articulate, a brain dump of thoughts that don’t quite connect yet, or a few different directions you’re considering. Not because it’s ready, but because it’s real.

From there, I’m not asking ChatGPT to create something for me. I’m asking it to work with me. To help organize what I’m thinking, identify patterns or themes, point out where something feels inconsistent, refine the structure, or challenge the direction entirely.

And this is where things start to shift.

Because now you’re not asking ChatGPT to create something out of nothing. You’re asking it to help you make sense of something that already exists. And when you bring your own thinking into the process, even if it’s incomplete, the output has something to build on. It has context, direction, and a point of view. It sounds more like you because it started with you.

It also changes your role in the process.

You’re not sitting back waiting for it to come up with something good. You’re actively shaping the outcome as you go. And in my experience, that’s when ChatGPT stops feeling like a content generator and starts feeling like a genuinely useful collaborator.

3. Give It a Perspective to Think From

Once you stop letting ChatGPT affirm everything you say and start giving it something real to work with, the next shift is this:

You have to guide how it’s thinking, not just what it’s doing.

Because if you don’t, it defaults to neutral. And neutral is usually where things start to feel a little flat. The feedback isn’t wrong, and it’s not necessarily unhelpful, but it’s also not giving you the level of insight you actually need to move something forward.

Here’s what I mean … 

Most people will ask ChatGPT to review something, refine something, or give feedback, which is great. That alone is already a more useful way to leverage the tool. But what’s usually missing is the nuance of how you want it to do that — in other words, what lens it should be looking through when it gives you feedback.

Without that, it defaults to a very general, middle-of-the-road response that could apply to almost anything. Not because it’s doing something wrong, but because it’s not sophisticated enough to intuit your intent or pick up on the subtext behind your ask.

For instance, you might know that when you ask ChatGPT to review something, it should be reviewing it as if it’s your target customer … but ChatGPT doesn’t know that. So, what I’ve found is that the quality of the output changes pretty significantly when you assign it a specific perspective.

For example, instead of asking:

“Can you review this?”

Ask things like:

“Think like my ideal client and tell me what feels confusing or unclear.”

“Review this like a strategist. Where would you push back or simplify?”

“Look at this from a skeptical point of view. What wouldn’t land?”

“If you were seeing this for the first time, what questions would you have?”

The content itself hasn’t changed, but the feedback becomes much more specific and much more useful, because now it’s not just analyzing your work — it’s analyzing it through a specific lens.

And what you’ll find is that different perspectives will surface different gaps.

Your “strategist lens” might catch structure issues.

Your “audience lens” might catch confusion.

Your “skeptical lens” might catch weak points or assumptions.

It’s powerful, and eye-opening. Because you’re no longer getting one general layer of feedback. You’re actively choosing the angle of analysis based on what you need in that moment.

And just like the other shifts we’ve talked about, this isn’t about coming up with a more clever prompt. It’s about being more intentional in how you’re using the tool and more specific about the kind of thinking you want it to bring to the table.

Because when you guide how ChatGPT thinks, not just what it produces, the output becomes a lot more relevant, a lot more insightful, and a lot more aligned with what you’re actually trying to do.

4. Train It in Your Voice

If you’ve followed me for any length of time, you’ve probably heard me say this a million times. And (spoiler alert) I’m going to keep saying it until I’m blue in the face:

Train. ChatGPT. On. Your. Voice.

Is this a new tip? No

Is it groundbreaking? Also no. 

Is it one of the biggest reasons people get frustrated with ChatGPT? Y.E.S. 

They expect it to sound like them without ever actually teaching it how, and then assume the tool just isn’t capable when the output feels generic or a little off. But that’s not actually what’s happening.

ChatGPT isn’t a mind reader. 

I repeat: ChatGPT cannot read your mind. 

It can’t pick up on your tone, your perspective, your audience, or your brand nuances unless you explicitly give it that context. If you skip that step, it defaults to what it knows, which is broad, generalized output that technically works but could belong to almost anyone. That’s how you end up with generic responses that sound like … well, a robot … wrote them. 

Now, there is a tactical side to this that matters. You do need to give it something to work with. That might look like sharing examples of your past content, explaining your tone and communication style, describing your audience, or continuing to build context over time so it has a clearer understanding of how you think and communicate.

But the part I care more about is what sits underneath that. Because in order to train ChatGPT in your voice, you actually have to know what your voice is.

And this is where things tend to break down.

A lot of business owners can talk about their offers and their services all day long, but when it comes to clearly articulating how they sound, what they stand for, and what makes their communication style distinct, it gets a little fuzzy. And if it’s fuzzy for you, it’s going to be fuzzy for ChatGPT.

This is the exact same issue that shows up when someone hires a copywriter without a clear brand foundation. The copy might be technically solid, and it might even sound polished, but it won’t quite feel right because there’s nothing concrete anchoring it. ChatGPT just exposes that gap a lot faster.

And honestly, I think that’s a good thing, because it forces you to get clearer.

It forces you to define how you actually communicate, what you want to be known for, and how you want your content to feel when someone reads it. Once that clarity is there, training ChatGPT becomes a completely different experience. You’re no longer trying to fix something generic. You’re giving it something specific to mirror.

That’s when the output starts to shift. It feels more aligned, more natural, and a lot more usable, not because ChatGPT suddenly got better, but because you gave it better input.

At the end of the day, ChatGPT isn’t here to replace your voice, but it can get significantly better at supporting it if you take the time to actually teach it.

5. Use It to Refine and Pressure Test

Up until this point, we’ve talked about how you interact with ChatGPT while you’re building something. This tip is the shift in when you use it, and honestly, it’s one of the most practical changes you can make.

Most people default to using ChatGPT at the very beginning of the process. They open it up, ask it to generate something from scratch, and then either run with what it gives them or spend way too much time trying to fix something that never quite felt right to begin with. And while that can be helpful in certain situations, it’s also where a lot of the frustration tends to come from, because you’re asking it to create something strong without giving it much to work with yet.

Where I’ve found ChatGPT to be significantly more useful is a little later in the process, once something already exists. Once I have a draft, an outline, or even just a messy version of an idea that’s at least partially thought through, I’ll bring ChatGPT in to help me refine it and pressure test it.

Not to rewrite everything or “make it sound better” in a vague, surface-level way, but to actually evaluate what’s there.

This is where I start using it to create a little bit of distance between me and the work, especially in moments where I’m too close to it or moving quickly and know I’m probably skipping over something. I’ll use it to gut check the thinking, to surface gaps, and to make sure what I’m about to move forward with actually holds up.

That usually sounds like asking where the weak points are, what assumptions I’m making that I haven’t fully addressed, what might be unclear or easy to misinterpret, or why something might not land the way I think it will. And depending on what I’m working on, I’ll also use it to sanity check or fact check certain pieces, especially if I’m referencing trends, examples, or broader observations that I want to make sure are actually accurate.

At that point, I’m not looking for more ideas. I’m trying to make sure the idea I already have is solid.

What I’ve found is that this creates a really helpful pause point in the process. Instead of running full speed ahead with something that sounds right, it gives you a chance to step back and ask whether it actually makes sense, whether it’s as clear as you think it is, and whether there’s anything you’re missing before you invest more time into it.

It also tends to save a lot of time on the back end, because it’s much easier to adjust something early than it is to realize later that you built an entire piece, plan, or direction on top of something that wasn’t fully thought through. (I may or may not be speaking from lots of experience here)

And just like everything else we’ve talked about, this shifts your role in the process. You’re not relying on ChatGPT to do the work for you. You’re using it to strengthen the work you’ve already done, which is ultimately what makes it a much more valuable tool.

Bonus Tip: Use It as a Verbal Brain Dump (Especially If You’re a Verbal Processor)

For all of you verbal processors out there (again, me), don’t sleep on this one. 

One of the simplest but most effective ways to use ChatGPT is to open the app, turn on voice mode, and just start talking. Seriously, simple as that. 

There’s no magic formula, no fancy prompt, and definitely no need to make anything sound polished. Just open ChatGPT and tell it you need to do a brain dump and you just need it to listen and catch your thoughts for a bit. 

Then unload. 

Because here’s the thing … if you’re anything like me, most of the time, the issue isn’t that you don’t have ideas. It’s actually the opposite. Everything is sitting in your head at the same time, half-formed and competing for attention, which makes it virtually impossible to figure out what’s worth pursuing and what’s just noise.

Enter the magic of the brain dump. 

Then, once you’ve done the equivalent of a hard refresh on the ideas swirling through your head, you can lean on ChatGPT to help you organize or make sense of everything (like we’ve talked about throughout this article). You can ask it to organize what you just said into something more structured, help you make sense of what’s actually worth pursuing versus what needs to wait, or look for patterns and themes in what you’re thinking about.

You can also take it a step further and ask it to shift perspectives, like having it think like a business coach and poke holes in your thinking, call out why you might feel all over the place, or identify where you’re overcomplicating something that could be simpler.

What’s helpful about this is that it removes the pressure to have everything figured out before you start. You’re not trying to present a clean, well-articulated idea. You’re starting with where you actually are, which is usually somewhere in the middle of a messy thought process.

And once it’s out of your head, it becomes a lot easier to work with.

A Few Final Thoughts

If there’s one thing I hope you take away from all of this, it’s this:

Getting better output from ChatGPT isn’t about finding the perfect prompt. It’s about how you engage with it after the prompt.

Because the difference between something that feels generic and something that actually moves your work forward usually isn’t in the first ask. It shows up in the follow-up questions you ask, the pushback you invite, the context you give it, and how willing you are to stay engaged in the process instead of treating it like a one-and-done tool.

And to be clear, none of this requires you to become some kind of AI expert or overhaul the way you work overnight. It’s really just a shift in how you approach it, being a little more intentional, a little more involved, and a lot less focused on getting it “right” the first time.

Because the reality is, ChatGPT is only as useful as the way you use it.

And when you shift from “tell me what to do” to “help me think this through,” everything changes. The output gets stronger, your ideas get sharper, and the work you’re putting out actually starts to feel like you again.

Which, at the end of the day, is the whole point … right?