10 ChatGPT Prompts for Meta Ads That Actually Work (2026)

10 ChatGPT Prompts for Meta Ads That Actually Work (2026)

September 17, 20268 min read

10 ChatGPT Prompts for Meta Ads That Actually Work (2026)

If you connected ChatGPT to your Meta ads and you are just staring at the screen not knowing what to type next, I got you. Here are the 10 prompts I use in my own account every day, to audit, work on the creative strategy, and plan my next steps. They are all copy and paste ready.

If you would rather watch me run all 10 on a real account, the video is here: ChatGPT + Meta Ads: 10 Prompts That Actually Work

Two rules before you start

Before the prompts, there are two rules we need to talk about.

First, you need the connection ready. You have to connect the MCP to your AI, and you can use Claude, Claude Code, Perplexity, or ChatGPT. In the video I use ChatGPT, and I have separate videos on connecting both Claude and ChatGPT if you need them.

Second, give context to the AI and make sure your tracking is correct. I mention this in every video, but if your tracking is off, AI cannot pull the data on your best purchases and your top ads. And context matters too: your website, your offering, your target audience. If you use AI every day it should have some context already, but it is still good practice to start each conversation with a brief and give the AI a role, for example "act as a media buying specialist." This foundation is the first thing we get right inside Meta Ads Systems.

I split the 10 prompts into five categories: audit and diagnose, reporting, scaling and budget, creative, and competitor research.

Category 1: Audit and diagnose

Say you are working on a new account. You need to audit it and see what happened before.

Prompt 1 is a senior media buyer audit. On my own account it compared the old setup to the current one and told me what is working. My copy testing is working well: I only spent $25 and already got six purchases, so the CPA is about $4, which is great for my Skool community. It also flagged the wasted spend on my old ads (one of the reasons I raised my community price, because at $9 a month it is hard to even break even on ads, but at $27 it is much easier to run profitable ads). Then it analyzed the top KPIs: my link CTR is around 2 to 3%, my click to landing page views is averaging 66%, and my actual landing-page-to-purchase conversion rate is 6%, which is nice for a paid community. And it gave me three tests to run next: keep the winning image fixed and run a real copy test, run an audience test, and iterate on my winning image (image four).

Prompt 2 focuses on wasted spend. It ranks every active ad set against my average. In my case it confirmed a decision I had already made, that I paused my creative testing yesterday and there was nothing else I needed to pause right now. Here is the important part: AI is really good at suggestions, but you make the final call. Sometimes it sees a high CTR and says "scale this," but that CTR might be high because clicks are coming from Audience Network. So instead of scaling that ad, you remove Audience Network as a placement, get clean data, and then decide. That judgment layer is exactly what we drill in our Meta ads training.

Category 2: Reporting

Reporting helps you know how your data is changing, and if you manage accounts or clients, you need to report weekly, daily, or monthly.

Prompt 3 is a month over month report. It only worked for 35 seconds and gave me the account-level performance: spend to date vs last month, what improved, which campaigns got more budget, where performance dropped.

Prompt 4 visualizes it. It builds an interactive HTML report you can click through, with color-coded ROAS. On my account, ROAS was up from 2x to over 8x this month, CPA down over 50%, CPM down, CTR about the same. You can switch between metrics, focus on a single campaign or even a single ad, and adjust the benchmark KPIs at the top (I want CTR above 1.5% and at least 2.5x ROAS). I like seeing my data, not just reading the numbers, and it is far easier to present on a client call.

Category 3: Scaling and budget

Now it is time to scale what works.

Prompt 5 is my favorite of all 10. It is a super-prompt that reads your winners, your top performing countries, and your best placements, and then builds the scaling campaign for you. It gives me the exact layout: a CBO campaign, two ad sets, my best placements, my winning ads and copies, and a starting budget. Then I just type "approved, build the campaign for me," and it goes inside my ad account and comes back with a URL to the live campaign. It builds everything paused so you can review first. It is pretty magical: you have someone working for you, you go grab a coffee, come back five minutes later, and you have a scaling campaign live without doing the manual work.

When I reviewed it in Ads Manager, it had followed my previous campaigns exactly, which is why clean naming conventions matter so much, the AI now uses the same naming and it is scarily accurate. It even used ad set budgets instead of sharing budget across ad sets, following best practice. One thing to catch: it left Audience Network on as a placement, which we do not want, so I turned it off. This is why you always review. The ads it built pointed to my about page with no related media and no AI suggestions from Meta, my primary text, my learn more CTA, and correct tracking. This is the same structured approach behind our AI creative workflows.

Prompt 6 is a budget-shift simulator. It builds an interactive slider: if I move 5% of my budget, my ROAS would be about 1.7% higher; if I move 40%, about 13% higher. It works off your existing numbers, so it is not perfectly realistic (Meta ads do not keep a 10x ROAS when you double spend), but it is great to illustrate on a client call, like planning a Black Friday campaign and showing expected ROAS at different spend levels.

Category 4: Creative

Creative is the number one lever in Meta right now, so the better your creative, the better your ads perform.

Prompt 7 pulls your top ads by ROAS and tells you the winning format, hook, and angle. On my account the winner was a static image with simple text on a white background, and it analyzed the hooks in my copy so I know the creative pattern that is working.

Prompt 8 makes five new ad concepts based on what is already winning. It gave me the concepts, the hooks, the angles, and the formats, plus why each is likely to work and which to test first. What is smart is that it analyzes not just the image but the ad copies, so you can take your top-performing copy and build new videos, images, or checklists around it.

Category 5: Competitor research

Prompt 9 is a competitor spy, and it is super powerful. It analyzes your about page, landing page, offering, and ads, then finds competitors in your niche without you giving any names or URLs. Even if you do not know who your competitors are, it finds them, shows their main hooks and angles (one was leaning hard on a $9 a month price), and gives you links straight to their Ad Library so you can see exactly what they are running and what they are testing that you are not. In my case, competitors were more focused on price, identified the customer profile harder, and had more proof in the hook, all ideas I can use.

Prompt 10 turns those competitor angles into your own concepts, grounded in what already works for your account. It gave me five ideas with the angle, the exact format, and why it would work, plus which to prioritize.

The pro move: chain and automate

The real power is chaining these together in one conversation. Run your audit, then the scaling prompt, then the creative research, all in the same chat so the AI keeps your context. Then automate them. You can tell the AI to run a creative research every morning or send new creative ideas by 8am, and you get an email and a notification when it is ready. So you start every week with a report and fresh creative ideas already in your inbox.

One honest note: the MCP is not rolled out to all locations and accounts yet, so check if you can use it. And always review the AI output before you publish.

Want the full prompt library?

These 10 are the ones I use most, but there is a full library plus the whole system for using them inside Meta Ads Systems, my Skool community. You get the prompts, the campaign structure they plug into, and real reviews of your actual account, with real member results. There is a 7-day free trial, so there is no risk on your side.

Join Meta Ads Systems on Skool and put all 10 to work on your account.

And if you want to watch every prompt run live, the video is here: ChatGPT + Meta Ads: 10 Prompts That Actually Work

Máté Hunyor
Máté Hunyor is the founder of Wupscale, a Meta Ads-focused business built around AI systems.
Back to Blog