Solved: How long to run ads with ChatGPT support?

Hi there,

Thanks for reaching out!

Happy to give you some of my initial thoughts on your question. It's probably the most common one I hear, and honestly, getting the answer right is what separates the campaigns that fly from the ones that burn cash. That advice from ChatGPT about 50 conversions... it's not wrong, but it's dangerously incomplete. It's a bit like a doctor telling you to just take two paracetamol for any illness. Sometimes it works, but often it's masking a much bigger problem.

The real question isn't just "how long do I run an ad for?", but "what am I actually looking for to tell me if this is working?". Below I've put together a pretty detailed breakdown of how I'd approach this, moving from the simple rule of thumb to the deeper strategy that actually drives results.

You'll need to understand the '50 Conversion' myth...

Alright, let's get this one out of the way first. That 50 conversions figure comes directly from the ad platforms themselves, like Meta and Google. When a new ad set or campaign starts, it enters what they call the "learning phase". During this time, the algorithm is frantically testing your ad on different pockets of your target audience, trying to figure out who is most likely to take the action you want (e.g., click, sign up, or buy).

The platform needs about 50 of these actions within a 7-day period to gather enough data to stabilise and start delivering your ad more efficiently. Once it exits the learning phase, performance should become more consistent and, hopefully, cheaper. So, in a perfect world, yes, you let it run until it gets those 50 conversions. It gives the machine the best chance to learn.

But we don't live in a perfect world. The problem is, this rule has no concept of your budget or your profitability. If a conversion is costing you £100 and your target is £20, waiting for 50 of them means you'll have spent £5000 just to learn that your ad is a disaster. That's not a strategy; it's just setting money on fire.

So, the rule is a technical guideline for the *algorithm*, not a business rule for *you*. You need to know when to pull the plug way before you hit that number if the early signs are bad. It's a balancing act: give the algorithm enough room to work its magic, but not so much rope that it hangs your bank account. Here’s a rough guide on how I think about it.

I'd say you need to diagnose the funnel, not the ad...

Most people make the mistake of only looking at the final number: sales, leads, whatever. They see zero conversions and panic. But that's like looking at the scoreboard at the end of a football match without watching the game. You know you lost, but you have no idea why.

A good paid ads expert is a detective. You have to follow the trail of clues the user leaves behind. This is the real way to know if an ad "is not going to work".

Let's imagine you're running an eCommerce store. Here's the journey and the questions I'd be asking at each stage:

• Impressions -> Clicks (Click-Through Rate - CTR): Are people even clicking on your ad? If you have thousands of impressions but a really low CTR (say, under 1%), it tells me the ad itself isn't grabbing attention. The image might be boring, the headline might be weak, or your targeting is just completely off and you're showing it to the wrong people. The ad is the problem here.


• Clicks -> Landing Page Views: Okay, so people are clicking, but are they actually sticking around to let the page load? If you see a big drop-off here, it's often a technical issue. Your website is probably too slow, especially on mobile, and people are bouncing before it even loads. This isn't an ad problem; it's a website problem.


• Landing Page Views -> Adds to Cart: You've got people on your product page, they're looking around, but nobody is adding to cart. This is a classic sign of a mismatch. Either the traffic you're bringing from the ad is wrong (e.g., your ad promised a huge discount that isn't on the page) or, more likely, the product page itself is the issue. Are the photos good? Is the description persuasive? Is the price clear? Is it too high? Does the site look trustworthy? This is a product/offer problem.


• Adds to Cart -> Purchase (Conversion Rate): The final hurdle. People are adding to cart but not checking out. This is almost always a problem with your checkout process. Unexpectedly high shipping costs are the number one killer here. A complicated checkout form, not enough payment options, or forcing people to create an account can also wreck your conversion rate. This is a checkout/process problem.

You see? An ad that's "not working" can mean a dozen different things. By looking at these 'leading indicators', you can diagnose the problem long before you get 50 conversions. If your CTR is 0.2%, you don't need to spend any more money to know the ad creative or targeting is wrong. If you're getting loads of clicks but no adds-to-cart, you know the problem is on your website, not the ad. Fix the real problem first, then run the ad again.

We'll need to look at your offer, it's probably not the ads...

I'm going to be brutally honest here, because it's the most valuable lesson I've ever learned in this industry. Nine times out of ten, when someone tells me their "ads aren't working," the ads are the last thing I look at. The real problem is almost always the offer.

You can have the most brilliantly targeted, beautifully designed ad campaign in the world, but if what you're offering is something nobody wants, or it's presented in a way that creates friction and confusion, you will fail. Full stop. I remember one B2B software client who came to us, they were burning through cash. They were selling a complex accounting system. Their website was a mess, their main selling point was "privacy" which most businesses dont really care about for accounting, and they didn't offer a free trial - just a "request a demo" button. They were asking for marriage on the first date.

We had to tell them to stop all advertising immediately. No amount of ad testing would fix that. Their offer was fundamentally broken for cold traffic. They needed a completely free trial to get people in the door, and they needed to change their messaging to focus on benefits businesses actually care about, like saving time or money. Once they fixed the offer, the same ads started working.

Before you spend another pound, ask yourself these questions:

• Is my offer instantly understandable? Can someone look at my ad and landing page and know *exactly* what I'm selling and for whom in under 5 seconds?
• Is it compelling? Does it solve a real, urgent, and expensive problem for a specific group of people? Your ad copy needs to agitate that problem and present your offer as the clear solution. Don't sell "AI implementation services," sell "stop your best engineers from quitting over broken workflows."
• Is it low-friction? How easy are you making it for someone to say yes? The "Request a Demo" button is one of the highest-friction calls to action there is. It screams "I'm going to have to sit through a sales pitch." Can you offer something of value upfront instead? A free trial, a free tool, a helpful checklist, an audit. For our agency, we offer a free strategy session. We solve a small part of their problem for free to earn the right to solve the whole thing.

A weak offer forces your ads to do all the heavy lifting. A brilliant, no-brainer offer makes your ads' job easy. It pulls people in. Fix your offer first.

You probably should obsess over targeting, not timing...

Once your offer is solid, the next piece of the puzzle is who you're showing it to. This is where most people get lazy. They'll pick a few broad interests on Facebook like 'Business' or 'Marketing' and hope for the best. This is like trying to catch a specific type of fish by throwing a giant net in the middle of the ocean. You'll catch a lot of junk.

The goal isn't just to find people, it's to find your *Ideal Customer Profile* (ICP). And your ICP isn't a demographic. It's not "women aged 25-40 who live in London". It's a person with a specific, career-threatening nightmare that you can solve. Your Head of Sales client isn't just a job title; he's a guy who's terrified of missing his quarterly target and having to explain it to the board.

Once you understand that pain, you can find them. What podcasts do they listen to? What newsletters do they actually read? What software do they already use? Who do they follow on LinkedIn? That is your targeting. On Meta, for example, instead of targeting "Amazon," which is full of shoppers, a smart agency targeting eCommerce stores would target interests like "Shopify," "WooCommerce," or follow pages of eCommerce influencers. It's about precision.

I structure my campaigns based on the sales funnel. This ensures I'm showing the right message to the right person at the right time.

If your BoFu retargeting ads aren't working, that's a massive red flag about your checkout or pricing. If your ToFu ads are getting clicks but no one ever moves to MoFu, your landing page is probably the problem. Structure gives you clarity.

You'll need to know your numbers to scale...

This is the final, and most important, piece of the puzzle. It brings everything together. You can't possibly know if an ad is "working" or not if you don't know how much a customer is actually worth to you.

Most people obsess over a low Cost Per Lead (CPL) or Cost Per Acquisition (CPA). They want the cheapest clicks and the cheapest leads. This is a trap. I would rather pay £200 for a lead that I know turns into a £10,000 customer than pay £5 for a lead that never buys anything.

To do this, you need to calculate your Customer Lifetime Value (LTV). This is the total profit you can expect to make from a single customer over the entire time they do business with you. Here's the simple maths for a subscription business:

LTV = (Average Revenue Per Customer Per Month * Gross Margin %) / Monthly Customer Churn Rate

Once you know your LTV, you can work backwards. A healthy business model often aims for a 3:1 LTV to Customer Acquisition Cost (CAC) ratio. This means if your LTV is £9,000, you can afford to spend up to £3,000 to acquire that customer and still have a very healthy business. Suddenly that £200 lead doesn't look so expensive anymore, does it? It looks like a bargain.

Knowing this number is liberating. It frees you from the tyranny of cheap clicks and allows you to advertise on more expensive, but higher quality, platforms like LinkedIn or Google Search. It gives you the confidence to let an ad run, even if the early CPA is a bit high, because you know the value of the customer you're trying to find. I've built a simple calculator for you to play around with your own numbers.

This is the main advice I have for you:

So, to bring it all back to your original question. How long do you let an ad run? You let it run long enough to give you a clear signal on where your funnel is broken. Sometimes that's after spending £10. Sometimes it's after £500. The 50 conversion rule is the finish line, but you should be checking your progress at every hurdle along the way.

This stuff is a lot, I know. It's a world away from just following ChatGPT's advice. It's a process of constant diagnosis, testing, and understanding the deep psychology and economics of your business. Below is a summary table of the framework I've laid out.

Getting this right isn't easy, and it's where having an expert can make a massive difference. We've seen so many accounts, like the medical job matching SaaS where we took their cost per user from £100 down to just £7, not by finding one 'magic' ad, but by systematically applying this exact kind of diagnostic process over and over again.

It's about having the experience to spot the patterns quickly and know which lever to pull. If you ever feel like you'd like a second pair of expert eyes on your campaigns to see where the real opportunities are, we offer a completely free, no-obligation strategy consultation. We can jump on a call, share screens, and I can walk you through your own account using these principles.

Hope this helps!

Regards,

Team @ Lukas Holschuh