Solved: Facebook Ads Reporting Inflated Conversion Data

Hi there,

Thanks for reaching out! I had a look over the issue you described, and I'm happy to give you some initial thoughts and guidance. Don't worry, what you're seeing is a proper common problem, especially with Meta Ads. It's frustrating as hell when the numbers don't line up, but it doesn't automatically mean the whole thing is broken or the algorithm is doomed. It usually points to a tracking issue that, once sorted, can make everything a lot clearer.

It's one of the most frequent things we see when we first look at a new account. People get really bogged down by the daily numbers, but getting the foundations right is what makes the real difference. I'll walk you through how I'd approach this, from the technical tracking stuff right through to the actual strategy for getting your newsletter signups at a decent cost.

We'll need to look at the data discrepancy first...

Right, so the first and most obvious problem is the gap between what Facebook's Ads Manager is telling you and what your own backend is showing. A 16 to 4 difference is massive and you're right to be concerned. If the algorithm is optimising for what it *thinks* are 16 conversions when you only got 4, it'll start making poor decisions about who to show your ads to. It's basically getting rewarded for the wrong actions.

There are a few likely culprits here, and it's usually a mix of them.

1. Attribution Windows

This is the big one that most people miss. Meta's default attribution setting is usually '7-day click or 1-day view'. This means if someone clicks your ad and converts within 7 days, it's counted. But, and this is the important part, if someone just *sees* your ad (without clicking) and then signs up within 1 day, Meta also takes the credit. Your backend system, on the other hand, is almost certainly only tracking direct clicks. So you get this massive gap. Someone could see your ad on their phone while on the bus, forget about it, then later get home, type your newsletter name into Google on their laptop, and sign up. Your backend sees a 'direct' or 'organic' signup. Meta says "Aha! They saw our ad earlier. That's our conversion". Neither is technically wrong, they're just measuring different things. But for your purposes, it's inflating the results and making the campaign look better than it is. You can change this attribution setting in your ad set settings to be '1-day click' only, which will give you a much more conservative and likely more accurate number, though it might still not match perfectly.

2. Pixel Firing Issues & Deduplication

You mentioned the conversion is tracked when a user hits the thank-you page. This is the correct way to do it. But you need to be absolutely sure the pixel is firing *only* on that page and that it's firing correctly. A common mistake is having the same 'Subscribe' or 'CompleteRegistration' event accidentally firing somewhere else, maybe on the initial landing page load as well. You can check this using the Meta Pixel Helper browser extension. Go through the signup flow yourself and watch what the pixel helper shows you. It should only show the conversion event on the final thank-you page.

Another, more technical point, is event deduplication. If you're using both the Meta Pixel (which runs in the browser) and the Conversions API (which sends data from your server), you could be double-counting. Each event needs a unique 'event_ID' so that when Meta receives the event from both the browser and the server, it knows it's the same one and can merge them. If this isn't set up propperly, it'll just count both. One from the browser, one from the server. Boom, two conversions reported for one actual signup. Go into your Events Manager in Facebook, look at the 'Diagnostics' tab. It'll often flag issues like this. It might seem a bit technical, but getting this right is absolutly fundamental for accurate tracking.

3. Cross-Device Conversions

This links back to attribution. Meta is brilliant at tracking people across different devices. Your backend probably isn't. So if someone clicks an ad on their work computer, gets distracted, then signs up on their personal phone later, Meta can connect the dots because they are logged into Facebook/Instagram on both devices. Your backend will see two different users and won't be able to link the ad click to the final conversion. There's not much you can do about this one, but it's important to understand it's happening and contributes to the discrepancy.

So, the immediate action here is to go into your ad set settings and change the attribution to '1-day click'. This will give you a much cleaner view. Then, use the Pixel Helper and Events Manager to diagnose any technical issues. Fixing the data feed is the first step before you can even think about optimising performance.

I'd say you need to set realistic cost expectations...

You mentioned your CPA basically tripled on one day. I know that feeling, it's a gut punch. But with a budget of $30 a day and only 8 days of data, wild flucutations are completely normal. You're still in what we call the 'learning phase'. The algorithm has very little data to work with, so one bad day can skew the averages massively. Don't panic and start changing everything based on a single day's performance. You need to look at the trends over at least a week, ideally longer.

Let's talk about what a 'good' CPA for newsletter signups even is. The answer is, it depends. But we can work out a realistic range. I've put together a table based on what we typically see for campaigns with a 'signup' objective in developed, English-speaking countries. It's all a function of two numbers: your Cost Per Click (CPC) and your Landing Page Conversion Rate (CVR).

As you can see, your CPA could reasonably be anywhere from under £2 to as high as £15. If your daily budget is $30 (£24ish), on a bad day where your CPC is high and your conversion rate is low, you might only get one or two signups. On a good day, you might get ten. That's a huge swing, but it's all within the bounds of normal performance. Your CPA tripling could just mean you went from a £3 CPA to a £9 CPA for that one day. Both are perfectly normal figures in this range.

The goal isn't to have a great CPA every single day. The goal is to get the *average* CPA down over a period of weeks. We've worked on several software campaigns where we were able to acquire 4,622 registrations at a cost of $2.38 per registration. You're only 8 days in. You need to give the algorithm time and data to learn and stabilise before you can properly judge the performance.

You probably should rethink your targeting strategy...

This is where we address your fear about the algorithm "showing my ads to the wrong people". The algorithm is incredibly powerful, but it's not a mind reader. It can only optimise within the constraints you give it. If you give it a poor-quality audience to begin with, it will find the 'best' people within that poor audience, but they'll still be the wrong people overall. Garbage in, garbage out, as they say.

For a newsletter, you need a structured approach to targeting. I usually break this down into different stages of the funnel.

Stage 1: Top of Funnel (ToFu) - Finding New People

This is where you are now. You're trying to reach cold audiences who have never heard of you. Your main tool here is 'Detailed Targeting' (interests, behaviours, demographics). The key is to be specific. A lot of people make the mistake of choosing very broad interests. For example, if your newsletter is about sustainable fashion, targeting the interest 'Fashion' is a terrible idea. It's way too broad. You'll be targeting millions of people who couldn't care less about sustainability.

Instead, think about what makes your ideal subscriber unique. What specific brands do they like (e.g., Patagonia, Everlane)? What magazines or blogs do they read (e.g., The Good Trade)? What influencers do they follow? Who are your direct competitors? Target *those* as interests. The audience size for each ad set should ideally be in the 1-5 million range. Any bigger and it's probably too broad. Any smaller and the algorithm might struggle to find people. I'd recomend setting up a few different ad sets, each one testing a different 'theme' of interests. For example:

• -> Ad Set 1: Competitor brands & newsletters
• -> Ad Set 2: Related media, magazines, blogs
• -> Ad Set 3: Broader related interests (e.g., 'Ethical Fashion', 'Slow Fashion')

This way you can see which 'angle' resonates most with your audience and delivers the best CPA.

Stage 2: Middle & Bottom of Funnel (MoFu/BoFu) - Nurturing & Converting

This is what you should set up next. This is all about retargeting. You'll need seperate campaigns for this. These audiences are 'warm' because they've already had some interaction with you. They almost always convert at a much lower cost.

• -> Website Visitor Retargeting: Create a custom audience of everyone who has visited your landing page in the last 30 days but has *not* visited the thank-you page (i.e., they didn't sign up). Show them a slightly different ad. Maybe it addresses a common objection or highlights the number one benefit of your newsletter.
• -> Ad Engager Retargeting: Create audiences of people who have watched a certain percentage of your video ads or engaged with your Facebook/Instagram page. These people have shown interest but might not have clicked through. Give them another nudge.

Stage 3: Lookalike Audiences

This is the holy grail of Meta advertising, but you need data for it to work. Once you have at least a few hundred, ideally 1,000+, subscribers on your list, you can upload that list to Facebook and create a 'Lookalike Audience'. Meta will analyse the common characteristics of your existing subscribers and then go and find millions of other people on its platform who look just like them. This is far more powerful than manual interest targeting because it's based on your actual conversion data. A 1% Lookalike of your subscribers list will almost always be your best-performing cold audience. You can't do this yet, but it should be your medium-term goal. Once you can, this will be your primary way to scale.

You'll need a solid testing structure...

So, how do you put all this into practice without getting overwhelmed? You need a simple, methodical structure. Rushing into changes based on one day's data is the quickest way to waste money. Here’s a basic campaign structur I would suggest starting with:

Campaign 1: ToFu - Cold Traffic (Objective: Conversions/Signups)

• Budget: Use Campaign Budget Optimization (CBO) set at your $30/day. This lets Meta decide which ad set is performing best and allocate more spend to it automatically.
• Ad Set 1: Testing Interest Theme A (e.g., Competitors).
• Ad Set 2: Testing Interest Theme B (e.g., Media/Magazines).
• Ad Set 3: Testing Interest Theme C (e.g., Broader related concepts).
• Ads: Inside each ad set, have 2-3 different ads. Maybe one is an image, one is a short video, and one is a carousel. Write different headlines and copy for each. Let Meta figure out the winning ad/audience combination.

Let this run. Don't touch it for at least 4-5 days. After a week, look at the results. Is one ad set getting a much lower CPA than the others? Turn off the bad ones. Is one ad creative clearly outperforming the others? Turn off the losers and create new variations of the winner to test against it.

This process of methodical testing and iteration is what drives costs down over time. It's not about finding one magic bullet, it's about a relentless process of elimination and optimisation. I recall one client where we were able to reduce their cost per lead by 84% by testing different audience combinations. It's a grind, but it's a process that works if you stick to it.

Finally, I've put together a summary of my main recomendations for you in a table below. This is basically the action plan I'd follow if I was in your shoes.

This is the main advice I have for you:

As you can probably see, there's quite a lot to dig into here, from the technical weeds of the pixel to high-level campaign stratgey. It can be a bit of a minefield, especially when you're also trying to actually write the newsletter and run your business. This is often where getting some expert help can make a real difference.

Having someone who's navigated these issues hundreds of times can drastically shorten the learning curve, help you avoid common costly mistakes, and implement a professional-grade structure from day one. It means you can focus on creating great content for your subscribers, while we focus on finding them for you efficiently.

If you’d like to have a more in-depth chat about this, we offer a free initial consultation. We could have a proper look at your ads account together and map out a more detailed plan. There's no obligation at all, of course, but it could be really helpful to get a second pair of expert eyes on it.

Hope this has been helpful either way!

Regards,

Team @ Lukas Holschuh