Hi there,
Thanks for reaching out!
I’ve had a good look at the problem you're facing with your Meta campaign, and I'm happy to give you some initial thoughts. To be brutally honest, I think you're focusing on the wrong problem. The fact that your impressions are skewed 99% to Android and 90% to Facebook isn't the actual issue; it's a symptom of a much deeper strategic flaw. You're asking how to stop the car from veering left without being willing to touch the steering wheel.
The core of the issue is your reliance on "best practice" without understanding the mechanics behind it. Right now, you're paying the world's most sophisticated advertising machine to actively find you the worst possible audience for your product. Let's get into why that's happening and how to fix it properly.
We'll need to look at why 'best practice' is failing you...
First off, let's tackle this idea of "broad targeting and consolidated campaigns" being the holy grail. It's a myth pushed by people who either don't understand the nuances or are managing accounts with millions in spend and years of pixel data. For most businesses, especially those still scaling, going broad from the get-go is like setting your money on fire. It only works when your Meta Pixel has so much high-quality conversion data that it can build an incredibly accurate profile of your ideal customer on its own. Without that history, you're just telling the algorithm: "Here's my money, go find anyone on the planet who might, by sheer chance, complete a registration for the lowest possible price."
And that’s exactly what it's doing. You've set the objective to 'Conversions' with the goal of 'Complete Registration'. The algorithm's one and only job is to achieve that goal as cheaply as possible. It has discovered that the cheapest, most plentiful cohort of users who will complete this action are people using Android devices, browsing the Facebook feed. It’s not a bug; it’s the system working perfectly as designed.
You're essentially running an awareness campaign disguised as a conversion campaign. You are paying Meta to find the people whose attention is cheapest because they are the least likely to be targeted by other advertisers for valuable actions like purchasing. These users might be more prone to clicking and filling out forms without any real intent. The iOS users, on the other hand, are more difficult and expensive to track and convert since Apple's ATT privacy changes. So the algorithm, taking the path of least resistance, avoids them entirely. You're getting a high volume of a low-quality event from a low-cost audience segment.
Asking to fix this without splitting the campaign by device or placement is asking the impossible. You can't force the algorithm to spend more money on more expensive users (iPhone/Instagram) when its primary directive is to find the cheapest conversions. You need to regain control. The whole "consolidated is best" thing is guidance, not gospel. When the machine's optimisation logic goes against your actual business goals, you have to intervene and give it stricter rules. But before we even do that, we need to give it much, much better inputs.
I'd say you need to define your customer by their nightmare, not their device...
The root of this problem is that you haven't told Meta who you actually want to reach. "Broad" tells them nothing. Forget the sterile, demographic-based profile you might have in a document somewhere. "iPhone users aged 25-45" tells you nothing of value. It leads to the generic, wallpaper-paste ads that you're probably running right now, which get ignored by everyone you actually want to reach.
To stop burning cash, you must define your customer by their pain. By their specific, urgent, expensive, career-threatening nightmare. Your Ideal Customer Profile (ICP) isn't a person; it's a problem state. You need to become an obsessive expert in that problem.
Who is registering for your app? Why do they need it? What frustration in their life is so acute that they'd stop scrolling through cat videos to sign up for something new? Is your Head of Sales client not just a job title, but a leader terrified of her team missing quota because their CRM is a mess? Is your user a student overwhelmed by revision, staring at a mountain of notes with no idea where to start? That is the nightmare. That is what you sell a solution to.
Once you've isolated that nightmare, you can find them. Where do they hang out online? What niche podcasts do they listen to? What industry newsletters do they actually open? What specific software tools (like HubSpot, Asana, or Slack) do they already pay for? Are they members of specific Facebook Groups? Do they follow particular influencers on Instagram? This intelligence isn't just data; it's the blueprint for your entire targeting strategy. You have to do this work first. Without it, you have no business spending another pound on ads.
Here’s how you can start thinking about this. Let's imagine you have a project management app for small creative agencies.
| Vague Demographic ICP | Nightmare-Based ICP | Resulting Targeting Interests |
|---|---|---|
| Small business owners in the UK, aged 30-50. | The agency owner who lies awake at 3 AM worrying about missing a client deadline, whose team is threatening to quit because of chaotic workflows, and who feels like they're constantly drowning in admin instead of doing creative work. |
-> Interests: Adobe Creative Suite, Dribbble, Behance -> Behaviours: Business Page Admins, Facebook Page Admins -> Software they use: Slack, Asana (competitor), Trello (competitor) -> Publications they read: Creative Review, Awwwards -> Job Titles (if on LinkedIn): Founder, Creative Director, Agency Owner |
This is the level of detail you need. Now you have a list of concrete interests you can start testing, instead of just throwing your message into the void and hoping for the best.
You probably should rebuild your targeting from the ground up...
Right, so let's scrap the broad campaign for now. We're going to build a proper, structured account that guides users from being vaguely aware of you to becoming a registered user. I've audited hundreds of Meta accounts, and the ones that succeed consistently use a funnel-based approach. It seems a lot of people just test random audiences that dont align with their goals. Here’s the priority list I use for testing audiences. You build from the top down as you gather more data.
META ADS AUDIENCE PRIORITISATION
Top of Funnel (ToFu) - Prospecting for new users:
- Detailed targeting (interests, behaviours): This is your starting point. Use the 'Nightmare ICP' work you just did to build ad sets around themes of related interests. Don't just lump everything together. Test one theme per ad set (e.g., one for competitor software users, one for industry publication readers).
- Lookalike audiences: Once you have enough *quality* conversion data (I'd say at least 1,000 'Complete Registration' events from your highly targeted campaigns), you can start building lookalikes. You want to build them from your best user signals. In order of value:
- Lookalike of paying customers (if applicable)
- Lookalike of 'Complete Registration' events
- Lookalike of users who initiated checkout/added payment info
- Lookalike of all website visitors
- Lookalike of 50% video viewers
- Broad targeting: This is the LAST thing you test, only once you have thousands of quality conversions and your other campaigns are consistently profitable.
Middle of Funnel (MoFu) - Engaging warm leads:
- Website visitors (last 30-90 days, excluding converters)
- Users who watched 50%+ of your video ads (last 30 days)
- Instagram/Facebook page engagers (last 90 days)
Bottom of Funnel (BoFu) - Closing the deal:
- Users who visited the registration page but didn't complete (last 7-14 days)
- Users who added to cart / initiated checkout (if applicable)
For a new account, you start with ToFu using detailed targeting. This is non-negotiable. You need to feed the machine good, clean data about who you want. You need at least 100 people in a custom audience for retargeting, but realistically you want thousands before it's effective. So you must start with prospecting to get that traffic.
Here's a sample campaign structure to make it concrete. This gives you control and clarity.
| Campaign (Objective: Conversions) | Ad Set | Audience | A Note on Placements |
|---|---|---|---|
| C1 - ToFu - Prospecting | Ad Set 1: Interest Group A | Interests: Asana, Trello, Monday.com (Competitors) | For now, I'd split them. Create one ad set for iOS/IG, one for Android/FB etc. Or at the very least, use the 'Breakdown' view in Ads Manager to see performance and confirm if the cheap Android/FB traffic is actually valuable. You are refusing the main tool for control here. You have to be willing to split if the data shows a huge performance gap. |
| Ad Set 2: Interest Group B | Interests: Creative Review, Behance, Dribbble (Publications/Communities) | ||
| C2 - MoFu/BoFu - Retargeting | Ad Set 3: Warm Audiences | Website Visitors (30d) + Video Viewers (50%, 30d). Exclude 'Complete Registration'. |
You run these campaigns long-term. You test new interest groups in the ToFu campaign. You turn off the ones that dont perform after they've spent enough money (a common rule is 2-3x your target cost per registration). This is how you systematically find winning audiences instead of just hoping broad targeting figures it out for you.
You'll need a message they can't ignore...
Having the right targeting is only half the battle. If you're showing the perfect audience a bland, feature-led ad, they'll just scroll past. Your ad copy and creative needs to speak directly to the nightmare you identified earlier.
Most B2B/SaaS ads are terrible. They just list features. "Our platform integrates with X and has Y feature." So what? Nobody cares. You need to sell the outcome, the relief from the pain.
For a product that needs a registration, the best framework is Before-After-Bridge. You paint a picture of their current "hell" (the Before), show them the "heaven" they want (the After), and position your app as the thing that gets them there (the Bridge).
Let's stick with our project management app example:
| The "Before" Ad (What most people write) | The "After" Ad (What you should write) |
|---|---|
|
Headline: The Best PM Tool for Creatives Body: Our platform has Gantt charts, task management, and time tracking. Integrate with your favourite tools. Register for a free trial today! |
Headline: Another project veering off the rails? Body: (Before) Your team's drowning in emails, client feedback is lost in Slack, and you have no idea if you're actually profitable on this project. Sound familiar? (After) Imagine knowing exactly where every project stands in 30 seconds. All comms, files, and deadlines in one place. Your team is happy, your clients are impressed. (Bridge) Our platform is the bridge that takes you from chaos to calm. Register for a free, no-risk trial and get your first project organised in 15 minutes. |
See the difference? The first one is about the tool. The second one is about the customer and their problem. It hooks them with a relatable pain point and sells them a transformation, not software. This is what compels a high-value user to stop and register, not just a random person on an Android phone who likes clicking buttons.
Speaking of registrations, one of my clients, a B2B software company, was struggling to get users. We restructured their campaigns with this kind of targeting and messaging, and we drove 4,622 registrations at just $2.38 each. It's possible, but it requires this foundational work.
I'd say you need to reconsider the numbers that matter...
This brings me to my final point. Your entire post is framed around the cost of impressions and a platform split. You're optimising for the wrong metrics. The real question isn't "Why are my impressions cheap on Android?" but "What can I afford to pay to acquire a customer who will actually stick around and be valuable?"
You need to understand your Customer Lifetime Value (LTV). If you dont know this number, you are flying blind. Let's run a quick, hypothetical calculation. This is the maths that unlocks intelligent, aggressive growth.
Average Revenue Per Account (ARPA): What do you make per registered user, per month? (Let's say £20)
Gross Margin %: What's your profit margin? (Let's say 80%)
Monthly Churn Rate: What percentage of users do you lose each month? (Let's say 5%)
The calculation is: LTV = (ARPA * Gross Margin %) / Monthly Churn Rate
LTV = (£20 * 0.80) / 0.05
LTV = £16 / 0.05
LTV = £320
In this example, each user is worth £320 in gross margin to you. A healthy business model aims for at least a 3:1 LTV to Customer Acquisition Cost (CAC) ratio. This means you can afford to spend up to £106 to acquire a single customer.
Suddenly, does it matter if a registration from an iPhone user on Instagram costs you £15, while one from an Android user on Facebook costs £2? No, not if the iPhone user is 10 times more likely to become a long-term, paying customer. You might find that 99% of your cheap Android registrations churn within a week, making their true value zero. You'd be better off getting one expensive, high-quality user than 50 cheap, worthless ones. Focusing on the cost per registration in isolation, without considering the downstream value, is a classic mistake that leads to optimising for vanity metrics while your business slowly bleeds out.
This is the main advice I have for you:
This has been a lot to take in, I know. It's a fundamental shift from what you're currently doing. To make it clearer, I've broken down my main recommendations into a table for you to implement.
| Your Current Problem | My Recommendation | Why This Is a Better Approach |
|---|---|---|
| Over-reliance on broad targeting is attracting low-quality traffic. | Pause the broad campaign immediately. Rebuild from the ground up starting with hyper-specific Detailed Targeting (interests/behaviours). | This feeds the algorithm high-quality data about who your real customers are, forcing it to find more valuable users instead of just the cheapest ones. |
| The algorithm is defaulting to the cheapest placements (Android/Facebook). | Stop refusing to use the tools available. Either manually select placements or split ad sets by device/platform to regain control and force spend where you want it. | It allows you to test if the more expensive placements (iPhone/IG) deliver more valuable users, overriding the algorithm's flawed "cheapest is best" logic. |
| Your audience definition is likely too vague, leading to generic ads. | Define your Ideal Customer Profile by their "nightmare" – their most urgent pains and frustrations. | This enables you to write powerfully resonant ad copy (Before-After-Bridge) and select targeting interests that your true customers engage with. |
| Your campaign structure is a single, uncontrollable "black box". | Implement a proper ToFu/MoFu/BoFu funnel structure with separate campaigns for prospecting and retargeting. | This provides clarity, control, and a systematic way to test audiences and move users logically towards conversion, rather than just hoping they show up. |
| You are focused on surface-level metrics like impression splits and CPL. | Calculate your LTV and determine a maximum affordable CAC. Shift your success metric from cheap registrations to profitable customer acquisition. | This focuses your budget on acquiring users who actually make you money, freeing you from the tyranny of cheap, low-quality leads. |
As you can see, fixing your campaign isn't about finding a secret button to tick inside Ads Manager. It's about a complete overhaul of your advertising strategy. It requires research, structure, discipline, and a deep understanding of your customer.
Doing this properly takes a significant amount of time and expertise, especially when you're also trying to run your business. It involves continuous testing, analysis, and optimisation that can easily become a full-time job. This is often where businesses decide to bring in expert help.
If you'd like to go through your account together and map out a specific, actionable plan based on what we've discussed, we offer a completely free, no-obligation initial strategy session. We could audit your campaigns in real-time and show you exactly where the opportunities are.
Either way, I hope this detailed breakdown gives you the clarity you need to get your ads back on track.
Regards,
Team @ Lukas Holschuh
