Summary of approach:
- Most B2B software founders assume you must target US-only broad audiences to get scale, burning thousands on high CPMs.
- Our single-account dataset of 4,622 registrations shows US registrations cost $38.57, while Worldwide targeting dropped the cost to $0.31—a 124x difference in top-of-funnel acquisition cost.
- Stacking Worldwide top-of-funnel traffic with highly constrained 1% Lookalike audiences and mid-funnel retargeting yields a blended $2.38 CPA without sacrificing backend trial quality.
- Below, you will find interactive calculators, geographical benchmark columns, and our exact audience targeting framework derived directly from the data.
There is a lot of rubbish advice floating around regarding B2B software acquisition. If you listen to most forum threads, they will tell you to restrict your Meta ads entirely to the US market, turn on broad targeting, and let the algorithm sort it out. I have audited enough accounts to know this is a brilliant way to burn through your runway without seeing a single qualified trial.
You cannot just hand Mark Zuckerberg your credit card, say "find me software users," and expect cheap registrations. You have to engineer the system.
In this investigation, I am setting out to answer a specific question: How does geographic constraint and specific audience layering (small business owners and lookalikes) actually affect top-of-funnel (TOFU) registration costs for B2B SaaS? We are looking strictly at how you scale past 4,000 registrations at a blended cost of $2.38, rather than paying the typical £50+ per lead most agencies accept as standard.
The Dataset and Methodology
Before we look at the numbers, let us establish exactly what we are looking at. This is not some aggregated, smoothed-out industry average. This is a forensic analysis of a single Meta Ads account over a specific timeframe, promoting a B2B software tool designed to help Amazon sellers manage and automate their feedback and reviews.
The raw dataset comprises $11,013.42 in total ad spend, generating exactly 4,622 completed registrations at an average Cost Per Action (CPA) of $2.38. The account utilises a multi-stage funnel approach, splitting campaigns across Top of Funnel (TOFU) prospecting, Middle of Funnel (MOFU) retargeting, and Bottom of Funnel (BOFU) trial and purchase conversion.
My methodology here is straightforward. I have extracted the raw campaign spend, reach, frequency, and registration volumes from this specific account. I will calculate the cost differentials across geographic locations, compare the efficacy of lookalike audiences against broad interest groups, and analyse the distribution of ad creative performance. We will rely on the math, not the theory.
The Geographic Premium: Why US-Only Targeting Bankrupts Startups
Let us tackle the biggest misconception first. The standard playbook says you should only target high-income, English-speaking countries—specifically the United States—because that is where the purchasing power sits. The data from this account paints a brutally clear picture of what that assumption actually costs you.
I calculated the CPA for the exact same objective (Completed Registrations) across three distinct geographic TOFU campaigns in the dataset:
- LH - TOFU (US): $925.60 spent for 24 registrations.
- LH - TOFU (Developed Countries): $1,337.42 spent for 232 registrations.
- LH - TOFU (Worldwide): $336.75 spent for 1,083 registrations.
Do the math on that. The US campaign yielded a $38.57 CPA. The Developed Countries campaign (which typically includes the UK, Australia, Germany, and others) dropped to $5.76. The Worldwide campaign? An absurdly low $0.31 per registration.
Top-of-Funnel Registration CPA by Geography
Audited registration costs across US, Developed, and Worldwide targeting within the dataset.
You are looking at a 124x difference in acquisition cost between the US and Worldwide targeting. Now, I know exactly what you are thinking. "Worldwide traffic is rubbish. It is all bots and unqualified leads from developing nations who will never pay for a subscription."
Fair play, that is a valid concern. The conversion rate from free registration to paid trial will absolutely be lower on worldwide traffic. But you have to look at the unit economics. If a US registration costs 124 times more, the US lead has to convert to a paid user at a rate 124 times higher just to break even on the ad spend. That almost never happens. Even if 95% of your worldwide traffic is entirely unqualified, the sheer volume of cheap data you feed the Meta pixel allows it to optimise far faster than starving it with a tiny $38.57 trickle of US leads.
By blending the $0.31 Worldwide TOFU registrations with the $5.76 Developed registrations, you get a massive pool of users into your software ecosystem. You then rely on your product onboarding and Middle-of-Funnel retargeting to filter the serious buyers from the time-wasters. This is how you achieve a blended account CPA of $2.38.
Targeting Diagnostics: Why Broad Burns Budget
If you are selling consumer goods, broad targeting works. If you are selling a highly specific B2B tool for Amazon sellers, going broad is an excellent way to pay Facebook to show ads to bored teenagers. You need to narrow the field.
Looking at the Ad Set breakdown within the dataset, we tested various detailed targeting options against Lookalike (LLA) audiences. The raw registration volume reveals a stark hierarchy of intent.
The "Small Business Owners" detailed targeting interest drove 27 registrations. That is a decent baseline. However, when we examine the Lookalike audiences based on actual pixel data, the performance shifts dramatically. The 1% Lookalike of Complete Registrations drove 150 signups. The 1% Lookalike of Purchases drove 38.
Why does this happen? "Small Business Owner" is a massive, noisy category on Meta. It includes freelance graphic designers, local plumbers, and cafe owners—almost none of whom need an Amazon seller feedback tool. A 1% Lookalike audience seeded by people who have actually completed a registration or made a purchase forces the algorithm to find users with the exact behavioral footprint of your best customers.
B2B Meta Ads Audience Prioritisation Framework
Layer 1: High-Intent Lookalikes (Primary Focus)
1% LLA of Purchases (38 Regs) and 1% LLA of Complete Registrations (150 Regs). Requires minimum 100 pixel events. Highest quality prospecting available.
Layer 2: Niche B2B Interests (Data Gathering)
Targeting "Small Business Owners" (27 Regs) or specific marketplace interests. Use this only when starting out to feed the pixel enough data to build Layer 1.
Layer 3: Broad Targeting (Avoid for Niche B2B)
Turned off in this dataset due to underperformance. Broad targeting fails when the Total Addressable Market (such as Amazon Sellers) is a tiny fraction of the general population.
The lesson here is simple. If you are starting from zero, you have no choice but to use detailed targeting like "Small Business Owners" or "Marketplace Interests" to get your first 100 conversions. The second you cross that threshold, you must build 1% Lookalike audiences and shift your budget there. It is the only way to scale without your CPA spiralling out of control.
Creative Concentration: The Power of Pain-Led Copy
Let us talk about the ads themselves. You can have the most sophisticated targeting structure in the world, but if your creative is dodgy, you will still fail. The dataset shows a massive concentration of results in just a few winning ads.
I looked at the performance split across the seven tested ad creatives. Two ads absolutely dominated the entire campaign.
Registration Volume by Ad Creative
The disproportionate impact of winning creatives. Two ads generated over 80% of the measured registrations.
The ad titled "Automate your Amazon feedback" drove 1,046 registrations. In stark contrast, "Email and alerts" drove a pathetic 2 registrations. That is a 523x difference in volume.
Why the massive discrepancy? It comes down to pain. "Email and alerts" is a feature. Nobody wakes up in a cold sweat worrying about their lack of email alerts. But "Automate your Amazon feedback" is a direct solution to a very specific, painful nightmare for an Amazon seller: getting banned or losing rank due to poor reviews they could not manage manually. Your Ideal Customer Profile is not a demographic; it is a problem state. The creative that speaks most directly to that problem state will win the auction, every single time.
If you do not aggressively cut the losers—such as turning off the "Email and alerts" ad—you are just giving Mark Zuckerberg your money for nothing. We use the aggregate performance across ad sets to ruthlessly decide what to keep and what to bin.
The Funnel Architecture: MOFU and BOFU Compounding
Getting cheap clicks at the top of the funnel is useless if they do not eventually buy. This is why a single campaign strategy almost always fails for software. You need a multi-stage funnel. Let us look at the mid and bottom-funnel numbers from the dataset.
We ran a specific Middle-of-Funnel (MOFU) campaign targeting website visitors who had not yet registered. This campaign generated 1,122 completed registrations at a cost of $0.88 each. It spent $987.93 and reached 18,805 people.
Notice the frequency here: it sits at a massive ►17.46. Most amateurs panic when their frequency goes above 3. In retargeting, especially B2B software where the sales cycle is considered and requires trust, high frequency is not just acceptable—it is required. People do not buy enterprise or business software on the first click. You have to stay on their mind until the business need becomes urgent enough to act.
Further down, we have the Bottom-of-Funnel (BOFU) campaigns targeting people who started a trial to push them to purchase, or those who registered to push them to a trial. The BOFU Trials campaign drove 156 conversions at $2.02 each. The BOFU Purchase campaign drove 36 conversions at $5.71 each.
Account Diagnostic & Action Blueprint
How to restructure your ad account based on the data findings.
| Audit Stage | Common Failure Point | Diagnostic Action Item | Data-Backed Impact |
|---|---|---|---|
| 1. Geographic Targeting | Restricting TOFU exclusively to the US market. | Open a separate 'Worldwide' campaign to feed the pixel cheap data. | CPA drops from $38.57 to $0.31 |
| 2. Audience Definition | Using broad targeting or overly generic interests. | Shift budget to 1% Lookalike Audiences of Complete Registrations. | 5.5x more volume than 'Interests' |
| 3. Funnel Structure | Running a single campaign for all traffic temperatures. | Isolate MOFU retargeting; allow high frequency (15+). | $0.88 MOFU CPA |
| 4. Creative Testing | Leaving underperforming feature-led ads running. | Bin the losers quickly. Concentrate spend on pain-led copy. | 97% of volume from top 3 ads |
This multi-stage architecture is exactly how you achieve a blended account CPA of $2.38. You let the TOFU Worldwide campaigns bring in massive volume at pennies on the dollar. You use US and Developed targeting to sprinkle in high-quality intent. Then, you aggressively retarget all of them with MOFU and BOFU campaigns to squeeze out the actual registrations and trials.
The Unit Economics of Scale
I see founders obsessed with driving their CPL down to zero. That is the wrong metric. The real question is: How high a CPA can you afford to acquire a great customer? This comes down to Lifetime Value (LTV). If your users bring in more money, you can spend more to acquire them and still be highly profitable.
The math is not difficult, but it is often ignored. You calculate your Average Revenue Per Account (ARPA), factor in your gross margin, and divide by your monthly churn rate. If your LTV is high enough, a $38.57 US registration is not a disaster—it is just a different tier of unit economics compared to a $0.31 worldwide registration.
To help you visualise this, I have built a quick modeler based on the logic we use when auditing accounts. Plug in your own hypothetical numbers to see your allowable acquisition cost.
Allowable Customer Acquisition Cost (CAC) Modeler
Adjust the sliders to model your unit economics. Assuming a standard 80% gross margin and a healthy 3:1 LTV to CAC ratio.
When you know your allowable CAC, you stop worrying about getting $0.31 registrations and start focusing on the quality of the funnel. The reason we scaled this account to 4,622 registrations at $2.38 is not because $2.38 was the magical breaking point. It is because the mix of cheap global data to season the pixel, combined with targeted lookalikes and heavy MOFU retargeting, produced a blended cost that fit perfectly into the client's allowable CAC model.
It is not just about setting up an ad and hoping for the best. It is about understanding your audience's painful nightmare, optimising your geographic targeting to feed the pixel efficiently, ruthlessly culling losing creatives, and fine-tuning your landing page offer (such as a free trial without credit card friction).
If you are tired of burning budget on broad interests and want a data-driven system engineered for actual revenue growth, we can definitely help you do this. I would like to invite you to schedule in a free strategy consultation. We will have a look at your Facebook and Instagram ad account together, figure out the right funnel strategy for you, and see if we can get your unit economics sorted. I look forward to speaking to you on that call.
Cite this Research & Data
If referencing this research, data, or strategy in your publication, please use the attribution below.
Lukas Holschuh
Founder, Growth & Advertising Consultant
Great campaigns fail without expertise. Lukas and his team provide the missing strategy, optimizing your entire advertising funnel—from ad creatives and copy to landing page design.
Backed by a proven track record across SaaS, eLearning, and eCommerce, they don't just run ads; they engineer systems that convert. A data-driven partnership focused on tangible revenue growth.