TLDR;
- You are likely being told to use broad targeting on Meta, assuming the algorithm will just find your customers.
- Our analysis of a B2B software campaign with 4,622 registrations shows this is poor advice for specific, high-value markets. In our test in developed countries, broad targeting delivered zero results.
- Lookalike audiences built from high-intent actions (like completed registrations and purchases) drove over 80% of conversions, proving to be far more effective at finding actual B2B software users.
- Below, we have built interactive tools, including a performance simulator and a searchable targeting database, to help you apply these findings.
There is a lot of noise out there about Meta ads strategy, especially for B2B software. The current thinking, pushed by Meta themselves and various market commentators, is to simply go broad. Chuck your creative into the machine, remove all targeting, and let the algorithm work its magic. The theory is that with enough data, the system performs better than you at finding customers.
Does that actually stack up in the real world, especially when you are not selling a cheap t-shirt but a specific piece of B2B software? When every click costs real money and you are trying to find a business owner who needs your solution, is going broad an effective strategy or just a fast way to burn through your budget? That is the question we wanted to examine.
To get a proper answer, we can look at the data from a real, successful campaign we ran. This is a look at a single B2B SaaS account for a product called FeedbackWhiz, which helps Amazon sellers manage their feedback. Over the course of the campaign, we generated 4,622 completed software registrations for a total spend of $11,013.42. That is a blended cost of just $2.38 per registration. The campaign worked. Now we can pick it apart to see what really drove those results, and settle the debate between going broad and using sharp, lookalike-driven targeting.
Does your funnel stage and location change what you pay?
Before we can talk about specific audiences like broad or lookalikes, we need to understand the bigger picture. Not all traffic is created equal. The cost to get a registration is massively different depending on where you are targeting in the world and what part of the marketing funnel you are in. People throw around average CPA figures, but they are meaningless without this context.
Looking at the campaign data, the difference is stark. We ran campaigns targeting various stages: Top-of-Funnel (TOFU) to find new people, Middle-of-Funnel (MOFU) to retarget website visitors, and Bottom-of-Funnel (BOFU) to retarget people who had already started a trial. We also tested different geographic targets: a low-cost worldwide campaign, a mid-tier developed countries group, and a high-cost US-only campaign.
The results vary significantly. A TOFU campaign targeting the US cost $38.57 per registration. The exact same funnel stage, but targeting worldwide, got registrations for just $0.31. That is over 120 times cheaper. Retargeting (MOFU) sits in the middle at a respectable $0.88. This tells you straight away that your geographic strategy is one of the biggest levers you can pull on cost. It also shows that paying to re-engage people who already know you is far more efficient than constantly chasing brand new, cold leads in expensive markets.
Take a look at the chart below. It breaks down the cost per registration for the main campaign variations. You can see how much location and funnel stage influenced the actual cost to get a new user to sign up.
Cost Per Registration by Campaign Strategy
Analysis of cost per completed software registration across different top-of-funnel (TOFU) geographic targets and middle-of-funnel (MOFU) retargeting campaigns from the FeedbackWhiz account data.
What happened when we put Broad and Lookalikes head-to-head?
Knowing that geography matters so much, this is where we can test the theory of going broad. For the developed countries campaign, we have a clean breakdown of the different audiences we tested against each other. This is as close to a controlled experiment as you can get in paid advertising. The campaign objective was 'Complete Registration', the location was set, and the ads were the same across the different ad sets. The only variable was the audience itself.
We tested a mix of audiences: a few different Lookalikes (LLAs) built from high-value customer actions, some detailed interest targeting groups, and a fully broad ad set with no targeting layers at all.
The results were not subtle. Of the 232 registrations generated in this campaign, zero came from the broad ad set. The same goes for most of the interest-based audiences like marketplace interests and eCommerce-related interests—they spent money but delivered nothing. All the conversions came from a handful of specific audiences.
The clear winners were the lookalike audiences. Specifically, a 1% lookalike of users who had already completed a registration drove the majority, pulling in 150 of the 232 registrations. A lookalike of actual purchasers was the next best. The only interest-based audience that worked was small business owners, which makes sense for this product. The chart below breaks down where every single registration came from.
Registration Sources: Developed Countries Campaign
Breakdown of the 232 software registrations generated from the top-of-funnel campaign targeting developed countries, showing which audiences were effective versus those that failed to convert.
If lookalikes are the answer, which ones actually work?
The data shows that for this B2B software in a developed market, lookalikes were the correct path. However, simply using lookalikes broadly is not enough. A lookalike of website visitors is very different from a lookalike of paying customers. The quality of your source audience matters completely.
The data from our experiment gives us a strong indication here. We can rank the audiences that actually worked by the volume of registrations they produced:
- LLA 1% Complete Registration: 150 registrations
- LLA 1% Purchases: 38 registrations
- Small Business Owners (Interest): 27 registrations
- LLA 1% Website Visitors: 13 registrations
- LLA 1% Trials: 4 registrations
There is a clear pattern. The further down the funnel your source audience is, the better the resulting lookalike performs. A lookalike of people who have already completed the desired action is the most effective at finding similar users. A lookalike of paying customers follows closely behind. These are high-intent audiences.
On the other end of the scale, a lookalike of general website visitors performed poorly. It is too broad. A website visitor could be anyone—a competitor, a job seeker, or someone who clicked by accident without showing intent. The algorithm receives a noisy signal and returns a noisy audience.
This provides a clear hierarchy for building and testing audiences. You should always prioritise building lookalikes from data sources that are as close to the final conversion as possible. Start with purchasers, then trial starts or completed registrations, and only then consider wider audiences like video viewers or website visitors.
B2B SaaS Lookalike Audience Hierarchy
Tier 1: High-Intent (Highest Priority)
Lookalikes built from your most valuable customer actions. These are people who have proven their worth. Prioritise these first.
Examples: LLA of paying customers, LLA of completed registrations, LLA of users who activated a core feature.
Tier 2: Mid-Intent (Test for Scale)
Lookalikes from audiences who have shown interest but haven't converted yet. Use these to expand your reach once you have proven Tier 1 works.
Examples: LLA of trial starts, LLA of people who initiated checkout, LLA of pricing page visitors.
Tier 3: Low-Intent (Use with Caution)
Broadest sources, highest risk of poor quality. Only test these if you have a large budget and have exhausted higher-intent options.
Examples: LLA of all website visitors, LLA of video viewers, LLA of social media page engagers.
How does this change your budget and expectations?
Understanding this performance difference directly impacts how you should plan your budget and what results you can realistically expect. The difference between a campaign that works and one that burns cash often comes down to allocating spend to the right audiences from the start. To make this concrete, we have included an interactive simulator below. You can plug in a monthly registration goal and it will show a projected breakdown of where those registrations might come from, based on the performance distribution from our developed countries campaign.
Lookalike Performance Simulator
Enter your target number of monthly registrations to see a projected breakdown based on our B2B SaaS campaign data.
Projected Registration Sources
Is broad targeting ever a good idea?
Given the results, it is fair to ask if there is ever a place for broad targeting. The answer is conditional. It depends on your cost per click. The reason broad targeting failed in the developed countries campaign is likely because clicks there are relatively expensive. Meta's algorithm did not have enough inexpensive data to learn from quickly. It spent budget exploring, found no immediate traction with the cold B2B audience, and was turned off.
What about the worldwide campaign? As noted earlier, the cost per registration there was an economical $0.31. When you can buy thousands of impressions and hundreds of clicks for minimal cost, you provide the algorithm with a large volume of data to test. It can learn rapidly and eventually find pockets of converting users even within a large, undefined audience.
This suggests a two-speed strategy. In high-cost, competitive markets like the US, UK, or Western Europe, you cannot afford to let the algorithm guess blindly. You need to provide strong signals using high-intent lookalike audiences. When targeting lower-cost regions, or once your pixel has accumulated thousands of conversions and operates reliably, broad targeting becomes a viable option to test for expansion.
To help model your own numbers, we have built another tool. This one models the potential outcome of a set budget across the different geographic strategies we have discussed. You can see how choosing a low-cost worldwide audience versus a high-cost US audience leads to different outcomes in volume and cost.
Geographic CPA Scenario Modeler
Select a monthly ad spend and a target geography to see the projected impact on registrations and cost, based on our campaign's performance data.
What is the practical plan?
This research points to a clear, data-driven approach for targeting on Meta ads for B2B SaaS. Success involves applying the correct strategy for the right situation rather than relying on a single approach. Below is a blueprint for how to approach this, moving from initial testing to scaling.
B2B SaaS Targeting Blueprint
A phased approach to audience testing and scaling on Meta ads, based on the findings from the FeedbackWhiz campaign data.
| Phase | Primary Goal | Recommended Audiences | Geographic Focus |
|---|---|---|---|
| 1. Validation | Get initial traction and prove campaign viability efficiently. | Start with your highest-intent audiences: LLA of Purchasers, LLA of Completed Registrations. Add 1-2 specific interests (e.g., small business owners). | Developed countries or specific high-value markets (e.g., US, UK, DE). |
| 2. Expansion | Expand reach beyond the core converting audiences. | Test mid-intent LLAs (e.g., trial starts, website visitors). Continue to run winning audiences from Phase 1. Avoid broad audiences. | Continue with developed countries. Keep budgets focused. |
| 3. Scaling | Maximise registration volume at the lowest possible CPA. | Launch a new campaign to test broad targeting. Requires a mature pixel with thousands of conversions. Run alongside your LLA campaigns. | Worldwide (excluding low-income countries). This is where broad targeting has the best chance to work. |
To help find specific interests for Phase 1, a searchable database is provided below. This includes interest categories that work well for B2B software clients, including options mentioned in the campaign strategy documents. You can filter by theme to find ideas relevant to your product.
B2B Interest Targeting Database
A searchable list of detailed targeting ideas for B2B SaaS, inspired by successful campaigns. Use the filter to find relevant categories.
| Interest / Behaviour | Category | Best For |
|---|
Final Thoughts
The pressure to go broad on Meta is significant, but as this analysis shows, it is not a universal solution. For a specific B2B software product targeting users in competitive, developed markets, broad targeting was ineffective. The path to efficient user acquisition came from a disciplined, layered approach. It started with specific, high-intent lookalike audiences that provided the algorithm with clear signals about qualified customers.
This does not mean broad targeting is useless. It has a place when aiming for scale in low-cost environments where the algorithm can learn affordably, or when your pixel is mature enough to predict conversions with high accuracy. For most B2B businesses starting out or seeking predictable growth, the evidence indicates starting specific. Build your foundation on data you own, such as existing customers and prospects. Once you master that, testing broad audiences becomes an option.
Engineering a system like this—finding the right audiences, building effective creative, and managing budgets across different stages and geographies—requires careful execution. If you have struggled to make Meta ads work for your software business and feel your budget is not optimized, a structured approach helps. We offer a free strategy consultation where we review accounts and help map out a practical plan to turn ad spend into a predictable growth engine.
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.