What Meta Video Ads Cost a B2B Software Company: Real Numbers From a $50-a-Day Test
A Meta video ad program for a B2B software company has three costs: the ad spend you pay Meta, the fee to manage it, and the videos themselves. A first test doesn’t need the $3,000 to $10,000 a month that most agency guides quote. One software company we work with ran a two-week test at $50 a day, spent $690.90, and signed four new customers from it.
That’s a small test, and I’ll be straight with you about what it proves and what it doesn’t. But it’s a real account with real numbers, which is more than most pricing guides give you.
The three costs
1. Ad spend, paid to Meta. This is the money that actually buys the ads. It goes from your card straight to Meta, not through an agency, and nobody should be marking it up. For a first test, we usually recommend $750 to $1,500 a month across Facebook and Instagram.
2. Management. Someone has to build the campaign, set up the tracking, watch it every day, and change what isn’t working. Published pricing from SaaS ad agencies runs from flat retainers starting around $1,250 a month for one channel, often with a $1,000 to $2,000 setup fee, to 10% to 20% of your ad spend. Ours is $990 a month for our first three software clients, for their first six months, then $1,500 a month. There’s no setup fee.
3. The videos. This is the cost founders forget, and the one AI has changed the most. A studio shoot or a batch of paid creator videos is its own bill, and you need fresh ones every month. In our program, the founder talks to a phone camera for about 20 minutes and sends a few screen recordings of the product. We turn that into 6 to 9 captioned ads in the first month and 4 to 6 new ones each month after. The editing is part of the monthly fee. If nobody on your team wants to be on camera, an AI presenter can read the scripts instead, for $150 a month more.
What $50 a day bought
Here’s the test, start to finish:
- Who: a B2B company selling field-service software into the trades
- When: 14 days, ending in July 2026
- Budget: $50 a day
- Spend: $690.90
- Demo requests: 14, at $49.35 each
- New customers: 4 of the 14, a 28.6% close rate
- Cost per customer: $172.73 in ad spend
Three decisions did most of the work. Every ad was a video, because buyers want to see the product working before they’ll book a demo. Every ad pointed at one action, a demo request on the company’s own site, and the campaign optimized for that and nothing else. And the budget held steady at $50 a day rather than reacting to every slow afternoon.
Now the honest part. This is one account, one market, two weeks, and 14 leads. That’s enough to show the channel can work on a small budget. It isn’t enough to promise you $49.35 demos. Your number will depend on your market, your offer, and your videos. The full case study shows how each figure was calculated.
Why the guides say you need $3,000 to $10,000 a month
They aren’t making the number up. It comes from how Meta’s system learns.
When you launch a campaign, it enters what Meta calls the learning phase. The algorithm shows your ad to different kinds of people, watches what happens, and adjusts. Meta’s guidance is that an ad set needs roughly 50 optimization events in a week to get through learning and settle down. If your event is a demo request and each one costs $50, that’s $2,500 a week, or about $10,000 a month, just to hand Meta the volume it asks for.
So agency guides size their budgets around that. One SaaS agency’s guide says campaigns under $3,000 a month usually never make it out of learning. Another puts the minimum test budget at $10,000 a month. Keep in mind that both come from agencies that sell ad management, so read them as vendor benchmarks, not independent research.
Where they’re right: if you need stable, predictable numbers fast, or you want to test several audiences and offers at once, a small budget won’t get you there. A $50-a-day campaign often stays in learning. Results come in bumpy, and you judge the campaign by demos and signed customers over weeks, not by the dashboard on a Tuesday.
Where they’re wrong for most early-stage companies: the first question isn’t “is this campaign stable?” It’s “can this channel produce demos worth paying for?” A small, steady test answers that question for a few hundred dollars to a couple thousand. You can raise the budget once you have the answer.
What moves your cost per demo
Targeting on Meta isn’t a menu of interests anymore. After Apple’s 2021 privacy changes and Meta’s own cuts to detailed targeting, most of what decides who sees your ad, and what each demo costs you, comes down to things you control.
Your video is your targeting. Meta decides who sees an ad partly by watching who stops for it. An ad that opens with the exact problem your buyers bring up on demo calls gets shown to more people who have that problem. An ad that opens with your logo and a tagline tells Meta almost nothing about who to find.
Your offer is a quality dial. A short demo form gets more requests. A longer one with a qualifying question gets better ones. If you’re getting demos that go nowhere, add a question. If you’re getting none, take one away.
Your conversion teaches the system what “good” means. Meta gets very good at producing whatever you tell it to optimize for. Optimize for clicks and it finds people who click on everything. Optimize for a completed demo request on your own page, then send your closed deals back to Meta through the Conversions API, and it starts looking for people who resemble your actual customers. That last step is the one most accounts skip.
Fresh videos keep the cost from creeping up. When one ad works, the move is to build new versions of it, with a new opening line or a different proof point, rather than run the same video until people tune it out. That’s why we plan new ads every month.
Where AI cuts the cost
A few years ago, testing ten versions of an ad meant paying a copywriter and an editor for ten versions. That’s a big part of why video ads used to be out of reach for small software companies. Now AI handles the slow parts:
- It pulls the exact words buyers use out of sales call recordings, demo notes, and reviews. Those words become the first line of the ad, because people stop scrolling when they hear their own problem said back to them.
- It drafts scripts on a proven structure in minutes, and a person edits them for accuracy and tone.
- It helps turn one winning ad into several new versions.
- It reads the results and spots which openings hold attention and which ads bring demos that actually book.
What it doesn’t do is decide what’s true about your product, what Meta’s policies allow, or which demos were real. That still takes a person who knows the business and the platform.
How to set your first month’s budget
Work backward from a customer, not forward from a budget. It’s the same method we use to set Google Ads budgets for local businesses:
- Decide what you can pay to win a customer. Start from what a new customer is worth to you in their first year, and pick the share of that you’d happily spend to get them.
- Divide by your close rate. If you close one demo in four, you can afford to pay a quarter of that number for each demo.
- Multiply by the number of demos you want. That’s your monthly ad spend.
If the math comes out under $750 a month, a test will take a long time to tell you anything. If it comes out well above $1,500, start smaller anyway, and raise the budget once the first month shows you what a demo really costs in your market.
If you’d like a real number for your company, here’s how our software program works, and every price we charge is published.
