In August 2026 we spent $104.24 on Meta ads for Gantra, our own AI growth team SaaS, across Spanish and English campaigns. The headline numbers: $1.84 per registration in Spanish-speaking LATAM, zero registrations in English-speaking markets with the same creatives, and a conversion lesson downstream that mattered more than either. Most Facebook-Ads-for-SaaS content publishes averages; averages hid every decision that mattered for us, so here are the splits.
This is first-party data from our own ad account. Small budget, honestly reported, including what failed.
The setup
Two campaign generations. V2: image creatives with a price-led hook, ES and EN variants of the same designs, optimized for registrations. V3: the same winning visual themes reframed around a “$1 unlocks your full audit” offer, again ES and EN. Budgets between $5 and $10 per day per campaign. Deliberately small: the goal was signal, not scale.
What the ES campaigns did
The Spanish campaign produced 9 registrations at $1.84 each, concentrated in Perú and Argentina. Creative-level data was decisive: of four visual concepts, one theme (a physical scoreboard metaphor for the site audit) drove the majority of conversions, and knowing the winner shaped everything we made afterward. Registration quality was real, users completed onboarding and ran audits.
Then the honest downstream number: conversions to a $129/month plan from that cohort: zero. The ad performed; the price-market fit underneath it didn’t. A $1.84 signup from a market where $129 is a meaningful monthly cost is not a $1.84 customer, it’s a cheap lead into an expensive product. That mismatch, not the ads, was our actual problem to solve.
What the EN campaigns did
The same creatives, translated and targeted at English-speaking markets: zero registrations. Not expensive ones, none. Two readings, both useful. First, CPMs in those markets run multiples of LATAM’s, so small budgets buy far fewer impressions. Second and more important: the hook didn’t translate. A price-led offer lands where price is the active pain; in markets where $129 is an unremarkable SaaS expense, “cheap” is not a reason to click. The creative wasn’t bad, it was answering a question that audience wasn’t asking.
The three lessons we’d pay for again
- Split every number by geography before reading it. Our blended CPA would have been a meaningless ~$11. The split, $1.84 ES, ∞ EN, is where all the information lived.
- The hook is market-specific; the design travels, the message doesn’t. Same image, different first line, opposite outcomes. For less price-sensitive markets the hooks worth testing sell time and replacement (“your agency charges $5,000 for this”), not discounts.
- Optimize past the registration. Cheap signups from a market that can’t pay your price look like winning and aren’t. The metric that should steer the budget is cost per activated user at minimum, per paying user ideally, which small budgets can’t measure quickly. That’s an argument for smaller price points, regional pricing, or patience; it’s not an argument for more spend on the same funnel.
What we changed
Both campaigns are paused as of early September while we act on lesson three. The next test targets English-speaking markets with non-price hooks and lower-CPM geographies (UK, Canada, Australia before the US), and the LATAM question becomes a pricing question rather than an ads question. We’ll publish the follow-up numbers when they exist.
If you’re weighing paid against organic at a small budget: the ~$100 experiment bought us more decision-grade information than most dashboards show in a quarter. And if the product you’re advertising is a site whose SEO is quietly broken, fix that first, it’s the free channel. Gantra audits any site at no cost, fixes written, which is exactly where our own funnel sends the clicks.
Frequently asked questions
How much does a SaaS signup cost on Facebook Ads?
In our campaigns: $1.84 per registration in Spanish-speaking LATAM, and effectively infinite in English-speaking markets, the same creatives produced zero registrations there. Published benchmarks vary wildly because they average across markets; the split by geography is the number that matters.
Do cheap Facebook signups convert to paying customers?
Not automatically, and this is the trap in optimizing for cheap registrations. Our $1.84 LATAM signups engaged with the product but did not convert at a $129/month price point: the price-market fit problem sat downstream of the ad. Cheap traffic that cannot pay is not cheap.
Should a SaaS run price-based hooks in ads?
Only where price is the buyer's active pain. Our price-led hook performed in LATAM and completely failed in English-speaking markets with the identical creative. For less price-sensitive markets, hooks about time saved, replacement of an expensive alternative, or a concrete product demo outperform discounts.
What budget is enough to test Facebook Ads for a SaaS?
We got statistically useful signal, 9 registrations, a clear geo split, and a creative winner, on roughly $100 of total spend across two weeks. Enough to learn is much cheaper than enough to scale: run small, read the split, then decide.