Test whether your app makes money with $300–500 of ads
$300–500 of ads can show whether strangers will pay for your app, if you spend it in three or four small rounds. At 2026 average prices, $400 buys roughly 200 to 670 clicks on Facebook and Instagram. That's enough to compare ads and see where people give up, but it may bring only a handful of purchases.
Use the early rounds to fix the path from ad to payment, and the last round to test the price.
What $400 buys
These are WordStream and LocaliQ's 2026 averages for mostly small-business ads, not apps, so your costs will differ. A lead is someone who leaves their contact details.
| Where (2026 averages) | Average cost | What $400 buys |
|---|---|---|
| Facebook and Instagram, ads for visits | $0.60 per click (1.93% of viewers click) | About 670 clicks |
| Facebook and Instagram, ads for leads | $1.80 per click, $27.39 per lead | About 220 clicks or 15 leads |
| Google Search | $5.42 per click, $66.69 per lead | About 74 clicks or 6 leads |
How many of those clicks will pay
FunnelFox's 2026 report (FunnelFox sells software for this) found that 3% of visitors buy in established web sign-up flows. At that rate, 670 clicks would bring about 20 purchases. A new, rough app at 1% would bring about 7.
Or count from sign-ups: at about $5 per sign-up with 4% of sign-ups paying, $400 brings roughly 3 paying customers.
Get your app ready before you buy ads
Before you spend, check you have:
- One ad idea and one first screen that repeats the ad's promise in the same words.
- A few quick questions, or one try of the main feature, such as a first result.
- A real price and a working checkout (Stripe or Paddle, switched to live), not a waitlist.
- A way to count each step: arrived, continued, finished, saw the price, started paying, paid.
- The whole path tested from inside Instagram and Facebook, where Google sign-in is blocked and Stripe's Link and PayPal buttons don't work.
Run it in rounds
- Round 1 (about $100): try three to five ad ideas, all sent to the same app. Keep the one with the most clicks and the cheapest taps on your first screen. Don't edit ads mid-round: Facebook says edits restart its learning.
- Round 2 (about $100): run the winning ad and fix the step where most people give up. Judge by the cost of each person who gets a first result.
- Round 3 (about $150): same ads, now focused on the price screen: price, plan options, trial or no trial. Judge by people who start paying and purchases.
- Round 4 (optional, about $100): rerun the best version and check that purchases hold up.
Decide your rules before you start
Bolt's idea-testing guide recommends writing down what success looks like, and when you'll stop, before you launch. The rules below are examples to adapt, not industry standards.
| What you see after round 3 | Example decision |
|---|---|
| Each purchase costs no more than a customer pays you in their first one to three months | Keep going and raise the budget slowly |
| People buy, but each purchase costs several times the price | Fix the weakest step or the price, then run another round |
| People see the price but nobody starts paying | The offer or price is wrong. Change one of them before spending more |
| Clicks, but almost nobody gets past the first screen | The ad and the first screen don't match. Rewrite the first screen |
Facebook or Google?
Facebook and Instagram ads reach people who weren't looking for you, which suits new kinds of apps. Google Search reaches people already typing the problem, like 'invoice generator for freelancers'. That costs more: in WordStream's 2026 data, an average Google Search click cost about nine times an average Facebook click for visits.
Either way, a test this small rarely gets Facebook the 50 results a week per ad set it needs to learn from purchases. Aim it at an earlier step that happens more often, and count purchases yourself.
How small tests go wrong
- Changing ads or budgets halfway through a round.
- Counting bots as visitors. Bolt's hosting stats, for example, include them.
- Testing on free hosting that goes offline when it hits its traffic limit.
Your next step
Make sure a stranger can get from an ad to a finished payment on a phone, write down your stop rule, and start round 1 with about $100. Foundation finds your first paying customers: it makes the ads and the page people land on, and shows which ads brought buyers. It's waitlist-only right now.
Questions
Is $300 enough to test an app idea with ads?
It's enough to compare ads and find where people give up, yes. It's often not enough to judge your price, because at a 3% purchase rate a few hundred clicks bring only a handful of buyers.
Can I judge my app after $50 of ads?
No. At about $5 per sign-up with 4% of sign-ups paying, $50 brings about 0.4 paying customers, less than one buyer. Spend in rounds of about $100.
Should I test with a waitlist or with real payments?
A waitlist or a fake price button (Bolt's help center describes one) is a cheap first signal, but only a real payment proves people will pay.
When should I stop a test?
When it hits the stop rule you wrote before launch. For example: after two first-screen rewrites and two price changes, still nobody starts paying.
Sources
- WordStream: Facebook Ads Benchmarks 2026
- WordStream: Google Ads Benchmarks 2026
- FunnelFox: State of Web2App 2026
- Meta Business Help Center: About the learning phase
- Bolt help center: Test your business idea before you build it
- Bolt help center: Hosting analytics
- Bolt help center: Bolt Cloud hosting plans
- Stripe docs: Express Checkout Element (in-app webview support)
- Google for Developers: OAuth changes for embedded webviews (2021)