The right push notification service depends on who sends the messages.
If your developers send pushes from your own backend, Firebase Cloud Messaging is free and reliable.
If marketers need a dashboard, segments and A/B tests without engineering help, OneSignal is the cheapest place to start, and Braze or Airship fit larger teams running full lifecycle programs on sales-quoted contracts.
This roundup compares the four on delivery, pricing and testing, using each vendor's public pricing pages and docs.
We haven't used these tools ourselves, so nothing here comes from hands-on testing.
Where we can, we tie features back to real app tests from our database, because a push tool is only worth what you test with it.
| Service | Best for | Starting price | Push A/B testing | Not for |
|---|---|---|---|---|
| OneSignal | Small and mid-size apps that want push, in-app and email in one dashboard | Free up to 1,000 mobile push MAU; Growth from $19/month | 2 variants on Free and Growth; up to 10 on Professional and Enterprise | Teams that need an auto-sent winner or deep cross-channel orchestration |
| Firebase Cloud Messaging | Developer-led teams that want a free delivery layer | No cost | Notifications composer experiments through Firebase A/B Testing | Marketers who want journeys and segments without code |
| Braze | Larger teams running cross-channel lifecycle programs | Custom quote | Up to 8 variants, plus control group | Indie and early-stage apps that need predictable costs |
| Airship | Mobile-first brands with complex app messaging | Custom quote (self-service Essentials option) | Up to 26 variants, plus control group | Small teams that want public, self-serve pricing |
For the strategy side (when to ask for permission, what to send, how often), see our push notification guide. This page is about picking the tool.
OneSignal
OneSignal is the most common starting point for apps that want a marketer-friendly push dashboard without a sales call.
It bundles mobile push, web push, in-app messages and email, with segments and automated Journeys on top.
The free plan is generous for small apps.
As of November 2026, OneSignal's pricing page lists unlimited mobile push for up to 1,000 monthly active users, unlimited in-app messages, 10,000 free emails and 3 active Journeys.
Strengths:
- A free tier that covers real use, not just a trial
- Push, in-app messages and email in one place, so a trial-ending reminder and an in-app offer can live in the same flow
- Self-serve signup and published per-MAU pricing on the Growth plan
Downsides:
- Costs scale with mobile push MAU once you pass 1,000, which adds up for apps with large free audiences
- Free and Growth plans cap Journeys (3 and 5 active) and A/B tests at 2 variants
- Its docs say you must pick the winning variant manually; there's no automatic winner send yet
Who it's not for: large apps with a data team that wants a full customer data platform, or teams that need dozens of concurrent automated journeys on a low budget.
Firebase Cloud Messaging
Firebase Cloud Messaging (FCM) is Google's delivery service for Android, iOS and web.
It's a pipe, not a marketing suite: your app registers a device token, and your server (or the Firebase console) sends messages to it.
Many other push tools use FCM under the hood for Android delivery.
The price is hard to beat. Firebase's pricing page lists Cloud Messaging, In-App Messaging and A/B Testing as no-cost products on both the Spark and Blaze plans.
Strengths:
- Free at any scale, with a default quota of 600,000 messages per minute, according to FCM's quota docs
- Works with Remote Config and Google Analytics, so push experiments can be scored on retention or purchase revenue
- No vendor lock-in on pricing; you own the sending logic
Downsides:
- Segmentation, scheduling logic, frequency caps and lifecycle flows mostly have to be built by your team
- The console composer is basic compared with a dedicated engagement platform
- Delivery limits apply per device: on Android, up to 240 messages per minute and 5,000 per hour to one device, and collapsible messages are throttled to protect battery
Who it's not for: marketing or growth teams without developer time.
If nobody can write the trigger logic, the free price won't help.
Braze
Braze is a customer engagement platform built for large consumer brands.
Push is one channel among email, SMS, in-app messages, content cards and more, all driven by a shared user profile and a visual journey builder (Canvas).
It's strong where lifecycle programs get complex: many channels, many teams, real-time triggers and personalization.
The trade-off is cost and setup. Braze offers four editions (Go, Select, Pro and Enterprise), and its pricing page doesn't list prices.
Cost is based on monthly active users and a pool of "action credits" spent on messages and AI features, and contracts include services.
Strengths:
- Deep cross-channel orchestration, so a push, email and in-app message can work as one sequence
- Solid testing: up to eight variants plus a control group, and optional AI-driven variant optimization
- Built for scale and for teams with many people sending
Downsides:
- No public prices; you'll need a sales process and an annual-style contract
- Credit-based pricing makes it harder to predict cost as channels and volume grow
- More platform than a small app needs, with a real implementation effort
Who it's not for: indie developers, early-stage apps and anyone who only needs push.
Airship
Airship has a long history in mobile push and positions itself as an app-first engagement platform.
It covers push, in-app experiences, web notifications, email, SMS and mobile wallet, with journey orchestration and an experimentation suite that includes feature flags.
Airship also publishes one of the most useful benchmark reports in the category.
Its 2026 report puts the median push opt-in rate at 53.3% on Android and 48.85% on iOS.
We collect more figures on our push opt-in rate benchmarks page.
Strengths:
- Mobile depth: push is the core product, not an add-on to email
- The largest variant limit of the four for message tests (26 variants), with a control group option
- Feature flags and journeys in the same platform as messaging
Downsides:
- The main platform is quote-based, so budgets are hard to plan before a sales call
- Its feature set overlaps with Braze, and both cost more than small apps need
- Third-party pricing guides describe Airship as custom-contract only, while Airship's own pricing page shows a self-service Essentials option with a "Start free" signup; check the current terms before you commit
Who it's not for: small teams that want a published price list, or developer-led apps that only need delivery.
Pricing compared
All prices below are as of November 2026, from each vendor's public pricing page. Only two of the four publish numbers.
| Service | Free option | Paid entry point | What drives cost |
|---|---|---|---|
| OneSignal | Unlimited mobile push up to 1,000 MAU | Growth from $19/month plus $0.012 per mobile push MAU | Mobile push MAU, email sends, web push subscribers |
| Firebase Cloud Messaging | No cost on Spark and Blaze | None needed for FCM itself | Your own servers and engineering time |
| Braze | None listed | Custom quote | MAUs and action credits |
| Airship | Self-service Essentials signup | Custom quote for the full platform | Scale and channels used |
A quick worked example for OneSignal: an app with 20,000 mobile push MAU on Growth would pay the $19 base plus about 19,000 × $0.012, or roughly $247 a month, assuming the first 1,000 MAU are covered by the platform fee as the pricing page describes.
Run your own MAU count through it before you choose.
Don't forget the hidden cost on the free option.
FCM costs nothing to send, but someone has to build segments, schedules, frequency caps and reporting.
For a team with no spare developer time, a paid dashboard can be cheaper overall.
A/B testing support
Every tool here can split-test a push. The differences are in how many variants you get, whether there's a no-message control group, and what metric decides the winner.
- OneSignal: 2 variants on Free and Growth, up to 10 on Professional and Enterprise. By default 25% of the audience gets the test, and you pick the winner by hand. Journey split branches handle tests of wait times and paths.
- Firebase: Notifications composer experiments start with one baseline and one variant, and you can add more. You can leave the baseline blank to send no message, which gives you a true holdout. Goals include retention and purchase revenue, but Google Analytics must be enabled.
- Braze: up to eight variants with an optional control group, plus AI-driven optimization that sends the winner to a holdout after the test.
- Airship: up to 26 notification variants, with a control group you can switch on and custom audience splits.
The control group matters more than the variant count.
Open rate tells you which copy got tapped; a no-message holdout tells you whether the push changed retention or revenue at all.
Braze's own docs warn that a control group can't be judged on opens or clicks, so pick a downstream goal.
What should you test? Our database has few clean push-copy tests, but our app tests on timing and offers point to what tends to move revenue:
- Timing over volume. Wizz showed fewer offers at better moments and saw +81% on the primary offer flow and LTV up about 20%, as reported on RevenueCat's blog. That was a paywall test, but the lesson carries to push scheduling.
- The channel, not just the copy. Lose It! moved offers into in-app messages and made about 10x more money, according to RevenueCat's write-up. Test push against in-app delivery for the same offer.
- Win-back offers. Unscripted sent lapsed users a personal reactivation trial offer and saw over +240% trial starts (vendor-reported in a Superwall case study). A push is a natural way to deliver that offer.
- Trial-end reminders. Blinkist's paywall told users when the trial would end and saw a significant rise in trial opt-in and 4% fewer trial cancellations (credited to Marcus Burke). Stardust's trial timeline lifted conversion 13.6%, with trial-to-paid unchanged (vendor-reported by Superwall). If your paywall promises a reminder, your push tool has to send it on time.
- Offers during the trial. Vahe Baghdasaryan's "end your trial now" discount tests averaged +15–17% ARPU. Push is one way to surface that offer mid-trial.
Before you call a winner, check your sample.
Push tests often reach thousands of users but measure rare events like purchases, so use our sample size calculator first.
Browse more results in our retention tests, win-back tests and push notification element page.
A simple way to choose: start with FCM if developers own messaging, OneSignal if marketers do and budget is tight, and talk to Braze or Airship once you run push, email and in-app as one program across several teams.
If your offers end on a paywall, pair the push tool with one from our paywall tools roundup, and see app A/B testing tools for testing beyond messages.