How to Analyse Push Notification Performance in Braze

A smartphone with notification cards beside a performance review checklist.

To analyse Braze push notification performance, read the report in the order a push travels. Start with sends, then deliveries and bounces, then direct and influenced opens, then conversions against a control. Judge creative on direct opens, check bounces and errors before blaming the copy, and use the Push Performance dashboards for the channel-wide view.

Key Takeaways

  • Read a push report in the order a push travels: sends, deliveries and bounces, opens, then conversions.
  • Judge creative on direct opens, which measure response to the notification itself.
  • On iOS and Android, total opens add influenced opens; on web push they count direct opens only.
  • Sends for a scheduled campaign include messages still waiting because of rate limiting.
  • Braze automatically unsubscribes bounced push users from future push notifications.
  • Use the Push Performance dashboards for the channel-wide view of performance, insights and deliverability.

Which metrics show Braze push notification performance?

Braze reports push results on each campaign's analytics page, and on Canvas analytics for messages sent in a Canvas. The metrics look familiar, but several are calculated in ways that change how you should read them.

MetricWhat Braze countsHow to read it
SendsAll messages for a launched scheduled campaign, including those still waiting because of rate limitingVolume intended, not volume delivered
BouncesPushes that failed, for example because there was no valid tokenA reach problem, and bounced users are unsubscribed from future push
Direct opensUsers who tapped the notification, as a share of deliveriesThe cleanest signal of whether the message itself worked
Influenced opensApp opens soon after the push without a tap, as a share of deliveriesWider effect on mobile, credited by each user's habits
Total opensDirect plus influenced on iOS and Android, direct only on web pushNot comparable between web and mobile
ConversionsThe events you defined, tracked after the initial sendBusiness effect, best read against a control

The row that catches people out is total opens. On iOS and Android it adds influenced opens to direct opens, while on web push it counts direct opens only.

Why should you read a push report in order?

Each stage of a push depends on the one before it. A weak open rate can be a copy problem, but it can also be a delivery problem that makes the copy look bad.

Read from the top down for that reason. If bounces rose or deliveries fell, fix reach first, because no change of wording will help users who never received the message.

The order also settles arguments quickly. A debate about wording ends once the earlier stages show a drop in reachable devices.

What is the difference between direct and influenced opens?

Braze logs a direct open when a user taps the notification and lands in your app. It logs an influenced open when the user opens the app without tapping, either less than thirty minutes after receiving the push or under half the average time since that user's last session.

That second rule matters. A heavy user who opens the app constantly gives the push little credit, while an occasional user who returns soon after the push is far more likely to count.

Influenced opens also differ from a conversion event set to app opens. Braze attributes every open within the conversion window to the campaign, whereas influenced opens use a window based on each user's own behaviour.

Use direct opens to compare creative, because they measure response to the notification itself. Use influenced and total opens to understand the push's wider effect on mobile.

Why do sends and recipients sometimes look wrong?

Some numbers are correct but counted differently from what you'd expect. For a launched scheduled campaign, sends include messages still waiting because of rate limiting, so early sends can run ahead of what has actually gone out.

Unique recipients can also include users who bounced, because Braze counts users it targeted for that send day, not only successful deliveries.

Two dashboard defaults cause confusion too. Campaign analytics show the last 90 days by default, so an older campaign displays zeros until you widen the range. Braze also notes that dashboard figures and Snowflake can differ slightly, and recommends Snowflake as the more precise source.

What do push bounces and errors tell you?

Treat bounces as an audience health signal. Braze automatically unsubscribes bounced push users from future push notifications, so a run of bounces shrinks the base you can reach.

The error messages point to the cause. Braze describes a NotRegistered error as usually meaning the app was deleted from the device, and a MismatchSenderId error as an authentication failure. A token removed from a profile doesn't by itself mean the user has push disabled, because they may still have other valid tokens.

Some errors start in your own content. Braze lists oversized payloads among invalid payload errors, including unexpectedly large values rendered by Liquid. If personalised pushes bounce more than plain ones, check what the Liquid returns.

Messaging Observability helps with the earlier stages. It reports how Braze processed a message, not its final delivery status. For push, Braze hands the message to the platform's push service, which then delivers it to the device. Access needs the View Dashboard Reports permission, and the release note advises contacting your customer success manager.

How should you use the Push Performance dashboards?

Campaign reports tell you how one message did. The Push Performance dashboards, generally available from the September 17, 2026 release, give a channel-wide view of performance, insights and deliverability across campaigns and Canvases.

Three views are worth a regular slot. The performance dashboard shows engagement across all campaigns and Canvases without rolling up individual messages. The frequency report shows how open rate changes as users receive more pushes, so you can find where extra sends stop earning engagement.

The deliverability dashboard tracks the health of your push audience over time, showing how your messaging affects your reachable base. Note that the frequency and cadence reports use a minimum three-month analysis window, so they won't reflect last week's change on their own.

In what order should you review a push campaign?

Work through these checks in order, and stop to fix the first stage that looks wrong.

  1. Confirm the reporting time range covers the whole campaign.
  2. Compare sends with deliveries, allowing for messages still queued by rate limiting.
  3. Check bounces and the push changelog errors, and fix reach before anything else.
  4. Compare direct open rates between variants or against earlier sends on the same platform.
  5. Read influenced and total opens for the wider effect on mobile, keeping web push separate.
  6. Check conversions against a control group to judge business impact.
  7. Review the channel dashboards monthly for frequency and audience health.

Step four is the creative test. Steps one to three decide whether that test is valid at all.

Which reading mistakes should you avoid?

Name the mistakes so a reviewer can flag them quickly. A platform mix-up compares total open rates across web and mobile, where the metric means different things.

A queued-send illusion reads early sends as delivered volume while rate limiting is still releasing them. A 90-day zero mistakes the default time range for a failed campaign.

A copy blame decides the wording failed when bounces or uninstalls were the real change. Check reach before creative every time.

Where does this connect to the rest of the programme?

Reading email results follows a similar discipline, covered in email campaign reporting. Audience choices that shape push results are covered in segmenting push notifications, and taking users to the right screen is in using deep links for mobile engagement.

For conversions to mean anything, the events behind them must be tracked well, as set out in tracking user events in Braze. Push rarely works alone, and multichannel engagement campaigns covers how it fits with other channels.

As a Braze implementation partner, CustomerIK recommends reading reach before response and response before revenue. Whenever a push number looks surprising, check the earlier stages first.

Frequently Asked Questions

1. What is an influenced open in Braze?

An app open without a tap that happens soon after a push, either within thirty minutes or under half the user's average time since their last session.

2. Why is my web push total open rate lower than mobile?

On web push, total opens count direct opens only. On iOS and Android they add influenced opens, so the two aren't comparable.

3. Why does an old push campaign show zero results?

Campaign analytics default to the last 90 days. Widen the reporting time range to see an older campaign's results.

4. What happens to users whose push bounces?

Braze automatically unsubscribes bounced push users from future push notifications.

5. Where can I see push performance across all campaigns?

The Push Performance dashboards show channel-wide performance, insights and deliverability across campaigns and Canvases.

Sources

CustomerIK is a Braze implementation partner working across onboarding, technical integration, marketing operations and customer data management.

If your push numbers don't add up and you want a second pair of eyes on the report, let's talk.

Preferences

Privacy is important to us, so you have the option of disabling certain types of storage that may not be necessary for the basic functioning of the website. Blocking categories may impact your experience on the website.

Accept all cookies
Accept all cookies

These items are required to enable basic website functionality.

Always active

These items are used to deliver advertising that is more relevant to you and your interests.

These items allow the website to remember choices you make (such as your user name, language, or the region you are in) and provide enhanced, more personal features.

These items help the website operator understand how its website performs, how visitors interact with the site, and whether there may be technical issues.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.