# Launch and measure

Test every journey on your own phone before real readers get its messages, then check whether it works.

## Test each journey before you launch it

A journey has no test mode, so once it is running, real readers who match its entry get its messages. Testing on your own phone first catches a missing event, an empty segment, or a wrong message before readers see it.

Run through this list for each journey:

1. Install the version of your app that includes the Pushwoosh SDK on your phone. If you do not have one, ask your developers for a test version.

   The phone must register in the application where the journey runs. To send test messages from presets, also register it as a [test device](/product/first-steps/start-with-your-project/test-your-integration/test-devices/).
2. Log in to the app with a test account.
3. Use the app and do what the journey needs, for example start an episode. Then open **Audience** → **User explorer**, find the test device, and check **Events history**.

   The events must be there with the expected attributes, or the segment and the journey do not see them.
4. If the journey starts from a segment, open it in **Audience** → **Segments** and refresh its size to see whether it counts any readers.
5. Launch the journey for your test readers only. Build a test segment from your test User IDs with the **User ID** filter. If the journey starts from a segment, use the test segment as its entry. See [How to test a journey with a test segment](/product/first-steps/start-with-your-project/test-your-integration/#how-to-test-a-journey-with-a-test-segment).

   <Aside type="caution" title="Journeys that start on an event">
   If the journey starts on an event (a [Trigger-based entry](/product/customer-journey/journey-elements/entry-elements/trigger-based-entry/)), every reader who triggers the event enters it, including real readers, and the entry has no segment filter. To keep real readers out of the messages, use one of these:

   * Right after the entry, add a [Condition split](/product/customer-journey/journey-elements/flow-controls/segment-split/) on your test segment and leave its **All others** branch empty. Real readers enter but leave the journey at once without messages, and they count in its entries. Remove the split before the real launch. If the journey has a [**Campaign Entry Limit**](/product/customer-journey/journey-settings/#campaign-entry-limit) such as **a month**, real readers can use up their entry during the test, so clone the journey instead.
   * [Clone the journey to a test application](/product/customer-journey/journey-settings/#clone) in the same account and run it there with a phone that is registered in that application. Events the test application lacks are created there, but check the segments and the content of the clone.
   </Aside>

6. On the test phone, do the action that the entry waits for.
7. Check that the message arrives, and follow the test reader with [Find user path](/product/statistics-and-analytics/journey-statistics/analyze-user-path/).
8. After a day, check how many readers entered the journey. See [Read the journey statistics](#read-the-journey-statistics).

To test a news alert, send it to the test segment first, with the same message type and settings as the real alert.

## Read the journey statistics

Journey statistics show whether readers enter a journey, get its messages, and reach its goal. They count goal conversions, not revenue. To compare revenue, use [Conversion events](/product/audience-data-and-segmentation/events/conversion-events/).

To see the journey statistics:

1. Open the journey on the canvas. To read the numbers for the whole journey, use the statistics panel on the left.
2. For numbers on each element, open Canvas settings (the gear icon in the bottom-right corner) and turn on **Show metrics**. See [Journey element statistics](/product/statistics-and-analytics/journey-statistics/journey-element-statistics/).

For what each metric means, see [Overall journey statistics](/product/statistics-and-analytics/journey-statistics/overall-journey-statistics/).

### Check the first-week signals

In the first week after launch, look for these signals:

| If you see | It usually means | What to do |
| --- | --- | --- |
| No entrances | The entry segment is empty or the entry event does not arrive | Repeat steps 3 and 4 of [Test each journey before you launch it](#test-each-journey-before-you-launch-it) |
| Far fewer messages sent than entrances | Many readers can't get push, or frequency capping and the silence period hold messages back | Open **Communications** on the statistics panel and check **Suppressions** in the **Frequency capping** card, add an in-app or email branch after a [Reachability check](/product/customer-journey/journey-elements/flow-controls/reachability-check/), and check your [delivery rules](/tutorials/industry-playbooks/media-and-streaming/prepare-for-communication/#set-delivery-rules-before-you-message-readers). News alerts also count toward frequency capping unless you send them as Transactional |
| Many drop-offs | Readers left the journey because of a problem, such as a device that was not found or an invalid push preset | Expand the **Drop-offs** block on the element to see the reasons |
| Messages are opened, but few readers reach the goal | The message works, but the content or the moment does not | Test another text or offer with an [A/B/n split](/product/customer-journey/journey-elements/flow-controls/abn-split/) |

<Aside type="tip" title="Journey does not behave as expected?">
If a journey does not behave as expected, follow one reader through it with [Find user path](/product/statistics-and-analytics/journey-statistics/analyze-user-path/), and look up why readers leave a journey or skip a message in [Drop-off and skip reasons](/product/troubleshooting/drop-offs-in-customer-journey/).
</Aside>

## Measure results

To measure whether a journey changes reader behavior, use the three tools below, from simplest to strongest.

* **Conversion goals:** every journey in this playbook has one. News alerts are one-time pushes and have none. The goal can be reached within the period you set, up to 30 days after the reader leaves the journey. See [Conversion goals](/product/customer-journey/journey-settings/#conversion-goals).
* **Holdout Split:** hold back 1% to 20% of the readers who reach a [Holdout Split](/product/customer-journey/journey-elements/flow-controls/holdout-split/) element and compare their goal rate with everyone else. The **Goal uplift vs holdout** report updates once a day.

  <Aside type="tip" title="Where to place a Holdout Split">
  * Put it right after the entry, before any message. Connect only the **In journey** branch and leave **Holdout** empty, so held-out readers leave the journey there and get nothing.
  * The journey needs a goal, because the report compares goal rates between the two groups.
  * To measure one part of a journey, put another Holdout Split in front of that part.
  </Aside>

* **Control group:** by default, holds out a share of the readers from all marketing messages in the application. Compare them with everyone else to measure the combined effect of your messaging. Set it up before you launch. See [Set up a control group before you launch](/tutorials/industry-playbooks/media-and-streaming/prepare-for-communication/#set-up-a-control-group-before-you-launch). To read the results, see [Analyze control group impact](/product/audience-data-and-segmentation/control-groups/analyze-control-group-impact/).

## Cases to check before launch

These cases decide whether your segments fill up and your journeys and news alerts reach readers. Check them before you launch.

### General cases

These cases are the same in every media app and in other industries.

| Case | What happens | What to do |
| --- | --- | --- |
| iOS push permission prompt | iOS shows it only once, so a reader who declines on the first launch can't be asked again | [Ask for push permission at the right moment](/tutorials/industry-playbooks/media-and-streaming/prepare-for-communication/#ask-for-push-permission-at-the-right-moment) |
| Repeating events such as `Episode started` | They fire many times a day, and each one after the previous run of the journey ended starts it again | Set a **Campaign Entry Limit**. See [Best practices for every journey](/tutorials/industry-playbooks/media-and-streaming/build-the-journeys/#best-practices-for-every-journey) |
| Two paths to revenue | If the app sends `PW_Conversion` and `Subscription started` also carries `__amount`, Pushwoosh records each subscription twice | Use one path per subscription. See [Track revenue from subscriptions](/tutorials/industry-playbooks/media-and-streaming/collect-reader-data/#track-revenue-from-subscriptions) |
| Readers without a push token | An Audience-based entry and segment filters with never take in only devices with a push token, so readers whose app has not asked for push permission yet are left out | Ask for permission early enough for your journeys. See [Build the segments](/tutorials/industry-playbooks/media-and-streaming/build-the-segments/) |

### Media cases

These cases are specific to media apps: news alerts, reader tracking, and subscriptions.

| Case | What happens | What to do |
| --- | --- | --- |
| News alerts of the Marketing type | They count toward frequency capping, wait for silent hours, and skip readers in the control group, so on a busy news day some readers don't get an alert or get it late | Decide the message type before the first alert. See [Choose how news alerts follow the delivery rules](/tutorials/industry-playbooks/media-and-streaming/prepare-for-communication/#choose-how-news-alerts-follow-the-delivery-rules) |
| A news alert to an offline phone | With the default **Time to Live** of 1 week, a phone that comes online later gets an alert about old news | Set **Time to Live** to a few hours for each alert. See [News alerts](/tutorials/industry-playbooks/media-and-streaming/build-the-journeys/reading-and-watching/#news-alerts) |
| A reader who has not given consent | The SDK sends nothing to Pushwoosh until the app turns communication on, so the reader is not in segments and gets no messages | Check that the app turns communication on after consent. See [Start the SDK after the reader gives consent](/tutorials/industry-playbooks/media-and-streaming/prepare-for-communication/#start-the-sdk-after-the-reader-gives-consent) |
| Store purchases without the reader's HWID | Pushwoosh receives the store notification but records no subscription event, so the Cancelled subscription journey never starts | Ask your developers to pass the HWID with each purchase. See [Track subscriptions from the stores](/tutorials/industry-playbooks/media-and-streaming/collect-reader-data/#track-subscriptions-from-the-stores) |
| Readers who don't log in | Without your own User ID, Pushwoosh counts each device as a separate reader, so what a reader does on one device does not count on another, and an episode started on a phone and finished on a tablet starts the Continue watching reminder | Set your account ID as the User ID for readers who log in. See [Identify readers by User ID](/tutorials/industry-playbooks/media-and-streaming/collect-reader-data/#identify-readers-by-user-id) |
| A subscription bought on your website | The app does not know about it, so the paywall offer can still reach a reader who just subscribed | Send `Subscription started` from your server with the reader's `userId`. See [Send the events from the app and your server](/tutorials/industry-playbooks/media-and-streaming/collect-reader-data/#send-the-events-from-the-app) |
| Inactivity segments right after launch | At-risk and Lapsed readers are empty until `Content read` has been recorded for 14 and 60 days | Use `PW_ApplicationOpen` in the same filters at first. See [Build the segments](/tutorials/industry-playbooks/media-and-streaming/build-the-segments/) |