Tara Event Analytics Report (25 Sept – 15 Oct 2026)

[TenzinMonlam]

Post-event report on the Dzongsar 21-Day Tara Puja & Teachings in the WeBuddhist app.

Sources: PostHog, Microsoft Clarity, Airbridge, Firebase.

1. summary

During 21 days, 3484 people used the app, 3467 of them (99.5%) for the first time, and 1974 of them (57%) opened the Tara event in the app.

What we counted	Number	In plain words
People who used the app	3,484	Almost everyone (99.5%) was using the app for the first time.
People who opened the Tara event	1,974	That is 57 out of every 100 app users. They opened it 27,099 times between them.
People who joined the live stream	1,528	They joined 27,136 times — about 18 times each over the three weeks.
People marked as attending	1,503	About half of those who gave a status joined online, half were there in person.
Hours of live stream watched	1,320 hours	That is 55 days of continuous viewing.
People using the app on a typical day	984	The busiest day was 10 October with 1,261 people.

• The three things that did not go well are on the app side:
  • many people struggled to log in
  • Android phones lost more people on the way into the live stream than iPhones did
  • Lot of taps on screen did nothing (people tapped the “Live” badge and the event title expecting them to open the stream).

Crash numbers also look alarming at first glance but mostly come from a reporting mistake.

2. Scope and data sources

In the app, the Tara event is the group event “Dzongsar 21-Day Tara Puja & Teachings” in the Dzongsar Khyentse Chökyi Lodrö group, a hybrid event with a daily live stream and in-person attendance. Its title is shown in English, Chinese (宗薩甚深精華度母十萬薈供) and Tibetan (རྫོང་སར་ཟབ་ཏིག་སྒྲོལ་མ་འབུམ་ཚོགས།) depending on the user’s UI language.

Platforms: iOS, iPadOS, Android, plus a handful of macOS sessions.

Tool	Used for
PostHog	Users, installs, Tara funnel, retention, sessions, screens, language, push opens, deep links, rage clicks
Microsoft Clarity	Sessions, dead and rage clicks, country/OS split, session recordings
Airbridge	Installs by channel/campaign, event-link attribution
Firebase	Installs/uninstalls, crash-free rate, push delivery and open rates

3. Acquisition

• 3205 devices installed the app during the event (PostHog) and 2059 of those came in the first seven days.
• The first day alone brought 443 installs. After the first week the pace settled at 30–75 a day, with one jump to 142 on 10 October when the Zangchot Monlam recitation event started in the same app.
• Firebase also saw 427 Android phones remove the app during the three weeks.

Acquisition metric	25 Sept – 15 Oct	Source
New users	2933 (Android 1597 · iOS 1336)	Firebase
Installs (unique devices)	3205 (Android 1803 · iOS 1368 · iPadOS 86 · macOS 13)	PostHog
Uninstalls (Android only)	427	Firebase
Active users	3464 (Android 1895 · iOS 1557 · web 12)	Firebase
Onboarding completed	2084 of 2119 started (98%)	PostHog
Chose UI language at onboarding	en 1363 · bo 474 · zh 345 · hi 32 · ne 14 · mn 3	PostHog
Deep-link opens (all)	1727 opens by 550 people	PostHog
Tara event link	147 opens by 72 people	PostHog
Airbridge-sourced links (home)	76 opens by 60 people	PostHog

Source: PostHog · people who used the app, by country, 25 Sep – 15 Oct 2026

• How people found the app: We cannot say much here. Airbridge, the tool that records which link or campaign an install came from, shows that 98 out of 100 installs had no source recorded, because only one tracked link existed (the in-app “App_OnBoard” link: 385 taps, 57 installs). No adverts were run, so there is no cost per install to report. Inside the app, only 72 people (147 opens) arrived at the Tara event through a shared link.

4. Engagement

The Tara event was the app’s main activity: 57% of all users opened it, 44% entered a live session, and 43% had attendance recorded. Average daily active users were 984 (PostHog), with a peak of 1,261 on 10 October.

Engagement metric	25 Sept – 15 Oct	Source
Average daily active users	984	PostHog
App opens	110,254 by 3195 people	PostHog
Opened the Tara event	1974 people, 27,099 opens	PostHog
Entered the live session	1528 people, 27136 entries (17.8 each)	PostHog
Watched a live session to its end	1080 people, 8906 sessions	PostHog
Attendance recorded	1503 people (1305 set a status: 659 online, 646 in person)	PostHog
Live watch time	1319.7 hours	PostHog
Opened the replay	19 people, 44 opens	PostHog
Shared the event	93 shares by 59 people	PostHog
Followed the Dzongsar group	43 (26 unfollowed)	PostHog
Opened the group chat / sent a message	206 / 18 people	PostHog

Firebase measures how much people used the app, not just how many used it:

Firebase engagement metric	25 Sept – 15 Oct
Time in app per active user	2 h 04 min
Engagement rate	83.8%

Source: PostHog · people using the app per day and people opening the Tara event per day, 25 Sep – 16 Oct 2026

PostHog · funnel insight, unordered, 21-day window, 25 Sep – 15 Oct 2026

Source: PostHog · hours of live stream watched per day, 25 Sep – 15 Oct 2026

• How much people watched. 1320 hours of live stream were watched in total. The busiest day was 28 September (147 hours by 258 people). Most visits to the stream were short check-ins of about a minute; a smaller group watched for a long time. People joined mostly at 8 in the morning (5942 joins at that hour), with smaller peaks at 2 pm and 4 pm. Only 19 people watched the replay afterwards.

• Languages. When the event was opened, 66 out of 100 people saw it in English, 17 in Tibetan and 17 in Chinese. At sign-up, 61% chose English, 21% Tibetan, 15% Chinese, and a few chose Hindi or Nepali.

• Other things people used. Beyond the event, 1948 people used the reader, 1636 looked at practice plans, 1178 looked at series (1522 joined one), 709 used the mala counter, 389 opened recitation collections and 303 used the timer. The most-opened screens were the practice plan page, the group event page, the reader and the home screen.

• Notifications. 1700 of the 2551 people agreed to receive notifications, 984 denied, and 175 of those appear in both counts because they refused at first and allowed later, or the other way round.
  Firebase: 116,575 notifications delivered and 4724 opened.

  • Most opened notifications (PostHog, which counts 10,428 opens):
    • group chat 4074 opens
    • verse of the day 733
    • Tara event reminders 107 opens by 95 people
    • other event and series notifications under 100 each

5. Retention

Of the 692 people who joined the app in the first few days, 65 out of 100 used it again the following week, 48 the week after, and 38 three weeks later. Looking day by day at the 437 people who joined on the very first day: half came back the next day, a third were still using the app a week later, and a quarter two weeks later.

Source: PostHog · retention insight

Among the 2071 people who joined in the first week, we compared those who opened the Tara event (1397) with those who never did (674). Tara participants used the app on 9 days out of 21 on average; the others on 4 days. Nine in ten Tara participants came back at least once, and nearly two thirds were still using the app in the final week, compared with one third of the others.

6. Stability and issues

1. Logging in was hard. 2825 people started to log in during the event. 1305 of them (46 out of 100) failed at least once before getting in and 2244 got in eventually. 641 people chose to continue as a guest instead.
2. Android phones lost more people on the way to the live stream. On iPhones, 85 out of 100 people who opened the event joined the stream and 58 watched a session to its end. On Android it was 69 and 50.
3. Many taps did nothing. Clarity records every tap in a visit and notes when a tap has no effect. Over the event there were 14,623 such taps in 15,732 visits, so almost one per visit. Most of them land on three things:

• The “Live” badge on the event page (over 7,000 taps across about 480 visits, in English and Chinese).
• The dark area around a pop-up sheet: people tap outside the sheet to close it, and it does not close (about 6,000 taps).
• Titles that look like buttons, such as the event name and “The Praise to the Twenty-One Taras”, and the “Copy” button, which failed repeatedly for a few people.
  
  
  
  
  
  
  Clarity · daily rage clicks, dead clicks and sessions, 25 Sep – 14 Oct 2026 (UTC days)

Blue bar(rage click) on 11 October is not a general problem. It came from two Android visits, one in Malaysia and one in India, where a screen stopped responding: in one, the user spent ten minutes tapping the calendar (over 1,800 taps); in the other, the timers screen took over 1,500 taps in seven minutes. Both look like the screen froze.

4. Crashes: Only 58 out of 100 Android users had a crash-free experience over the month. But 5,100 of the 6,600 recorded “crashes” are the same thing: a picture (a plan thumbnail stored on our server) failing to download, which the app wrongly reports as a crash.

For iPhones we cannot say anything yet, because 1,371 crash reports are waiting on files the developers need to upload before Firebase can read them. The “99.9% crash-free” figure shown for iPhone is therefore not real.

7. Data gaps and discrepancies

• Two tools, two answers. You will see 3,484 users in one place and 3,464 in another, or 1,065 people on 15 October in one tool and 659 in another. The tools count differently: Firebase only counts someone as “active” if they actually looked at the app for a moment, while PostHog counts any activity, including things that happen in the background. Neither is wrong; the report always says which tool a number comes from.
• How people found the app. Only one tracked link existed, so 98 out of 100 installs have no known source. We can see what people did once inside, not what brought them.
• Crashes. The Android crash number is inflated by picture downloads being reported as crashes, and the iPhone number is missing most of its reports. Neither can be used as a scorecard until these are fixed.
• Language. Firebase thinks only 1 in 100 users is Tibetan-speaking because it reads the phone’s language setting, and most Tibetan readers run their phones in English. The app’s own sign-up choice (21% Tibetan) is the right figure.
• Screens. Clarity and Firebase were not told which screen a person was on, so we cannot show tap heatmaps per screen. Screen names do appear inside individual recordings, which is how the frozen-screen sessions were found.
• Sharing. The share buttons inside the app are mostly not tracked, so the 93 event shares are a minimum, not a total.