Tealium AOM Takip Akisi

Versiyon: 4.0 | Guncelleme: 2026-06-11

---

1. Yonetici Ozeti

PostPaid ana sayfa acildiginda 3 paralel API cagrilir, tumu tamamlaninca (veya timeout olunca) AOM MVA call tetiklenir. 200K gercek kullanici verisiyla toplam sure cihaza gore 3.7s - 7.4s arasi degisir.

Ozet Sureler (200K Kullanici Verisi):

RAM Segmenti	Kullanici Payi	Dashboard → AOM Request	AOM Request → Response	Toplam Sure	Render Suresi
3-4GB RAM	30.5%	3.9s	3.5s	7.4s	1.8s
4-6GB RAM	26.3%	3.1s	3.0s	6.1s	0.9s
6-8GB RAM	20.6%	2.5s	2.4s	4.9s	0.8s
8-12GB+ RAM	3.0%	1.8s	1.9s	3.7s	0.4s

Kritik Bulgu: Kullanicilarin %56.8'i 3-6GB RAM segmentinde ve ortalama 6.8s yukleme yasiyorlar.

---

1a. Uygulama Kullanici Tabani - RAM Dagilimi

Global Installation Base: 11.06 Milyon Kullanici (1,379 farkli cihaz modeli)

RAM Grubu Bazli Dagilim

RAM Grubu	Cihaz Sayisi	Toplam Kullanici	Kullanici Payi	Ort. per Model
≤2GB	158	513,954	4.6%	3,252
2-3GB	194	1,634,190	14.8%	8,423
3-4GB	431	3,370,668	30.5%	7,820
4-5GB	7	25,573	0.2%	3,653
5-6GB	239	2,912,495	26.3%	12,186
6-8GB	288	2,278,809	20.6%	7,912
8-12GB	59	327,084	3.0%	5,543
>12GB	3	1,106	0.0%	368

Tespit Edilen Segments

• Dusuk RAM (3-4GB): 30.5% kullanici (3.37M) | 431 cihaz modeli | EN BUYUK SEGMENT
• Orta RAM (4-6GB): 26.3% kullanici (2.91M) | 246 cihaz modeli | IKINCI BUYUK SEGMENT
• Yuksek RAM (6-8GB): 20.6% kullanici (2.28M) | 288 cihaz modeli
• Premium RAM (8GB+): 3.0% kullanici (327K) | 62 cihaz modeli

Sonuc: Performance tuning 3-4GB ve 4-6GB RAM segmentlerine odaklanmali (%56.8 toplam pazar payi, 6.28M kullanici)

---

2. Architecture Overview - Component Diagram

+--------------------------------------------+
|            Yanimda App                      |
|                                            |
|  +--------------------------------------+  |
|  |  PostPaidHomeActivity (UI)           |  |
|  |  - LiveData observe                  |  |
|  |  - setTargetTimeoutTimer (2s guard)  |  |
|  |  - AOM_MVA confirmation              |  |
|  +------------------+-------------------+  |
|                     v                      |
|  +--------------------------------------+  |
|  |  PostPaidHomeViewModel               |  |
|  |  - isTargetRun (guard flag)          |  |
|  |  - isAllTargetDataEnabled()          |  |
|  |  - triggerTargetTimeout              |  |
|  +------------------+-------------------+  |
|                     v                      |
|  +--------------------------------------+  |
|  |  HomePageResponseState (State)       |  |
|  |  - postValue() [thread-safe bridge]  |  |
|  +------------------+-------------------+  |
|                     v                      |
|  +--------------------------------------+  |
|  |  TealiumHelper (SDK Wrapper)         |  |
|  |  - trackView() - HTTP POST           |  |
|  |  - homeRemoteCommand callback        |  |
|  +------------------+-------------------+  |
|                     v                      |
|  +--------------------------------------+  |
|  |  Tealium SDK                         |  |
|  +------------------+-------------------+  |
+---------------------+----------------------+
                      v
           +---------------------+
           |  Tag Management      |
           |  Server              |
           +---------------------+

---

3. Business Sequence Diagram

---

4. Business Flow Chart

---

5. Yukleme Fazlari

Faz 1: Dashboard Init

• PostPaidHomeActivity.startAomMva() cagirilir
• setTargetTimeoutTimer() baslatilir (2s guard)
• observeOfferModel() baslat

Faz 2: Paralel API Cagrilari

ViewModel init'te 3 API ayni anda tetiklenir:

API	Gorev
getAppIcons	Icon change response
getMainPageOffer	MCCM main page offer
setTobiWelcomeMessage	Tobi welcome message

Her API response geldiginde ViewModel'e postValue yapilir.

Faz 3: Karar Noktasi (Race Condition)

Iki yoldan biri ilk olan kazanir:

Yol A - Normal: Tum API'ler tamamlanir (isAllTargetDataEnabled = true)

• observeOfferModel observer tetiklenir
• isTargetRun = true (timer artik skip eder)
• trackAomCall(getTealiumData()) cagirilir

Yol B - Timeout: 2s dolar, API'ler hala gelmemis

• setTargetTimeoutTimer icindeki delay(millis) biter
• isTargetRun = true (observer artik skip eder)
• isAtLeastOneTargetDataEnabled() kontrol edilir
• Varsa partial data ile, yoksa null ile trackAomCall

Faz 4: AOM Response

• TealiumHelper.trackView("aom_mva") gonderilir
• Server remote command response doner
• PostPaidJourneyMetricsTracker.markAomResponse()

---

6. API Envanteri

#	API	Tip	Tetikleyici	Timeout Guard
1	getAppIcons	REST	ViewModel init	2s timer
2	getMainPageOffer	REST	ViewModel init	2s timer
3	setTobiWelcomeMessage	REST	ViewModel init	2s timer
4	fetch_aom (AOM MVA CALL)	Tealium trackView	Tum API complete VEYA timeout	-
5	homeRemoteCommand	Tealium Remote Cmd	Server response	-
6	aom_mva (AOM MVA RESPONSE)	Tealium trackView	Remote cmd response	-

---

7. Production Performance Data - 200K Kullanici Bazli

7.1 RAM Segmentlerine Gore Detayli Yukleme Sureleri

Dusuk RAM Segment (3-4GB) - 30.5% Kullanici = 3.37M

0ms            3900ms                 7400ms
|               |                      |
Dashboard     AOM Request           AOM Response
Init          Gonderildi            Alindi
              
              [---- 3.5s network ----]
              
              Render: 1.8s (Dashboard componentleri)

Metrikler:

Asamalar	Sure
Dashboard acilisi → AOM istegi gonderimi	3.9s
AOM istegi → AOM yaniti	3.5s
Toplam (Dashboard → AOM response)	7.4s
Dashboard komponentlerinin render suresi	1.8s

Karakteristikler:

• En buyuk kullanici segmenti (3.37M kullanici)
• Dashboard hazirlik suresi cok uzun (3.9s)
• AOM network latency yuksek (3.5s)
• Render suresi yuksek (1.8s) - UI cizim yavasi

---

Orta RAM Segment (4-6GB) - 26.3% Kullanici = 2.91M

0ms         3100ms              6100ms
|            |                   |
Dashboard  AOM Request        AOM Response
Init       Gonderildi         Alindi

           [--- 3.0s network ---]
           
           Render: 0.9s

Metrikler:

Asamalar	Sure
Dashboard acilisi → AOM istegi gonderimi	3.1s
AOM istegi → AOM yaniti	3.0s
Toplam (Dashboard → AOM response)	6.1s
Dashboard komponentlerinin render suresi	0.9s

Karakteristikler:

• Ikinci buyuk segment (2.91M kullanici)
• Dashboard hazirlik suresi orta (3.1s)
• AOM network latency orta (3.0s)
• Render suresi dusuk (0.9s)

---

Yuksek RAM Segment (6-8GB) - 20.6% Kullanici = 2.28M

0ms      2500ms           4900ms
|         |                |
Dash    AOM Req         AOM Resp
Init    Gonderildi      Alindi

        [-- 2.4s network --]
        
        Render: 0.8s

Metrikler:

Asamalar	Sure
Dashboard acilisi → AOM istegi gonderimi	2.5s
AOM istegi → AOM yaniti	2.4s
Toplam (Dashboard → AOM response)	4.9s
Dashboard komponentlerinin render suresi	0.8s

Karakteristikler:

• Ucuncu buyuk segment (2.28M kullanici)
• Dashboard hazirlik suresi iyi (2.5s)
• AOM network latency iyi (2.4s)
• Render suresi dusuk (0.8s)

---

Premium RAM Segment (8-12GB+) - 3.0% Kullanici = 327K

0ms   1800ms        3700ms
|      |             |
Dash  AOM Req     AOM Resp
Init  Gonderildi  Alindi

      [- 1.9s net -]
      
      Render: 0.4s

Metrikler:

Asamalar	Sure
Dashboard acilisi → AOM istegi gonderimi	1.8s
AOM istegi → AOM yaniti	1.9s
Toplam (Dashboard → AOM response)	3.7s
Dashboard komponentlerinin render suresi	0.4s

Karakteristikler:

• En kucuk segment (327K kullanici)
• Dashboard hazirlik suresi cok hizli (1.8s)
• AOM network latency hizli (1.9s)
• Render suresi cok dusuk (0.4s)
• Optimal performance

---

7.2 Karsilastirma Tablosu (200K Kullanici Production Verisi)

Metric	3-4GB RAM	4-6GB RAM	6-8GB RAM	8-12GB+ RAM
Kullanici Payi	30.5%	26.3%	20.6%	3.0%
Etkilenen Kullanici	3.37M	2.91M	2.28M	327K
Dashboard → AOM Request	3.9s	3.1s	2.5s	1.8s
AOM Request → Response	3.5s	3.0s	2.4s	1.9s
Toplam Dashboard Yukleme	7.4s	6.1s	4.9s	3.7s
Dashboard Render Suresi	1.8s	0.9s	0.8s	0.4s
RAM vs Baseline (8GB+) Fark	+100%	+65%	+32%	Baseline

Kritik Cikarimlar:

1. Dashboard hazirlik suresi RAM ile ters orantili: 3.9s → 1.8s (2.2x fark)
2. AOM network latency RAM ile ters orantili: 3.5s → 1.9s (1.8x fark) - muhtemelen istek hazirlik + network processing
3. Render suresi RAM ile ters orantili: 1.8s → 0.4s (4.5x fark) - en buyuk fark burda
4. %56.8 kullanici 3-6GB segmentinde, ortalama 6.8s yukleme yasiyorlar

---

7.3 Legacy Test Cihazi Verileri (Referans)

Onceki test cihazlariyla yapilan olcumler (referans amacli):

Metric	Device A (Dusuk)	Device C (Orta)	Device B (Yuksek)
getAppIcons	1463ms	368ms	256ms
getMainPageOffer	1474ms	368ms	275ms
setTobiWelcomeMessage	1473ms	562ms	420ms
AOM request	4395ms	615ms	335ms
Toplam	6913ms	1724ms	794ms

Not: Bu veriler 3 test cihazi bazlidir. Production verileri (Section 7.1-7.2) 200K kullanici bazli olup daha gercekci.

---

8. Tealium Target Entegrasyon Akisi

Akis:

1. Dashboard init - startAomMva() cagirilir
2. Timer baslar - setTargetTimeoutTimer(2000ms veya config'den)
3. 3 API paralel tetiklenir - ViewModel init icinde
4. Her response gelince - ViewModel LiveData gunceller
5. Karar noktasi:
   • isAllTargetDataEnabled() == true VE isTargetRun == false → Normal akis
   • Timer dolarsa VE isTargetRun == false → Timeout akis
6. isTargetRun = true - Sadece 1 kez calisir (guard)
7. trackAomCall() - TealiumHelper.trackView("aom_mva") + markAomCall()
8. Server response - homeRemoteCommand callback - markAomResponse()

Guard Mekanizmasi:

isTargetRun = false (baslangic)
  |
  +--> Normal path tetiklerse -> isTargetRun = true -> Timer artik skip
  |
  +--> Timer tetiklerse -> isTargetRun = true -> Observer artik skip

---

9. Tealium Event Envanteri

fetch_aom (AOM MVA CALL)

Ozellik	Deger
Event type	view
Tetikleyici	Tum API response VEYA 2s timeout
Guard	isTargetRun == false
Data	getTealiumData() veya null

aom_mva (AOM MVA RESPONSE)

Ozellik	Deger
Event type	view
Tetikleyici	Remote command response
Data	page_name, session_id, container_name, offer JSON, source

---

10. Delay and Timer Envanteri

Timer	Sure	Kaynak	Amac	Fail Davranisi
setTargetTimeoutTimer	2000ms (default)	ConfigManager.mccmTobiTargetData.timeoutDuration	API'ler gelmezse zorla AOM call	Partial data veya null ile devam
SDK Init	~33-485ms	Jetpack Startup	SDK hazir olmasi	AOM call gecikir
fetch_aom Network	8s	Tealium SDK	Server response	Timeout, no response
Target Offer Delay	0-1000ms	Remote config	UI settle	Hemen goster

setTargetTimeoutTimer detay:

• Config'den gelir: ConfigManager.getConfig().mccmTobiTargetData.timeoutDuration
• Null ise default: HomeConstants.TARGET_DELAY = 2000L
• Dispatchers.IO uzerinde calisir
• isTargetRun flag ile sadece 1 kez execute edilir
• isAtLeastOneTargetDataEnabled() kontrol eder (en az 1 API response veya error)

---

11. Mimari Notlar ve Cikarimlar

1. Race condition yonetimi: isTargetRun flag ile timer ve observer arasinda sadece ilk tetiklenen kazanir

2. Bottleneck 3-4GB RAM: Dashboard hazirlik 3.9s + AOM network 3.5s = 7.4s toplam (en buyuk segment)

3. Bottleneck 4-6GB RAM: Dashboard hazirlik 3.1s + AOM network 3.0s = 6.1s toplam (ikinci buyuk segment)

4. Render suresi RAM'e cok bagli: 1.8s (dusuk) → 0.4s (yuksek) - 4.5x fark

5. AOM network latency RAM'e bagli: Muhtemelen dusuk RAM cihazlar istek hazirlik + network processing'de yavasliyor

6. Toplam yukleme 2x fark: 7.4s (3-4GB) vs 3.7s (8GB+)

7. Thread-safe: Remote command [IO] - postValue() - Observer [Main]

8. Priority: Tealium > API (remote command data varsa override eder)

Kritik Path (200K Kullanici Verisi)

3-4GB RAM (30.5% kullanici - EN BUYUK SEGMENT):

Dashboard Init (3.9s) → AOM Network (3.5s) → Render (1.8s) = 7.4s total

4-6GB RAM (26.3% kullanici - IKINCI BUYUK SEGMENT):

Dashboard Init (3.1s) → AOM Network (3.0s) → Render (0.9s) = 6.1s total

6-8GB RAM (20.6% kullanici):

Dashboard Init (2.5s) → AOM Network (2.4s) → Render (0.8s) = 4.9s total

8GB+ RAM (3% kullanici - BASELINE):

Dashboard Init (1.8s) → AOM Network (1.9s) → Render (0.4s) = 3.7s total

Bagimlilik Grafigi:

Dashboard Init
  |
  +---> setTargetTimeoutTimer(2s) ----timeout---+
  |                                             |
  +---> getAppIcons --------+                   |
  +---> getMainPageOffer ---+-- isAllTarget? ---+---> trackAomCall (ilk gelen kazanir)
  +---> setTobiWelcomeMsg --+                   |
                                                |
  isTargetRun = true <--------------------------+

---

11a. RAM Segmentleri ile Performance Mapping - 200K Kullanici

200K Production Data ile Segment Analizi

RAM Segment	Kullanici Payi	Kullanici Sayisi	Toplam Sure	Dashboard → AOM	AOM Network	Render	Optimizasyon Onceligi
3-4GB	30.5%	3.37M	7.4s	3.9s	3.5s	1.8s	CRITICAL
4-6GB	26.3%	2.91M	6.1s	3.1s	3.0s	0.9s	HIGH
6-8GB	20.6%	2.28M	4.9s	2.5s	2.4s	0.8s	MEDIUM
8-12GB+	3.0%	327K	3.7s	1.8s	1.9s	0.4s	LOW

Segment Bazli Akis Analizi (Guncellenmis)

3-4GB RAM Segment - 30.5% Kullanici = 3.37M [EN KRITIK]

Karakteristikler:

• En buyuk kullanici segmenti (3.37M kullanici)
• 431 cihaz modeli (yuksek fragmantasyon)
• Dashboard hazirlik cok yavas (3.9s)
• AOM network latency yuksek (3.5s)
• Render cok yavas (1.8s)
• Toplam 7.4s - kullanici deneyimi ZAYIF

Optimizasyon Onerisi (ONCELIKLI):

1. Dashboard init optimizasyonu: 3.9s → 2.5s (-1.4s)

   • API cagrilarini parallelize et
   • Lazy loading uygula
   • Gereksiz view hierarchy azalt

2. AOM network iyilestirme: 3.5s → 2.5s (-1.0s)

   • Request payload kucult
   • Gzip compression
   • CDN kullan

3. Render optimizasyonu: 1.8s → 1.0s (-0.8s)

   • RecyclerView view recycling iyilestir
   • Overdraw azalt
   • Layout hierarchy basitlestir

Hedef: 7.4s → 5.0s (-2.4s, %32 iyilestirme)

---

4-6GB RAM Segment - 26.3% Kullanici = 2.91M [YUKSEK ONCELIK]

Karakteristikler:

• Ikinci buyuk segment (2.91M kullanici)
• 246 cihaz modeli (orta fragmantasyon)
• Dashboard hazirlik orta (3.1s)
• AOM network latency orta (3.0s)
• Render iyi (0.9s)
• Toplam 6.1s - kullanici deneyimi ORTA

Optimizasyon Onerisi:

1. Dashboard init: 3.1s → 2.0s (-1.1s)
2. AOM network: 3.0s → 2.0s (-1.0s)
3. Render zaten iyi (0.9s)

Hedef: 6.1s → 4.0s (-2.1s, %34 iyilestirme)

---

6-8GB RAM Segment - 20.6% Kullanici = 2.28M [ORTA ONCELIK]

Karakteristikler:

• Ucuncu buyuk segment (2.28M kullanici)
• 288 cihaz modeli
• Dashboard hazirlik iyi (2.5s)
• AOM network iyi (2.4s)
• Render iyi (0.8s)
• Toplam 4.9s - kullanici deneyimi IYI

Optimizasyon Onerisi:

• Minor iyilestirmeler (target: 4.0s)
• Regression onleme odakli

Hedef: 4.9s → 4.0s (-0.9s, %18 iyilestirme)

---

8-12GB+ RAM Segment - 3.0% Kullanici = 327K [DUSUK ONCELIK]

Karakteristikler:

• En kucuk segment (327K kullanici)
• 62 cihaz modeli
• Tum metrikler optimal
• Toplam 3.7s - kullanici deneyimi MUKEMMEL

Optimizasyon Onerisi:

• Optimizasyon gerekmez
• Baseline olarak kullan

---

Onem Sirasi (Impact x Frequency x Severity)

Oncelik	Segment	Kullanici	Sure	Impact Score	Optimizasyon ROI
1	3-4GB RAM	3.37M (30.5%)	7.4s	24,938	En Yuksek
2	4-6GB RAM	2.91M (26.3%)	6.1s	17,751	Yuksek
3	6-8GB RAM	2.28M (20.6%)	4.9s	11,172	Orta
4	8-12GB+ RAM	327K (3.0%)	3.7s	1,210	Dusuk

Impact Score = Kullanici Sayisi × Sure (saniye)

Hedef KPI'lar (RAM Segment Bazli) - Guncellenmis

Segment	Mevcut	Target	Iyilestirme	Etkilenen Kullanici	Risk
3-4GB RAM	7.4s	5.0s	-2.4s (-32%)	3.37M	Yuksek
4-6GB RAM	6.1s	4.0s	-2.1s (-34%)	2.91M	Orta
6-8GB RAM	4.9s	4.0s	-0.9s (-18%)	2.28M	Dusuk
8-12GB+ RAM	3.7s	3.7s	Baseline	327K	N/A

Optimizasyon Roadmap

Phase 1 (Q3 2026) - Quick Wins:

• 3-4GB: Dashboard view hierarchy azalt (-0.5s)
• 4-6GB: API request parallelization (-0.8s)
• Gzip compression tum segmentlerde (-0.3s)

Phase 2 (Q4 2026) - Major Improvements:

• 3-4GB: Lazy loading implementation (-1.0s)
• 3-4GB: Render pipeline optimization (-0.6s)
• 4-6GB: AOM request payload optimization (-0.7s)

Phase 3 (Q1 2027) - Polish:

• 6-8GB: Fine tuning (-0.5s)
• Tum segmentler: CDN migration (-0.4s)
• A/B testing ve final optimization

---

12. Test Stratejisi - RAM Segment Bazli (Guncellenmis)

Cihaz Secimi Kriterleri

Market Coverage: %56.8 kullanici icin 3-6GB cihazlar MANDATORY

Test Onceligi:

Oncelik	RAM Segment	Cihaz Secimi	Kullanici Payi	Baseline Sure	Target Sure
P0	3-4GB	Redmi Note 8, Galaxy A32	30.5%	7.4s	5.0s
P0	4-6GB	Pixel 4a, Redmi Note 10	26.3%	6.1s	4.0s
P1	6-8GB	Galaxy S20, OnePlus 8	20.6%	4.9s	4.0s
P2	8-12GB+	Pixel 6 Pro, S21 Ultra	3.0%	3.7s	3.7s

Regression Test Checklist (Guncellenmis)

3-4GB RAM Cihazlar (CRITICAL):

• Dashboard init sure <4.0s (target: 2.5s)
• AOM network latency <3.0s (target: 2.5s)
• Render completion <1.2s (target: 1.0s)
• Toplam sure <5.5s (target: 5.0s)

4-6GB RAM Cihazlar (HIGH):

• Dashboard init sure <2.5s (target: 2.0s)
• AOM network latency <2.5s (target: 2.0s)
• Render completion <1.0s (baseline)
• Toplam sure <4.5s (target: 4.0s)

6-8GB RAM Cihazlar (MEDIUM):

• Dashboard init sure <2.8s
• AOM network latency <2.5s
• Render completion <1.0s
• Toplam sure <4.5s (target: 4.0s)

8-12GB+ RAM Cihazlar (BASELINE):

• No regression (baseline: 3.7s)
• Tum metrikler stabil kalmali

Performance Testing Tools

• Firebase Performance: Track real user metrics
• Android Profiler: CPU, memory, network analysis
• Systrace: Frame rendering analysis
• Network Profiler: AOM request/response inspection

---

13. Sonuc ve Oneri

Mevcut Durum (200K Kullanici Verisi)

Kullanici Dagilimi:

• %56.8 kullanici (6.28M) 3-6GB RAM segmentinde
• Bu segment ortalama 6.8s yukleme yasıyor
• %30.5 kullanici (3.37M) 7.4s bekliyor (EN KRITIK)

Bottlenecklar:

1. Dashboard initialization: 1.8s - 3.9s (RAM'e gore)
2. AOM network latency: 1.9s - 3.5s (RAM'e gore)
3. UI rendering: 0.4s - 1.8s (RAM'e gore, 4.5x fark)

Oneri ve Aksiyon Plani

Immediate Actions (Sprint 1-2):

1. 3-4GB segment icin dashboard init optimizasyonu (target: -1.0s)
2. AOM request payload size reduction (target: -0.5s)
3. View hierarchy flattening (target: -0.3s)

Short-term (Q3 2026):

1. API parallelization tam implementasyon
2. Lazy loading stratejisi
3. Network optimization (gzip, CDN)

Long-term (Q4 2026 - Q1 2027):

1. Architectural refactoring (modularization)
2. Progressive loading implementation
3. Adaptive performance based on device tier

Success Metrics:

• 3-4GB segment: 7.4s → 5.0s (-32% improvement)
• 4-6GB segment: 6.1s → 4.0s (-34% improvement)
• 6.28M kullanici icin ortalama -2.3s iyilestirme

---

Doküman Sonu