[{"data":1,"prerenderedAt":610},["ShallowReactive",2],{"article-alternates":3,"article-\u002Fde\u002Fmarketing\u002Fios-17-nach-attribution-stack":13},{"i18nKey":4,"paths":5},"marketing-003-2026-06",{"de":6,"en":7,"es":8,"fr":9,"it":10,"ru":11,"tr":12},"\u002Fde\u002Fmarketing\u002Fios-17-nach-attribution-stack","\u002Fen\u002Fmarketing\u002Fios-17-post-attribution-stack","\u002Fes\u002Fmarketing\u002Fstack-atribucin-post-ios-17","\u002Ffr\u002Fmarketing\u002Fstack-attribution-ios-17-post","\u002Fit\u002Fmarketing\u002Fios-17-sonrasi-attribution-stack","\u002Fru\u002Fmarketing\u002Fios-17-sonrasi-atribucija-stack","\u002Ftr\u002Fmarketing\u002Fios-17-sonrasi-ad-attribution-stacki",{"_path":6,"_dir":14,"_draft":15,"_partial":15,"_locale":16,"title":17,"description":18,"publishedAt":19,"modifiedAt":19,"category":14,"i18nKey":4,"tags":20,"readingTime":26,"author":27,"body":28,"_type":604,"_id":605,"_source":606,"_file":607,"_stem":608,"_extension":609},"marketing",false,"","iOS 17 nach: Der neue Attribution-Stack","ATT, SKAdNetwork 4 und modeled conversions: So rekonstruierst du Attribution auf iOS – praktischer Leitfaden für die Post-Lookback-Maturitätsphase.","2026-06-02",[21,22,23,24,25],"ios-attribution","skadnetwork","att","modeled-conversions","mobile-measurement",9,"Roibase",{"type":29,"children":30,"toc":590},"root",[31,39,46,51,62,72,82,87,93,98,105,110,118,153,158,164,169,174,180,196,202,207,215,249,259,269,275,280,288,321,331,337,342,495,503,531,536,542,550,555,563,568,576,581,585],{"type":32,"tag":33,"props":34,"children":35},"element","p",{},[36],{"type":37,"value":38},"text","Fünf Jahre sind vergangen, seit Apple iOS 14.5 und App Tracking Transparency einführte. Seitdem haben sich die Grundannahmen des mobilen Performance Marketing grundlegend verschoben. Deterministische User-Level-Attribution ist tot, probabilistische und aggregierte Modelle sind zur Pflicht geworden. Mit iOS 17 und SKAdNetwork 4 wird das Spiel neu geschrieben: Das neue Conversion-Value-Schema, das Post-Lookback-Fenster und modeled conversions ermöglichen es dir, Attribution neu zu denken. Dieser Leitfaden zeigt dir, wie du iOS-Attribution 2026 aufbaust, welche Signale du in welcher Reihenfolge nutzt und wie du MMP + Incrementality-Tests kombinierst.",{"type":32,"tag":40,"props":41,"children":43},"h2",{"id":42},"anatomie-der-attribution-nach-att",[44],{"type":37,"value":45},"Anatomie der Attribution nach ATT",{"type":32,"tag":33,"props":47,"children":48},{},[49],{"type":37,"value":50},"Vor iOS 14.5 konnten MMP (Adjust, AppsFlyer, Kochava) die IDFA auf Geräteebene auslesen und jeden Conversion direkt einer Kampagne zuordnen. Mit ATT wurde dieser Mechanismus für 95 % der Nutzer geschlossen (Statista 2025: Opt-in-Rate bei 7 %). Jetzt gibt es drei Schichten:",{"type":32,"tag":33,"props":52,"children":53},{},[54,60],{"type":32,"tag":55,"props":56,"children":57},"strong",{},[58],{"type":37,"value":59},"1. Deterministisch (IDFA-Opt-in-Nutzer):",{"type":37,"value":61}," Die 7 % der Nutzer, die Tracking erlauben, nutzen weiterhin den klassischen MMP-Fluss. Click\u002FImpression-Timestamp, Install, In-App-Event — alles auf User-Ebene. Aber dieses Segment ist keine repräsentative Stichprobe mehr.",{"type":32,"tag":33,"props":63,"children":64},{},[65,70],{"type":32,"tag":55,"props":66,"children":67},{},[68],{"type":37,"value":69},"2. SKAdNetwork (aggregierte Postback):",{"type":37,"value":71}," Apples eigenes Datenschutz-Framework. Attribution Window: 0–72 Stunden; Conversion Value: 6-Bit-Encoding (0–63). Mit SKAdNetwork 4 wurden zweite und dritte Postback hinzugefügt (8–35 Tage), sodass D7–D30 Retention jetzt messbar ist.",{"type":32,"tag":33,"props":73,"children":74},{},[75,80],{"type":32,"tag":55,"props":76,"children":77},{},[78],{"type":37,"value":79},"3. Modeled conversions:",{"type":37,"value":81}," Machine-Learning-Prognosen von MMP. Sie kombinieren aggregierte Click-\u002FImpression-Daten, Install-Zahlen und SKAN-Signale. Zuverlässigkeit niedriger als deterministisch, aber skalierbar.",{"type":32,"tag":33,"props":83,"children":84},{},[85],{"type":37,"value":86},"Diese drei Schichten musst du zusammen nutzen. Keine reicht allein: IDFA ist zu eng, SKAN aggregiert und verzögert, modeled conversions sind prognostisch. Ein Stack, der diese drei ins Gleichgewicht bringt, ist jetzt Core Competency.",{"type":32,"tag":40,"props":88,"children":90},{"id":89},"was-skadnetwork-4-bringt",[91],{"type":37,"value":92},"Was SKAdNetwork 4 bringt",{"type":32,"tag":33,"props":94,"children":95},{},[96],{"type":37,"value":97},"SKAdNetwork 4 (iOS 16.1 eingeführt, iOS 17 gereift) hat drei große Neuerungen:",{"type":32,"tag":99,"props":100,"children":102},"h3",{"id":101},"conversion-value-hierarchie-und-postback-kette",[103],{"type":37,"value":104},"Conversion-Value-Hierarchie und Postback-Kette",{"type":32,"tag":33,"props":106,"children":107},{},[108],{"type":37,"value":109},"Statt eines einzelnen 6-Bit-Werts gibt es jetzt drei Postbacks: erster 0–2 Tage, zweiter 3–7 Tage, dritter 8–35 Tage. Jeder Postback hat seinen eigenen 6-Bit-Wert. So kannst du frühes IAP-Signal (Install-to-Purchase \u003C48h) im zweiten Postback von Retention-Signal (D3–D7 Session Count) im dritten trennen. Früher musstest du alle Signale in 64 Slots quetschen; jetzt hast du 64×3=192 Kombinationen (praktisch 64+64+64, sequential encoding).",{"type":32,"tag":33,"props":111,"children":112},{},[113],{"type":32,"tag":55,"props":114,"children":115},{},[116],{"type":37,"value":117},"Beispiel-Mapping:",{"type":32,"tag":119,"props":120,"children":121},"ul",{},[122,133,143],{"type":32,"tag":123,"props":124,"children":125},"li",{},[126,131],{"type":32,"tag":55,"props":127,"children":128},{},[129],{"type":37,"value":130},"Postback 1 (0–2 Tage):",{"type":37,"value":132}," D0 IAP-Status (0=kein Event, 1–10=Revenue-Bracket, 11–20=spezifische SKU, 21–63=Custom-Blend)",{"type":32,"tag":123,"props":134,"children":135},{},[136,141],{"type":32,"tag":55,"props":137,"children":138},{},[139],{"type":37,"value":140},"Postback 2 (3–7 Tage):",{"type":37,"value":142}," D3–D7 Retention-Tier (0=Churn, 1–20=Session-Count-Band, 21–40=Engagement-Tiefe)",{"type":32,"tag":123,"props":144,"children":145},{},[146,151],{"type":32,"tag":55,"props":147,"children":148},{},[149],{"type":37,"value":150},"Postback 3 (8–35 Tage):",{"type":37,"value":152}," D30 LTV-Proxy (0–63=kumulatives Revenue-Bracket)",{"type":32,"tag":33,"props":154,"children":155},{},[156],{"type":37,"value":157},"Um diese Struktur richtig zu nutzen, musst du dein Conversion-Value-Mapping wöchentlich überprüfen. Wenn sich das Nutzerverhalten ändert, ändert sich auch, welches Signal den meisten Informationswert liefert.",{"type":32,"tag":99,"props":159,"children":161},{"id":160},"source-identifier-und-hierarchische-source-id",[162],{"type":37,"value":163},"Source Identifier und hierarchische Source ID",{"type":32,"tag":33,"props":165,"children":166},{},[167],{"type":37,"value":168},"SKAdNetwork 4 zeigt die ID der Publisher-App und untergeordneter Netzwerke in einer vierstufigen Hierarchie. Du siehst nicht mehr nur \"von Meta\", sondern \"Meta → Audience Network → Publisher App X\" (wenn das Ad Network es offenlegt). So kannst du Sub-Publisher-Performance vergleichen.",{"type":32,"tag":33,"props":170,"children":171},{},[172],{"type":37,"value":173},"In der Praxis geben Walled Gardens wie Facebook, TikTok und Google dieses Feld nicht vollständig preis. Aber für programmgesteuerte und Rewarded-Video-Netzwerke schafft es kritischen Mehrwert.",{"type":32,"tag":99,"props":175,"children":177},{"id":176},"web-to-app-attribution-unterstützung",[178],{"type":37,"value":179},"Web-to-App-Attribution-Unterstützung",{"type":32,"tag":33,"props":181,"children":182},{},[183,185,194],{"type":37,"value":184},"Mit iOS 17.4 unterstützt SKAdNetwork auch Web-Clicks. Wenn ein Nutzer von einem Safari-Banner zum App Store geht und installiert, geht das in den SKAN-Postback ein. Für Marken mit kombinierter Web + App UA-Strategie wird es möglich, dieses Signal mit ",{"type":32,"tag":186,"props":187,"children":191},"a",{"href":188,"rel":189},"https:\u002F\u002Fwww.roibase.com.tr\u002Fde\u002Fppc",[190],"nofollow",[192],{"type":37,"value":193},"Performance-Marketing (PPC)",{"type":37,"value":195},"-Kampagnen zu verbinden und Cross-Channel-Incrementality zu berechnen.",{"type":32,"tag":40,"props":197,"children":199},{"id":198},"modeled-conversions-wie-es-funktioniert-wann-es-zuverlässig-ist",[200],{"type":37,"value":201},"Modeled Conversions: Wie es funktioniert, wann es zuverlässig ist",{"type":32,"tag":33,"props":203,"children":204},{},[205],{"type":37,"value":206},"Modeled conversions kombinieren MMP-Daten — SKAN-Postbacks, aggregierte Impression-\u002FClick-Zahlen und Install-Count — über maschinelles Lernen für probabilistische Attribution. AppsFlyer nennt es \"predictive analytics\", Adjust \"statistical modeling\" — technisch dasselbe: Regression + Bayesian Inference.",{"type":32,"tag":33,"props":208,"children":209},{},[210],{"type":32,"tag":55,"props":211,"children":212},{},[213],{"type":37,"value":214},"Bedingungen für Zuverlässigkeit:",{"type":32,"tag":216,"props":217,"children":218},"ol",{},[219,229,239],{"type":32,"tag":123,"props":220,"children":221},{},[222,227],{"type":32,"tag":55,"props":223,"children":224},{},[225],{"type":37,"value":226},"Ausreichende Datengröße:",{"type":37,"value":228}," Mindestens 500+ Installs pro Tag, 50+ Conversions pro Kampagne (SKAN oder IDFA). Darunter overfittet das Modell.",{"type":32,"tag":123,"props":230,"children":231},{},[232,237],{"type":32,"tag":55,"props":233,"children":234},{},[235],{"type":37,"value":236},"Konsistenz des SKAN-Signals:",{"type":37,"value":238}," Dein Conversion-Value-Mapping muss stabil sein. Wenn du es täglich änderst, kann das Modell historische Muster nicht erfassen.",{"type":32,"tag":123,"props":240,"children":241},{},[242,247],{"type":32,"tag":55,"props":243,"children":244},{},[245],{"type":37,"value":246},"Kalibrierung durch Incrementality-Test:",{"type":37,"value":248}," Mindestens einmal pro Quartal führst du ein Geo-Holdout oder Time-based Holdout durch. Du vergleichst modeled Zahlen mit echtem Lift und wendest Bias-Korrektur an.",{"type":32,"tag":33,"props":250,"children":251},{},[252,257],{"type":32,"tag":55,"props":253,"children":254},{},[255],{"type":37,"value":256},"Schlechtes Beispiel:",{"type":37,"value":258}," Du startest eine neue Kampagne, 3 Tage später 20 Installs, das MMP sagt \"modeled 15 IAPs\". Das ist rein Rauschen — die Sample-Größe ist unzureichend. Warte mindestens 2 Wochen.",{"type":32,"tag":33,"props":260,"children":261},{},[262,267],{"type":32,"tag":55,"props":263,"children":264},{},[265],{"type":37,"value":266},"Gutes Beispiel:",{"type":37,"value":268}," 30 Tage lang bringen Meta + TikTok + Google UAC insgesamt 50K Installs, aus SKAN kommen 3K Conversion-Postbacks. Das MMP modelliert das auf 8K. Im selben Zeitraum zeigt dein Geo-Test-Holdout (Frankreich vs. Deutschland) +12% Lift. Du revidierst die modeled Zahl auf 8K × 1,12 = 8,96K. Das ist zuverlässig.",{"type":32,"tag":40,"props":270,"children":272},{"id":271},"post-lookback-maturität-signale-nach-tag-35",[273],{"type":37,"value":274},"Post-Lookback-Maturität: Signale nach Tag 35",{"type":32,"tag":33,"props":276,"children":277},{},[278],{"type":37,"value":279},"Der dritte Postback von SKAdNetwork 4 erfasst Ereignisse zwischen 8–35 Tagen. Nach Tag 35 kommt kein SKAN-Postback mehr. Aber das echte Nutzerverhalten endet nicht bei Tag 35: D60 Retention, D90 LTV, jährliche Abonnementverlängerungen.",{"type":32,"tag":33,"props":281,"children":282},{},[283],{"type":32,"tag":55,"props":284,"children":285},{},[286],{"type":37,"value":287},"Lösungsansätze:",{"type":32,"tag":216,"props":289,"children":290},{},[291,301,311],{"type":32,"tag":123,"props":292,"children":293},{},[294,299],{"type":32,"tag":55,"props":295,"children":296},{},[297],{"type":37,"value":298},"Kohorten-basierte LTV-Projektion:",{"type":37,"value":300}," Passe mit deinen SKAN + modeled conversion Daten der ersten 35 Tage eine Kohorten-LTV-Kurve an (meist Power Law oder exponentieller Zerfall). Extrapoliere 90–180-Tage-LTV. Diese Prognose ist mit Unsicherheit behaftet, aber bei ausreichender Kohortengröße ist die Varianz gering.",{"type":32,"tag":123,"props":302,"children":303},{},[304,309],{"type":32,"tag":55,"props":305,"children":306},{},[307],{"type":37,"value":308},"Cross-Channel-Holdout und Incrementality:",{"type":37,"value":310}," Pausiere einen Kanal 2 Wochen, miss Änderungen bei organischen Installs und In-App-Revenue. Berechne Net Incrementality. Nutze das quarterly, um Post-35-Tage-Signale zu backfill.",{"type":32,"tag":123,"props":312,"children":313},{},[314,319],{"type":32,"tag":55,"props":315,"children":316},{},[317],{"type":37,"value":318},"Server-seitige Event-Anreicherung:",{"type":37,"value":320}," Schicke späte Events (Subscription Renewal, High-Ticket IAP), die nicht im SKAN-Postback sind, Server-to-Server ans MMP. Das ist nicht deterministisch, aber erzeugt aggregate Muster. Das MMP nutzt es als Modell-Input.",{"type":32,"tag":33,"props":322,"children":323},{},[324,329],{"type":32,"tag":55,"props":325,"children":326},{},[327],{"type":37,"value":328},"Achtung:",{"type":37,"value":330}," Apple verbietet nicht explizit, Server-seitige User-Level-Signale zu schicken, aber ein MMP kann dafür keine User-Level-Attribution-Claims machen. Als Aggregate-Modeling-Input ist es OK.",{"type":32,"tag":40,"props":332,"children":334},{"id":333},"praktisches-stack-setup-szenario",[335],{"type":37,"value":336},"Praktisches Stack-Setup-Szenario",{"type":32,"tag":33,"props":338,"children":339},{},[340],{"type":37,"value":341},"Angenommen, du hast eine Abonnement-Fitness-App. 60 % des Install-Base sind iOS, Ziel: 100K neue Installs monatlich. Dein Attribution Stack:",{"type":32,"tag":343,"props":344,"children":345},"table",{},[346,375],{"type":32,"tag":347,"props":348,"children":349},"thead",{},[350],{"type":32,"tag":351,"props":352,"children":353},"tr",{},[354,360,365,370],{"type":32,"tag":355,"props":356,"children":357},"th",{},[358],{"type":37,"value":359},"Schicht",{"type":32,"tag":355,"props":361,"children":362},{},[363],{"type":37,"value":364},"Tool",{"type":32,"tag":355,"props":366,"children":367},{},[368],{"type":37,"value":369},"Rolle",{"type":32,"tag":355,"props":371,"children":372},{},[373],{"type":37,"value":374},"Vertrauensbereich",{"type":32,"tag":376,"props":377,"children":378},"tbody",{},[379,403,426,449,472],{"type":32,"tag":351,"props":380,"children":381},{},[382,388,393,398],{"type":32,"tag":383,"props":384,"children":385},"td",{},[386],{"type":37,"value":387},"SKAN-Postback",{"type":32,"tag":383,"props":389,"children":390},{},[391],{"type":37,"value":392},"AppsFlyer",{"type":32,"tag":383,"props":394,"children":395},{},[396],{"type":37,"value":397},"Erstes 35-Tage-Conversion-Value + Source ID",{"type":32,"tag":383,"props":399,"children":400},{},[401],{"type":37,"value":402},"95 % (Apple verifiziert)",{"type":32,"tag":351,"props":404,"children":405},{},[406,411,416,421],{"type":32,"tag":383,"props":407,"children":408},{},[409],{"type":37,"value":410},"Modeled Conversions",{"type":32,"tag":383,"props":412,"children":413},{},[414],{"type":37,"value":415},"AppsFlyer Predictive",{"type":32,"tag":383,"props":417,"children":418},{},[419],{"type":37,"value":420},"SKAN + Aggregate für probabilistische Attribution",{"type":32,"tag":383,"props":422,"children":423},{},[424],{"type":37,"value":425},"70–80 % (Geo-Test-Kalibrierung)",{"type":32,"tag":351,"props":427,"children":428},{},[429,434,439,444],{"type":32,"tag":383,"props":430,"children":431},{},[432],{"type":37,"value":433},"IDFA Opt-in",{"type":32,"tag":383,"props":435,"children":436},{},[437],{"type":37,"value":438},"AppsFlyer Raw Data",{"type":32,"tag":383,"props":440,"children":441},{},[442],{"type":37,"value":443},"7 % deterministisches Segment",{"type":32,"tag":383,"props":445,"children":446},{},[447],{"type":37,"value":448},"100 % (aber geringe Repräsentativität)",{"type":32,"tag":351,"props":450,"children":451},{},[452,457,462,467],{"type":32,"tag":383,"props":453,"children":454},{},[455],{"type":37,"value":456},"Incrementality",{"type":32,"tag":383,"props":458,"children":459},{},[460],{"type":37,"value":461},"GeoLift (Meta) + Custom Holdout",{"type":32,"tag":383,"props":463,"children":464},{},[465],{"type":37,"value":466},"Kanal-Level-Lift-Messung",{"type":32,"tag":383,"props":468,"children":469},{},[470],{"type":37,"value":471},"90 % (statistisch, aber teuer)",{"type":32,"tag":351,"props":473,"children":474},{},[475,480,485,490],{"type":32,"tag":383,"props":476,"children":477},{},[478],{"type":37,"value":479},"LTV-Projektion",{"type":32,"tag":383,"props":481,"children":482},{},[483],{"type":37,"value":484},"Interne dbt + BigQuery",{"type":32,"tag":383,"props":486,"children":487},{},[488],{"type":37,"value":489},"Kohorten-Kurven-Fit, 90–180 Tage Prognose",{"type":32,"tag":383,"props":491,"children":492},{},[493],{"type":37,"value":494},"60–70 % (Modellgenauigkeit)",{"type":32,"tag":33,"props":496,"children":497},{},[498],{"type":32,"tag":55,"props":499,"children":500},{},[501],{"type":37,"value":502},"Fluss:",{"type":32,"tag":216,"props":504,"children":505},{},[506,511,516,521,526],{"type":32,"tag":123,"props":507,"children":508},{},[509],{"type":37,"value":510},"Tägliche SKAN-Postback-Abzüge pro Kampagne.",{"type":32,"tag":123,"props":512,"children":513},{},[514],{"type":37,"value":515},"AppsFlyer modeled conversions abrufen, aber bei Kampagnen-Level-CPA 20 % Vertrauensbereich einbauen.",{"type":32,"tag":123,"props":517,"children":518},{},[519],{"type":37,"value":520},"Monatlich ein Geo-Holdout ausführen (z. B. Meta in Spanien pausieren, in Portugal weiterlaufen). Net Lift berechnen.",{"type":32,"tag":123,"props":522,"children":523},{},[524],{"type":37,"value":525},"Quarterly: Kohorten-LTV-Kurve aktualisieren. SKAN-Signal der ersten 35 Tage gegen 90-Tage-Revenue-Korrelation regressieren.",{"type":32,"tag":123,"props":527,"children":528},{},[529],{"type":37,"value":530},"Budget-Allokation über gewichteten Durchschnitt von SKAN + modeled + incrementality.",{"type":32,"tag":33,"props":532,"children":533},{},[534],{"type":37,"value":535},"Ist dieser Multi-Layer-Ansatz teuer? Ja. Aber wenn iOS 60 % deines Traffic ist und CAC $30+\u002FUser beträgt, kostet ein Attribution-Fehler deutlich mehr.",{"type":32,"tag":40,"props":537,"children":539},{"id":538},"tradeoffs-und-gegenargumente",[540],{"type":37,"value":541},"Tradeoffs und Gegenargumente",{"type":32,"tag":33,"props":543,"children":544},{},[545],{"type":32,"tag":55,"props":546,"children":547},{},[548],{"type":37,"value":549},"„Modeled conversions sind unzuverlässig — warum nutzen wir sie?\"",{"type":32,"tag":33,"props":551,"children":552},{},[553],{"type":37,"value":554},"Weil es keine Alternative gibt. SKAN ist aggregiert, IDFA bei 7 %, gar kein Signal bedeutet vollständige Blindheit. Modeled conversions sind unvollkommen, aber kalibriert. Mit Holdout-Tests reduzierst du Bias und erreichst 75–80 % Accuracy — deutlich besser als nichts.",{"type":32,"tag":33,"props":556,"children":557},{},[558],{"type":32,"tag":55,"props":559,"children":560},{},[561],{"type":37,"value":562},"„Ist SKAdNetwork 4 ausreichend, oder sollten wir auf 5 warten?\"",{"type":32,"tag":33,"props":564,"children":565},{},[566],{"type":37,"value":567},"SKAdNetwork 5 (iOS 18, 2024 angekündigt) verspricht granularere Source ID und längeres Lookback-Fenster, aber volle Adoption fehlt noch. iOS 17 User-Base ist 70 %+, iOS 18 etwa 30 %. Es ist pragmatisch, deinen Stack auf SKAdNetwork 4 zu bauen und Features von 5 inkrementell hinzuzufügen.",{"type":32,"tag":33,"props":569,"children":570},{},[571],{"type":32,"tag":55,"props":572,"children":573},{},[574],{"type":37,"value":575},"„Brauche ich für jede Kampagne einen Incrementality-Test?\"",{"type":32,"tag":33,"props":577,"children":578},{},[579],{"type":37,"value":580},"Nein. Incrementality ist teuer und langsam. Ein quarterly Test pro Kanal reicht (Meta Q1, TikTok Q2, Google Q3). Kleine Kampagnen nutzen modeled + SKAN Blend; bei großen Budget-Bewegungen testen.",{"type":32,"tag":582,"props":583,"children":584},"hr",{},[],{"type":32,"tag":33,"props":586,"children":587},{},[588],{"type":37,"value":589},"iOS-Attribution ist nicht mehr deterministisch, sondern probabilistisch + aggregiert + test-getrieben. Die richtige Abbildung von SKAdNetwork 4s Drei-Postback-Struktur, die Kalibrierung von modeled conversions mit Holdout-Tests und die Projektion von Post-35-Tage-LTV über Kohorten-Projection — das ist der neue Standard 2026. Wenn du deinen Stack auf diese drei Schichten aufbaust — SKAN + modeled + incrementality — fliegst du auf iOS nicht blind, sondern datengestützt.",{"title":16,"searchDepth":591,"depth":591,"links":592},3,[593,595,600,601,602,603],{"id":42,"depth":594,"text":45},2,{"id":89,"depth":594,"text":92,"children":596},[597,598,599],{"id":101,"depth":591,"text":104},{"id":160,"depth":591,"text":163},{"id":176,"depth":591,"text":179},{"id":198,"depth":594,"text":201},{"id":271,"depth":594,"text":274},{"id":333,"depth":594,"text":336},{"id":538,"depth":594,"text":541},"markdown","content:de:marketing:ios-17-nach-attribution-stack.md","content","de\u002Fmarketing\u002Fios-17-nach-attribution-stack.md","de\u002Fmarketing\u002Fios-17-nach-attribution-stack","md",1782050759984]