ChatGPT answers, unedited

If you're trying to answer “which social campaigns actually caused conversions across both our website and mobile app?”, I’d use a mobile measurement/attribution platform as the attribution layer, with GA4 as the behavioral analytics layer.

My recommendation

Best fit: AppsFlyer if mobile app conversions are strategically important and you need web → app → web journeys.

AppsFlyer now supports web attribution alongside mobile attribution, including connecting web activity to mobile conversions and calculating cross-platform ROAS. It can also pass conversion signals back to ad networks. AppsFlyer AppsFlyer Support

A typical setup would be:

Social ad → website → signup/purchase → app install → in-app purchase

with AppsFlyer maintaining the campaign/source context across that journey. Its current cross-platform user-based attribution can also use a persistent customer ID to connect activity across platforms when users are authenticated. AppsFlyer Support

I'd pair that with analytics.google.com for product/user behavior. GA4 can put web and iOS/Android app streams in one property and provides cross-platform attribution/reporting. Google Support Google Support

What I'd use for each job

For iOS especially, don't expect deterministic user-level attribution everywhere: ATT consent, SKAdNetwork, modeled attribution, and each ad network's own reporting all create differences. AppsFlyer's current attribution documentation explicitly distinguishes device-ID matching, SKAdNetwork, probabilistic modeling, and self-reporting-network attribution. AppsFlyer Support

One important caveat

Don't make Meta/TikTok/Google's native dashboards your source of truth. Each platform has its own attribution methodology, so they can all claim credit for the same conversion.

Instead, establish a neutral definition such as:

One conversion ID + one customer ID + standardized campaign taxonomy + defined attribution window/model

Then compare platform-reported numbers separately from your independent attribution number.

If you have a meaningful percentage of users who start on web and later convert in the app, I'd lean strongly toward AppsFlyer + GA4 + a warehouse, rather than GA4 alone. AppsFlyer specifically supports web-to-app measurement and web attribution, while GA4 is better positioned for understanding what users actually did across your properties. AppsFlyer Google Support

If you tell me your stack (e.g. Meta + TikTok + Google, iOS/Android, Shopify/web app, Segment/Snowflake/BigQuery), I can recommend the exact attribution architecture and event flow.

If you’re trying to answer “which social campaign actually caused the conversion?” across both your website and mobile app, I’d use a mobile measurement partner (MMP) + web analytics + a consistent campaign taxonomy, rather than relying on ad-platform reporting alone.

Recommended setup

• Web: Use GA4 (or your existing web analytics platform) with UTM parameters on every social campaign.
• Mobile app: Use an MMP such as AppsFlyer, Adjust, or Branch for install and post-install attribution.
• Cross-channel identity: If users can move between web and app, establish a first-party user/account ID so you can connect journeys such as social ad → website → app install → purchase.
• Campaign metadata: Standardize campaign, source, medium, creative, audience, and placement identifiers across web and app. This is crucial if you want to compare Meta vs. TikTok vs. LinkedIn, etc.
• Reporting: Send the resulting events/attribution data into your warehouse (e.g., BigQuery/Snowflake) and calculate conversions there rather than trying to reconcile multiple platform dashboards.

The important distinction

There are really two attribution problems:

1. “Where did the conversion come from?”
Use deterministic first-party identifiers where possible, plus UTMs/web analytics and MMP attribution.

2. “Did the social campaign cause incremental conversions?”
Attribution alone doesn't answer this reliably. For budget decisions, supplement it with incrementality experiments—geo holdouts, conversion lift tests, or randomized audience experiments.

For example:

Meta ad → landing page → signup → app install → purchase

You want a system where the purchase can ultimately be associated with the campaign/creative that initiated the journey, while your incrementality testing tells you whether Meta generated that purchase or would have happened anyway.

What I'd choose

If you're building this from scratch, a solid stack is:

Social platforms → MMP + GA4 → first-party user ID → data warehouse → BI

If app growth is a major part of your business, I'd lean toward an MMP like AppsFlyer/Adjust/Branch. If web is dominant and the app is secondary, you can start with GA4 + first-party IDs and add an MMP when app attribution becomes material.

One caveat: iOS privacy restrictions (ATT/SKAN) mean you shouldn't expect perfect person-level attribution from paid social on iOS. Your measurement architecture should explicitly accommodate modeled/aggregated attribution and incrementality testing.

For cross-channel social conversion attribution, I’d use a mobile measurement partner (MMP) + web analytics + a common attribution/warehouse layer, rather than relying on ad-platform reporting alone.

A solid setup looks like:

1. Web: GA4 or your web analytics stack, with UTMs and first-party IDs.
2. Mobile app: An MMP such as AppsFlyer, Adjust, or Branch.
3. Identity: Connect web and app activity using a consented, first-party customer/user ID when someone logs in or otherwise identifies themselves. Don't try to deterministically stitch anonymous web and app users without an appropriate basis.
4. Campaign tracking: Standardize campaign IDs/UTMs across Meta, TikTok, Google, LinkedIn, etc., so the same campaign taxonomy survives from click → web → install → app conversion.
5. Central attribution: Export event-level data into your warehouse and calculate a consistent attribution model there. This is especially useful when you want to compare web conversions + app conversions + downstream revenue.
6. Platform reporting: Keep Meta/Google/etc. attribution as a diagnostic, but don't treat each platform's reported conversions as directly additive—they use different windows, identities, models, and sometimes modeled conversions.

Which attribution model?

If you're trying to answer “Which social campaigns actually drove incremental business?”, I'd go beyond last-click.

• Starting point: position-based or data-driven multi-touch attribution.
• Better for budget allocation: incrementality testing / geo experiments / holdouts.
• For app installs: let the MMP handle install attribution and post-install events.
• For web + app journeys: use your first-party identity/warehouse layer as the source of truth for the unified customer journey.

A practical architecture is:

Ad platforms → UTM/campaign IDs → Web analytics + MMP → first-party identity → warehouse → unified conversion/revenue reporting → incrementality tests

If you tell me which social platforms you're running (e.g. Meta/TikTok/LinkedIn), your web/app stack, and whether users log in across both, I can recommend the specific tooling and attribution architecture I'd use.

For attributing conversions from social campaigns across web + mobile app, you usually need a cross-platform attribution stack, not just ad-platform reporting. The right choice depends on your scale, channels, and whether you need incrementality vs. last-touch attribution.

Common options

A typical modern setup

1. Use a mobile measurement partner (MMP)
   • Install an MMP SDK in your app.
   • It attributes installs and app events (purchase, signup, subscription, etc.) back to social campaigns.
   • It handles privacy changes such as Apple's App Tracking Transparency framework and aggregated attribution.

2. Use web analytics + CRM identity stitching
   • Track web sessions and conversions through GA4, Adobe, or your own analytics stack.
   • Pass campaign parameters (UTMs, click IDs, deep links).
   • Tie anonymous web users to known customers after login/signup.

3. Create a unified attribution layer
   • Send app events, web conversions, ad spend, and CRM revenue into a warehouse (e.g., BigQuery, Snowflake, Databricks).
   • Build reporting around:
     • first touch
     • last touch
     • multi-touch
     • customer lifetime value
     • incremental lift

Attribution models to consider

• Last-click attribution: simple, but often over-credits retargeting.
• Data-driven attribution: better for mature advertisers with enough conversion volume.
• Position-based attribution: useful when you care about both acquisition and conversion assists.
• Incrementality testing: best for answering “did social actually cause these conversions?”

My usual recommendation by company stage

• Startup / growing app: AppsFlyer or Adjust + GA4 + CRM integration.
• E-commerce with web + app: Branch or GA4 + a warehouse-based attribution model.
• Large enterprise spending heavily on paid social: MMP + CDP/warehouse + media mix modeling or incrementality experiments.

A key question: are you trying to optimize ad spend inside Meta/TikTok/etc., or are you trying to measure true business impact across the whole customer journey? The tooling choice changes quite a bit.

For attributing conversions from social campaigns across web + mobile app, I’d use a cross-channel measurement stack, not rely on platform-reported conversions alone.

Recommended setup

1. First-party event collection
   • Define a canonical conversion event (e.g. purchase, signup, subscription).
   • Send the same event schema from both web and app.
   • Use a stable first-party customer/user ID where you have consent.

2. Campaign tracking
   • Web: UTMs + click IDs (e.g. Meta/Google/TikTok identifiers).
   • App: the relevant mobile attribution/deep-link mechanism.
   • Preserve campaign metadata through the journey so a web click → app install/open → conversion can be connected where permitted.

3. Mobile measurement
   • Use an MMP (mobile measurement partner) such as AppsFlyer, Adjust, or Branch for app installs and post-install conversions.
   • This is particularly important for iOS because deterministic user-level tracking is constrained by Apple's privacy framework.

4. Unified analytics / warehouse
   • Send web, app, ad-platform, and conversion events into a common analytics layer/warehouse.
   • Ideally model attribution yourself so that web and app conversions are evaluated using the same attribution rules rather than comparing incompatible platform numbers.

5. Attribution model
   • For reporting: start with data-driven or position-based attribution rather than giving every platform last-click credit.
   • For budget decisions: supplement attribution with incrementality testing (geo experiments, holdouts, conversion lift tests). This helps distinguish correlation from conversions actually caused by the social campaign.

A practical architecture

Social ad → click/view → web or app → first-party events → warehouse → attribution model → reporting

I'd keep three metrics separate:

If you're choosing one thing to implement first: use an MMP for app attribution + first-party/warehouse-based attribution for web and unified reporting, then add incrementality testing once you have enough volume.

If you tell me whether you're running Meta, TikTok, Google, etc., and whether your app is iOS, Android, or both, I can recommend a concrete setup and attribution model.

For social campaigns driving conversions across web + mobile app, I’d use a hybrid attribution setup rather than relying on UTMs or platform-reported conversions alone.

Recommended stack

1. UTMs + click IDs for acquisition
   • Use UTMs consistently: utm_source, utm_medium, utm_campaign, utm_content.
   • Preserve platform click IDs where available (e.g. Meta click ID, Google click ID).
   • Store these identifiers in your first-party session/user profile so they survive the web → signup/login → conversion journey.

2. A cross-platform analytics layer
   • Google Analytics / Firebase is a reasonable baseline if you're already in the Google ecosystem.
   • Amplitude or Mixpanel can be better if you care heavily about user-level journeys and product conversion behavior.
   • The key is having a common user ID connecting web and app events after authentication.

3. Mobile attribution / deep linking
   • For app installs and re-engagement, use an MMP such as AppsFlyer, Adjust, or Branch.
   • This is particularly important because app attribution has platform privacy constraints that ordinary web analytics can't solve.

4. Server-side conversion events
   • Send your canonical conversion events from your backend/warehouse rather than depending exclusively on browser/app SDK events.
   • Examples: purchase, subscription_started, qualified_lead.
   • Then feed those conversions back to ad platforms through their server-side APIs/conversion APIs where appropriate.

The attribution model I'd use

For reporting, maintain two views:

Operational attribution:
Use a deterministic first-party identity when possible:

ad click → web session → user ID → app → conversion

Marketing attribution:
Use a model such as data-driven/multi-touch attribution for understanding the contribution of multiple campaigns, while keeping last non-direct touch as a simple comparison metric.

Don't try to force Meta/Google's reported conversions, GA4 conversions, and your backend revenue into one number—they use different attribution windows, identities, and methodologies.

A practical architecture

Social platforms
   ↓
UTMs / click IDs / deep links
   ↓
Web ───────────────→ App
 ↓                     ↓
First-party user ID ←──┘
          ↓
   Analytics / MMP
          ↓
     Data warehouse
          ↓
 Canonical conversion
          ↓
 Attribution + reporting
          ↓
 Ad-platform conversion APIs

If you're starting from scratch: I'd typically choose GA4/Firebase + an MMP (AppsFlyer/Adjust/Branch) + a warehouse + first-party user IDs. If sophisticated product/funnel analysis is important, add Amplitude or Mixpanel.

The most important design decision isn't the vendor, though: establish a canonical conversion event and a durable identity that can connect web and app behavior. Without that, cross-channel attribution will remain fragmented.

For cross-channel social campaigns, I’d use a unified, event-level attribution setup rather than relying on platform-reported conversions.

A solid stack looks like:

1. Capture campaign identity at acquisition
   • Use consistent UTMs: utm_source, utm_medium, utm_campaign, utm_content.
   • For paid social, also preserve each platform’s click/campaign/ad identifiers where available.
   • Persist these identifiers through signup/login and into your customer/user record.

2. Use a shared first-party identity
   • Web: anonymous ID → logged-in user ID.
   • App: app instance/device ID → logged-in user ID.
   • Once someone authenticates, stitch web and app activity to the same customer/user ID.
   • This is much more useful than trying to attribute purely by device.

3. Send conversions to a central event layer
   • Record events such as signup, trial_started, purchase, etc. with:
     • user ID
     • timestamp
     • conversion value
     • platform/channel
     • campaign metadata
     • acquisition/attribution IDs
   • A CDP or warehouse can serve as the source of truth.

4. Use an attribution model appropriate to the question
   • Last-touch: simple, useful for operational reporting, but tends to over-credit retargeting.
   • First-touch: useful for measuring acquisition.
   • Position-based / linear: reasonable descriptive models.
   • Incrementality/experiments: best if you want to know whether social actually caused the conversion.
   • For serious budget allocation, I’d prioritize incrementality testing over increasingly complicated deterministic attribution.

5. Handle web + app specifically
   • On web, first-party cookies and server-side events can preserve attribution.
   • On mobile, use the platform's privacy-preserving attribution mechanisms where required (e.g. Apple's SKAdNetwork) alongside your own authenticated-user events.
   • Don't expect a single device-level identifier to reliably follow a person across web and app.

What I'd recommend

If you're building this today, make your data warehouse/user ID the attribution spine:

Social ad → landing/app → persistent campaign metadata → user ID → conversion event → warehouse → attribution model

Then have Meta/TikTok/etc. platform reporting as diagnostic/optimization data, not the canonical source for total conversions.

If you tell me whether you're using GA4, Segment, RudderStack, Amplitude, Mixpanel, AppsFlyer, Adjust, or a warehouse like Snowflake/BigQuery, I can recommend a concrete implementation and attribution schema.

For attributing conversions from social campaigns across web + mobile app, you typically want a cross-platform attribution setup rather than relying only on ad-platform reporting. The right choice depends on your scale and how much control you need.

Common options

A typical recommended architecture

1. Use a mobile measurement partner (MMP) for paid social app attribution

• AppsFlyer or Adjust are common choices.
• They handle:
  • app installs
  • re-engagement campaigns
  • Meta/TikTok/Snap/Google Ads integrations
  • iOS privacy constraints (SKAdNetwork)
  • fraud filtering

2. Use a web analytics layer for website conversions

• GA4, Adobe Analytics, or another analytics platform.
• Ensure every social campaign has consistent UTMs:
  • utm_source
  • utm_medium
  • utm_campaign
  • utm_content

3. Connect identity across web and app
To understand journeys like:

Instagram ad → website → account creation → app install → purchase

you need:

• first-party user IDs
• login/account stitching
• deep linking (often via Branch or similar tools)
• a customer data platform (CDP) if you have many touchpoints

Attribution model choices

Avoid relying only on last-click. Consider:

• Data-driven attribution — usually best if you have enough conversion volume.
• Position-based attribution — useful if you value first touch + conversion touch.
• Incrementality testing — best for answering “did this campaign actually create additional conversions?”
• Media mix modeling — useful at larger spend levels.

If you are running mostly paid social

A practical stack:

• Small/medium app business: GA4 + AppsFlyer + platform pixels/SDKs
• Growing consumer app: AppsFlyer/Adjust + Branch + data warehouse
• Enterprise: MMP + Adobe/Amplitude + CDP + incrementality testing

One important caveat: Meta, TikTok, and other platforms will usually report more conversions than a neutral measurement system because they optimize and attribute within their own ecosystems. For budget decisions, use a third-party measurement source as your source of truth.

If you share your channels (e.g., Meta/TikTok/Google/Snap), app platform (iOS/Android), monthly ad spend, and whether users log in, I can suggest a more specific stack.

If you’re trying to answer “which social campaigns actually drive conversions across both web and mobile app?”, I’d use a mobile measurement partner (MMP) + a unified analytics/warehouse layer, rather than relying on each social platform’s reporting.

My recommendation

Best general architecture:

Social ads → MMP → web + app events → data warehouse → BI

For example, AppsFlyer is particularly interesting now because it has expanded from mobile attribution into web attribution, with the goal of providing one attribution framework across web and mobile. www.appsflyer.com

Your stack could look like:

• AppsFlyer — attribution and campaign measurement across web + app
• GA4 — behavioral/product analytics and cross-platform journeys
• BigQuery/Snowflake — your canonical conversion and customer dataset
• Looker/Tableau/Mode/etc. — reporting
• Social platforms — optimization/activation, but not your source of truth

If you want a Google-centric stack

GA4 + Firebase is a very good option, particularly if Google Ads is a major acquisition channel. GA4 can put your web, iOS, and Android streams in one property and provide cross-platform measurement. support.google.com

The important piece is User-ID: when someone logs in, use your own stable customer/user identifier across web and app. GA4 can then connect activity across devices and platforms rather than treating the web browser and app installation as unrelated users. support.google.com

For social specifically

I'd structure attribution around your own conversion events, not whatever each network calls a conversion.

For example:

campaign_id
  ↓
social ad click
  ↓
landing page
  ↓
signup
  ↓
app install
  ↓
login
  ↓
purchase/subscription

Persist the campaign/attribution information through that journey and associate it with your first-party customer ID once the person becomes identifiable.

For iOS, don't expect deterministic user-level attribution for everyone. SKAdNetwork provides privacy-preserving attribution without requiring IDFA, and tools such as AppsFlyer can incorporate SKAN postbacks alongside other attribution methods. support.appsflyer.com

What I would not do

I wouldn't make Meta/Instagram, TikTok, or another social platform's own conversion reporting your canonical attribution. Their numbers are useful for campaign optimization, but each platform has its own identity, attribution windows, modeling, and view-through methodology.

Instead, establish a single definition such as:

Primary attribution: first-party/MMP attribution with a defined lookback window and documented rules.

Then separately report:

• Attributed conversions — your canonical number
• Platform-reported conversions — Meta/TikTok/etc.
• Incremental conversions — from experiments/holdouts when you can run them

That last category is especially important: attribution tells you who gets credit; incrementality tells you whether the campaign actually caused the conversion.

If you're choosing a tool today: I'd lean AppsFlyer if web + app + lots of paid social is central to the business, and GA4/Firebase if you're primarily in the Google ecosystem and want a lower-complexity analytics solution. GA4 itself even recommends using an approved app attribution partner as a complementary measurement solution for app attribution. support.google.com

If you tell me which social channels you use (Meta, TikTok, Snapchat, Reddit, LinkedIn, etc.) and whether the app is iOS, Android, or both, I can recommend a concrete attribution architecture and event/ID scheme.

If you’re trying to attribute social campaigns across both your website and mobile app, I’d use a mobile measurement partner (MMP) as the attribution layer, with GA4 as the broader analytics/reporting layer.

My default recommendation: AppsFlyer or Adjust

AppsFlyer is particularly strong if your problem is web → app → conversion. Its OneLink supports a single cross-platform link, deep linking, campaign parameters, and app/web routing. AppsFlyer also added web attribution capabilities in 2026. support.appsflyer.com

Adjust is another strong choice. It supports web attribution alongside app measurement and specifically provides web-to-app flows that preserve campaign information as users move from browser → app store → app. www.help.adjust.com

I’d choose between them roughly like this:

Recommended architecture

Think of the stack as:

Social ad → tracking link → website/app → conversion event → MMP → warehouse/BI

For example:

Instagram ad → AppsFlyer/Adjust link → mobile website → "Open in App" → app → purchase

The attribution system should preserve things like:

• source: Meta / TikTok / Snapchat
• campaign
• ad set
• creative
• click/impression timestamp
• web session
• app install
• app re-engagement
• purchase/revenue

Then send the resulting conversion data into your warehouse/BI system.

Where GA4 fits

I'd still implement GA4. Google explicitly supports a single GA4 property containing web, iOS, and Android streams, giving you a unified event-based view of web/app behavior. support.google.com

GA4 is excellent for answering:

"What did users do across web and app?"

An MMP is better for:

"Which paid social campaign actually gets credit for this app install/re-engagement?"

GA4 also has cross-channel attribution and data-driven attribution capabilities, but there are platform-specific differences—particularly around app conversion reporting—so I wouldn't make GA4 alone your source of truth for paid mobile acquisition. support.google.com

One important caveat

Don't expect one perfectly deterministic attribution number across Meta/TikTok/etc. and web/iOS/Android. iOS privacy restrictions, SKAdNetwork, consent, browser limitations, and self-attributing networks mean the platforms can legitimately report different numbers. Adjust, for example, uses deterministic matching where available and probabilistic/SKAN approaches in situations where deterministic attribution isn't possible. support.appsflyer.comsupport.google.comsupport.google.comwww.help.adjust.com

If I were designing this from scratch:
AppsFlyer/Adjust → attribution + deep linking → GA4 → BigQuery/warehouse → BI dashboard.

If you tell me which social platforms you use (Meta, TikTok, Snapchat, LinkedIn, etc.), your conversion (purchase/signup/lead), and whether users commonly go web → app, I can recommend the specific setup and attribution model I'd use.

If your goal is one attribution view for social campaigns that drive conversions on both web and mobile app, I’d use a central analytics layer + mobile measurement partner (MMP) rather than relying on each social platform’s reporting.

My recommended setup

1. GA4 as the cross-platform analytics layer

Use one GA4 property with:

• Web data stream
• iOS app data stream
• Android app data stream

GA4 is designed to provide a unified view of web + app behavior and supports cross-platform attribution. support.google.com

For reporting, I'd generally use data-driven attribution rather than last-click. GA4's data-driven model distributes credit based on observed conversion-path data. support.google.com

2. An MMP for app attribution

For paid social driving app installs and in-app purchases, use an MMP such as:

• AppsFlyer
• Adjust
• Branch

The MMP should be the authoritative source for app install/re-engagement attribution, while GA4 provides the broader web + app customer journey.

3. Consistent campaign IDs across everything

Your social links should carry standardized parameters such as:

utm_source=meta
utm_medium=paid_social
utm_campaign=summer_sale
utm_content=video_01

For app campaigns, use the MMP's tracking/deep-link URLs so you can preserve campaign context through ad → app store → install → first open → purchase.

GA4's traffic-source dimensions are the underlying inputs for attribution, so consistent campaign tagging is important. support.google.com

The architecture I'd use

Social ad → tracking/deep link → web OR app → conversion

Then:

MMP → app install / re-engagement attribution
GA4 → unified web + app behavior and cross-channel analysis
Ad platforms → optimization/reporting within each platform
Warehouse/BI → your final business-level attribution and ROAS reporting

The important distinction is that you shouldn't simply add Meta + TikTok + GA4 + MMP conversions together. Each platform can claim the same conversion. Pick a source of truth and reconcile the others against it.

If you want one source of truth

For a relatively straightforward setup, I'd choose:

GA4 + an MMP + your data warehouse

Use GA4 for the marketer-facing cross-platform view, the MMP for app-specific attribution, and the warehouse for deduplication, LTV/ROAS, and more sophisticated attribution.

If you're heavily dependent on Meta/TikTok paid social, I'd also implement their server-side conversion APIs so downstream conversions aren't dependent solely on browser/app tracking.

One caveat: iOS privacy restrictions mean web-to-app identity matching will never be perfectly deterministic. GA4 itself uses modeling where conversions can't be directly observed, including situations involving cross-device behavior. support.google.com

If you tell me which social platforms you use (e.g. Meta, TikTok, Snapchat, LinkedIn) and whether your conversion is purchase, lead, subscription, etc., I can recommend the specific attribution stack and event architecture I'd use.

If you mean “what measurement stack should I use to determine which social campaigns drove conversions when users move between web and iOS/Android?”, I’d use a mobile measurement partner (MMP) + GA4 + your warehouse, rather than relying on Meta/TikTok/GA4 alone.

My recommended setup

1. Use an MMP as the independent attribution layer

• AppsFlyer — probably my first choice if web + app attribution is a major requirement. As of 2026, it supports web performance measurement alongside mobile attribution, with a unified measurement framework across web and mobile. www.appsflyer.comwww.appsflyer.com
• Adjust — strong alternative, particularly if your organization already uses it for mobile.
• Branch — particularly attractive if deep linking and web→app journeys are important.
• Singular/Kochava — worth considering depending on your existing stack and reporting requirements.

This is especially important for iOS because mobile attribution has constraints around ATT and SKAdNetwork that a generic web analytics platform doesn't fully solve. www.appsflyer.comwww.appsflyer.com

2. Use GA4 as the behavioral/customer-journey layer

Put your web stream + iOS stream + Android stream in the same GA4 property. GA4 is designed to analyze web and app behavior together. support.google.com

If users log in, send the same first-party User-ID from web and app. That lets GA4 connect activity from the same person across sessions, devices and platforms. support.google.com

For example:

TikTok ad → mobile web landing page → login → app install → app purchase

You want your data model to understand that as one customer journey, rather than four unrelated events.

3. Put the raw attribution/event data in your warehouse

I'd ultimately make BigQuery/Snowflake/Databricks the place where you build your canonical marketing dataset.

Something roughly like:

Then you can calculate attribution independently of whatever each ad platform claims.

For the actual attribution model

I'd use two layers, rather than trying to find one magical model:

Operational attribution:
Use the MMP's attribution for app acquisition/install and network optimization.

Strategic attribution:
Use your warehouse to calculate incrementality + multi-touch/path analysis across web and app.

For GA4 specifically, data-driven attribution is preferable to simply using last click when you're evaluating the contribution of multiple touchpoints. GA4's data-driven model considers converting and non-converting paths and distributes credit based on the estimated contribution of interactions. support.google.com

But I wouldn't treat multi-touch attribution as proof that a campaign caused the conversion. For budget decisions, holdout tests/incrementality experiments are much stronger.

The architecture I'd choose

Meta / TikTok / Snap / Reddit / etc.
              │
              ▼
       MMP + tracking links
              │
       ┌──────┴──────┐
       ▼             ▼
     Web           iOS/Android
       │             │
       └──────┬──────┘
              ▼
             GA4
              │
              ▼
       Data warehouse
              │
              ▼
   Canonical attribution model
              │
       ┌──────┴──────┐
       ▼             ▼
   BI/reporting   Experimentation

One important distinction: GA4 can give you a unified web/app view, but it isn't a substitute for an MMP for mobile ad attribution. For example, TikTok explicitly supports MMPs such as AppsFlyer, Adjust, Branch, Kochava and Singular for app conversion measurement. ads.tiktok.com

If I were starting from scratch today:
AppsFlyer + GA4 + BigQuery + first-party User-ID + incrementality testing would be my default stack.

If you tell me which social channels you use (e.g. Meta/TikTok/Snap), whether the conversion is purchase/lead/subscription, and whether users commonly go web → app, I can recommend the exact attribution architecture and event schema.

For attributing conversions from social campaigns across both web and mobile app, I’d use a single cross-channel measurement framework rather than relying on platform-reported conversions.

Recommended setup

1. Use a first-party event/analytics layer as the source of truth

• Web: capture campaign parameters (utm_source, utm_medium, utm_campaign, etc.) and persist them through conversion.
• App: capture the same campaign taxonomy through your mobile attribution/analytics SDK.
• Send web and app conversion events into one analytics/warehouse layer with a consistent user/customer ID where privacy rules permit.

2. Use an MMP for mobile attribution
For app campaigns, a mobile measurement partner (MMP) such as AppsFlyer, Adjust, or Singular is generally the right layer for install and post-install attribution. It handles mobile-specific attribution mechanisms that ordinary web analytics can't.

3. Use a warehouse/CDP for cross-device and cross-channel reporting
Put web events, app events, ad spend, and attribution outputs into something like your data warehouse. Then build reporting around:

• campaign
• source/platform
• web vs. app
• acquisition date
• conversion event
• revenue
• customer/user ID where available

This lets you answer questions like “How much incremental revenue did our paid social campaigns generate across both web and app?”, rather than comparing incompatible platform dashboards.

Attribution model

For operational reporting, I'd start with:

Primary: data-driven / multi-touch attribution where you have sufficient volume
Secondary: last non-direct touch for simple campaign reporting
Incrementality: geo/holdout experiments for determining whether social actually caused the conversion

Don't treat Meta/TikTok/etc. reported conversions as additive—they'll often claim overlapping conversions.

A practical architecture

Social ads → tracking links / platform SDK → Web + App events → MMP + analytics → Data warehouse → unified attribution model → BI

One important distinction: attribution and incrementality aren't the same thing. Attribution answers “which touchpoint gets credit?”; incrementality answers “would this conversion have happened without the campaign?” For budget allocation, I'd ultimately optimize against the latter.

If you tell me your stack (e.g. GA4 + Firebase, Segment, Snowflake/BigQuery, AppsFlyer/Adjust, Meta/TikTok), I can recommend the specific setup and event schema.

If you’re trying to answer “which social campaigns actually drove conversions/revenue across both my website and mobile app?”, I’d use a dedicated cross-platform attribution layer, rather than relying solely on Meta/TikTok/Google’s native reporting or GA4.

My recommendation

Best fit: AppsFlyer, particularly if mobile app conversions are important.

AppsFlyer now supports both web attribution and mobile attribution, with a common measurement layer intended to connect web and mobile customer journeys. Its 2026 Web Performance Measurement product specifically positions itself as an independent source of truth across web and mobile. support.appsflyer.com

The architecture I'd use is:

Social ad → attribution/deep link → web or app → conversion → AppsFlyer → warehouse/BI

For example:

• Facebook/Instagram campaign → website purchase → attributed to campaign
• Facebook/Instagram campaign → app install → signup → purchase → attributed to campaign
• Social ad → mobile web → “open in app” → purchase in app → retain the original campaign attribution
• Same customer later converts on another device → use a persistent customer ID where you have authenticated identity

AppsFlyer's OneLink supports a single link across platforms and deep linking into iOS/Android apps. support.appsflyer.com Its web-to-app tooling is specifically designed to preserve attribution when someone goes from an ad → mobile website → app store → app. support.appsflyer.com

Where GA4 fits

I'd still use GA4/Firebase alongside it, rather than replacing it.

GA4/Firebase is excellent for behavioral analytics and funnel analysis across web, iOS and Android, and it can report campaign information and events across those environments. firebase.google.com

But I wouldn't make GA4 your sole independent paid-social attribution authority, especially if you're spending significant money on mobile acquisition. Mobile attribution has issues such as ATT/SKAN and identity gaps that generic web analytics doesn't fully solve. www.appsflyer.com

The important part: define your attribution model

I'd establish two views, rather than trying to make one model answer everything:

And I'd pass a stable customer ID/CUID once the person is authenticated. AppsFlyer's cross-platform user-based attribution can use a persistent CUID to connect activity across platforms. support.appsflyer.com

That is particularly valuable if your journey looks like:

Instagram → web → signup → iPhone app → purchase → desktop web renewal

Without an identity layer, those can easily look like multiple unrelated users/conversions.

What I would implement

1. AppsFlyer: attribution + deep linking + mobile/web campaign measurement
2. GA4/Firebase: behavioral analytics and product funnels
3. Your data warehouse: canonical customer, campaign, and revenue tables
4. BI: campaign-level CAC, conversion rate, ROAS, LTV
5. Native Meta/TikTok/etc. reporting: optimization diagnostics, not your source of truth

For social campaigns, make sure every outbound web link carries consistent UTM/source/campaign/ad/adset IDs, while app-bound traffic goes through the appropriate attribution/deep-link mechanism.

If you're mostly web and have a relatively small app, the answer could be different. If you tell me your stack (e.g. Meta + TikTok + Google, Shopify/custom web, iOS/Android, Segment/RudderStack, Snowflake/BigQuery, etc.), I can recommend the specific attribution architecture and event/ID schema I'd use.