Your Meta ads dashboard is lying to you. And it's not Meta's fault.

Over the past 18 months, attribution accuracy for Facebook and Instagram ads has deteriorated by 40-60%. The causes? iOS privacy updates that block tracking, browser restrictions that kill pixels, and ad blockers that strip conversion data before it reaches Meta's servers.

Then came January 12, 2026. Meta deprecated two critical attribution windows: 7-day view and 28-day view. Overnight, your reported conversions dropped 15-30%. Not because performance tanked. Because measurement got worse.

Here's the brutal reality: If you're still running Meta ads with pixel-only tracking in 2026, you're making budget decisions based on incomplete data. The gap between what's actually happening and what Meta reports is widening every week.

This guide breaks down the three major attribution changes that hit Meta ads in 2025-2026, why challenger brands are getting hit hardest, and three practical fixes you can implement this week to restore measurement accuracy.

No fluff. Just the official data and actionable frameworks.

The 3 Big Attribution Changes That Hit Meta Ads in 2025-2026

Change #1: iOS Privacy Updates Broke Cross-Device Tracking

Apple's been tightening privacy restrictions for years. But the 2024-2025 updates pushed Meta ads tracking past the breaking point.

April 2021: iOS 14.5 introduced App Tracking Transparency (ATT). Users got a pop-up asking permission to track them across apps and websites. Most users tapped "Ask App Not to Track." According to industry consensus tracked by Triple Whale, the opt-out rate exceeded expectations. Meta lost visibility into most iOS users' behavior overnight.

September 2024: iOS 17 added Link Tracking Protection. This feature strips tracking parameters (including fbclid, the identifier Meta uses to connect ad clicks to conversions) from URLs when users browse in Safari Private Browsing mode.

September 2025: iOS 18 expanded Link Tracking Protection to more browsing contexts. Now even non-private browsing strips fbclid and UTM parameters in certain scenarios. According to BrightVessel and WorkMagic analysis, this change further degraded Meta's ability to track iOS users.

November 2025: iOS 26 beta analysis from Measured confirmed what performance marketers were seeing in their dashboards: "Meta attribution increasingly unreliable for iOS users."

The impact? iOS users represent a "significant portion of traffic for many brands" (Measured, November 2025), but Meta can't accurately track their conversions anymore. Your iPhone users are clicking ads, converting, and Meta's dashboard shows nothing.

Why this matters: If 40-50% of your website traffic comes from iOS devices (common for B2C brands), you're flying blind on half your audience. The algorithm can't optimize what it can't see.

Change #2: Meta Deprecated Key Attribution Windows (January 2026)

On January 12, 2026, Meta removed two attribution window options from Ads Manager: 7-day view and 28-day view.

According to Meta for Developers (October 16, 2025 announcement), this change affects:

• API access to historical data (limited to 13 months for certain breakdowns)

• Comparison of current performance to historical benchmarks

• Long-attribution window strategies for B2B or high-consideration purchases

What attribution windows actually mean:

An attribution window defines how long after someone sees or clicks your ad Meta will credit that ad with a conversion.

• 7-day click, 1-day view (current default): If someone clicks your ad and converts within 7 days, Meta counts it. If they only see your ad (no click) and convert within 1 day, Meta counts it.

• 28-day click (deprecated): Meta used to count conversions up to 28 days after a click. This captured longer purchase consideration cycles.

• 7-day view, 28-day view (deprecated January 2026): These longer view windows are no longer available.

The practical impact: Your reported conversions just dropped. Not because fewer people are converting. Because Meta stopped counting conversions that happen beyond the 7-day click window.

For B2B brands with 2-4 week sales cycles, this is brutal. A prospect clicks your ad on Monday, evaluates solutions for 10 days, then converts on Thursday the following week. Meta no longer attributes that conversion to your ad. It shows up as "direct traffic" or "organic" in your analytics.

The data gap: According to industry analysis, moving from 28-day click to 7-day click attribution typically reduces reported conversions by 15-30%, depending on your sales cycle length.

Change #3: Conversion API Went From "Nice to Have" to Required

For years, Meta recommended running Conversion API (CAPI) alongside your pixel. Most advertisers ignored the recommendation. Pixel-only tracking was good enough.

Not anymore.

Here's why pixel-only tracking is broken in 2026:

Ad blockers kill pixel fires. Industry estimates suggest that 25-30% of web users run ad blockers. These tools block Meta Pixel JavaScript from loading. Result? No conversion data sent to Meta.

Browser restrictions block third-party cookies. Safari's Intelligent Tracking Prevention (ITP) and Firefox's Enhanced Tracking Protection limit how long cookies last. Your pixel can't track returning visitors accurately.

iOS opt-outs eliminate mobile Safari tracking. With most iOS users opted out of ATT, pixel tracking on mobile Safari is essentially dead.

Google's third-party cookie deprecation looms. Google keeps delaying the timeline (currently targeting late 2025/early 2026), but the direction is clear: third-party cookies are dying.

The symptoms you're seeing in your account:

According to Triple Whale's analysis of advertiser pain points:

• "Conversions don't match CRM" (your CRM shows 50 sales, Meta reports 32)

• "Pixel fires inconsistently" (works for some users, fails for others)

• Revenue attribution is way off (Meta says you made $10K, Shopify says $15K)

The adoption problem: AdExchanger reported in September 2025 that "CAPI adoption may have plateaued" at mid-market brands. Translation: Many advertisers still haven't implemented server-side tracking, even though it's now essential.

If you're not running Conversion API in 2026, you're losing 40-60% of conversion visibility. And Meta's algorithm is optimizing your campaigns based on incomplete data, which means you're wasting budget on audiences and creatives that look like they're not working but actually are.

Why This Matters More for Challenger Brands Than Enterprises

Enterprise brands have resources you don't.

They have dedicated data engineering teams who can build custom attribution models. They have budgets for $50K-$200K/year multi-touch attribution (MTA) platforms like Northbeam, Rockerbox, or Triple Whale's enterprise tier. They run marketing mix modeling (MMM) studies that cost $100K+ and require PhD-level statisticians.

Enterprise brands also have first-party data infrastructure already built: Customer data platforms (CDPs), data warehouses, CRM systems integrated with every marketing channel, and analysts who can stitch together cross-channel customer journeys without relying on Meta's dashboard.

Challenger brands are stuck in the middle.

You're too sophisticated to ignore attribution accuracy. You understand that a 30% measurement error on a $50K/month Meta budget means you could be wasting $15K/month on the wrong audiences or killing campaigns that are actually profitable.

But you're too lean to afford enterprise solutions. You don't have a data engineering team. You can't drop $100K on an attribution platform. Your CMO is also your paid media specialist, content strategist, and the person who updates the website.

The risk: You're making $500K+ annual budget decisions based on incomplete Meta data.

Example scenario: You're a B2B SaaS brand spending $50K/month on Meta ads. Your reported conversions dropped 30% after the January 2026 attribution window changes. Is performance actually down? Or is measurement broken?

Without first-party data infrastructure and CAPI implementation, you can't answer that question. So you either:

1. Cut budget (potentially killing a profitable channel)

2. Keep spending (potentially wasting money on a channel that's not working)

3. Guess (not a strategy)

This is why challenger brands need simple, practical attribution fixes. You can't out-engineer enterprise competitors. You need to implement the 20% of solutions that deliver 80% of the accuracy improvement.

That's what the next section covers.

How to Fix Meta Ads Attribution in 2026: 3 Practical Solutions

Solution #1: Implement Conversion API (CAPI) Alongside Pixel

What it does: Conversion API is server-side tracking. Instead of relying on browser-based pixels (which get blocked by ad blockers, browsers, and iOS restrictions), your server sends conversion data directly to Meta.

How it works: When someone converts on your website, your server fires an API call to Meta with the conversion details: user identifiers (hashed email, phone, IP address), event type (Purchase, Lead, InitiateCheckout), event value, and a unique event ID.

Key benefit: CAPI captures conversions that pixel misses. Ad blockers? Doesn't matter. iOS opt-outs? Bypassed. Browser cookie restrictions? Irrelevant. The data flows from your server to Meta's servers directly.

Setup complexity: Medium. If you're on Shopify, WooCommerce, or another major e-commerce platform, there are partner integrations that handle CAPI setup for you (apps like Elevar, Littledata, or Meta's official Shopify integration). If you have a custom website, you'll need a developer to implement the API calls.

The proof: Jellyfish case study documented Gentle Monster's CAPI implementation and reported that the brand "significantly boosted ROAS" after adding server-side tracking. Triple Whale's Sonar tool (which enriches CAPI data) helped Ampersand recover attribution accuracy and "improved Meta attribution and campaign efficiency" after iOS signal loss.

Meta's official guidance: According to Meta Blueprint training, Meta recommends running CAPI and Pixel together for "optimal full-funnel tracking." This is not "CAPI or Pixel." This is "CAPI plus Pixel."

Why both? Pixel captures client-side events (page views, scroll depth, time on site). CAPI captures server-side conversions (purchases, form submissions, qualified leads). Together, they give Meta the most complete picture of user behavior.

Implementation Checklist:

1. Set up Meta Business Manager (if you haven't already)

2. Install and verify Meta Pixel (check Events Manager to confirm it's firing)

3. Choose your CAPI setup method:

   • Option A: Partner integration (Shopify app, WooCommerce plugin, GTM template)

   • Option B: Manual API implementation (requires developer)

4. Configure event deduplication: Use the same event_id parameter for the same event sent from both Pixel and CAPI. This prevents double-counting the same conversion.

5. Send high-priority events: Purchase, Lead, InitiateCheckout, AddToCart (prioritize bottom-funnel events first)

6. Include first-party identifiers: Send hashed email, phone number, IP address, user agent, and fbp/fbc cookies if available. More identifiers = better matching.

7. Validate in Events Manager: Check Event Match Quality (EMQ) score. Aim for "Good" or "Great" (we'll cover this in Solution #3).

Timeline: If you use a partner integration, setup takes 1-3 hours. If you need custom development, budget 1-2 weeks for implementation and testing.

Cost: Partner integrations range from free (Meta's official apps) to $50-$300/month (third-party tools like Elevar). Custom development costs $2K-$10K depending on complexity.

ROI: If CAPI captures even 20% more conversions than pixel-only tracking, and you're spending $50K/month, that's 10,000 more conversion signals per year for Meta's algorithm to learn from. Better data = better targeting = better performance.

For step-by-step guidance on exporting and analyzing Meta Ads data, see our complete guide.

Solution #2: Adjust Your Attribution Settings for Reality

Most conversions happen within 7 days of a click. That's the reality of modern consumer behavior.

According to Meta's default attribution setting (7-day click, 1-day view), the platform only counts conversions within that window. If someone converts on day 8 or day 15, Meta doesn't attribute it to your ad.

What to do:

Step 1: Verify your current attribution setting in Ads Manager. Go to Campaigns > Select a campaign > Scroll to Attribution Setting. Check what window you're using.

Step 2: Switch to 7-day click, 1-day view if you're not already using it. This is Meta's current default and reflects realistic conversion timelines for most businesses.

Step 3: Use the new "Breakdown by Attribution Setting" feature in Ads Manager (rolled out January 2026). This lets you compare performance across different attribution windows in the same report. Run a test: View your campaign performance with 7-day click vs. 1-day click attribution. How much difference is there?

Advanced move for B2B marketers: If you have a 2-4 week sales cycle (common for high-ticket B2B offers), expect to lose 20-40% of attributed conversions with the 7-day click window. Those conversions are still happening. Meta just isn't counting them anymore.

How to report this internally:

Don't just show Meta's conversion numbers. Show "assisted conversions" in Google Analytics (conversions where Meta played a role but wasn't the last click) and match Meta conversion data to your CRM.

Example internal dashboard:

• Meta Ads Manager: 45 conversions (7-day click attribution)

• Google Analytics: 68 conversions with Meta assisted (multi-touch view)

• CRM matched conversions: 52 conversions from contacts who clicked Meta ads in the past 30 days

This gives leadership a fuller picture. Meta's dashboard is conservative. Your actual impact is higher.

Pro tip: For long sales cycle businesses, consider optimizing for upper-funnel events (Landing Page Views, Lead Form submissions) instead of purchases. Meta can optimize more effectively when it has more conversion volume within the attribution window.

Solution #3: Build First-Party Data Infrastructure

Third-party tracking is dying. First-party data is the future.

What this means in practice:

Capture emails earlier in the funnel. Don't wait until checkout to collect an email address. Use lead magnets (free guides, calculators, templates), gated content, or email capture pop-ups to get emails from prospects before they convert.

Why? Because you can send that hashed email to Meta via CAPI, and Meta can match it to user profiles even when pixels and cookies fail.

Use Meta's Advanced Matching. This feature (built into Meta Pixel and CAPI) sends hashed customer data (email, phone, first name, last name, city, state, zip code) alongside conversion events. Meta uses this data to improve attribution matching.

According to Meta's documentation, Advanced Matching can improve Event Match Quality scores by 10-30%, which translates to better attribution accuracy and campaign performance.

Integrate your CRM with Meta. Use tools like Zapier, Segment, or native CRM integrations to send offline conversion events back to Meta.

Example: A prospect fills out a Meta lead form. Your sales team calls them, qualifies the lead, and closes a $5K deal 3 weeks later. That sale happened because of the Meta ad. But Meta has no idea because the conversion happened offline (phone call, not website).

Solution: Send an offline conversion event to Meta via API with the original lead ID. Now Meta knows that lead form ad generated a $5K sale, and the algorithm can optimize for leads that turn into high-value customers.

Track Event Match Quality (EMQ) in Events Manager. EMQ measures how well Meta can match your conversion events to user profiles. Low EMQ means poor attribution matching.

Meta grades EMQ as:

• Great: 90%+ match rate

• Good: 70-89% match rate

• Fair: 50-69% match rate

• Poor: Below 50% match rate

How to improve EMQ:

1. Send more customer information parameters with every event (email, phone, IP address, user agent, external ID)

2. Hash data correctly (use SHA-256, lowercase, trim whitespace)

3. Include both fbp and fbc cookies if available

4. Send events via CAPI (server-side) with richer data than pixel can capture

Tools that help:

• Triple Whale's Sonar: Enriches CAPI data with first-party identifiers and attribution matching

• Segment: Customer data platform that routes conversion events to Meta with enriched user profiles

• Zapier: No-code integration to send CRM conversions back to Meta

The strategic shift: You're moving from "rely on Meta's tracking" to "own your customer data and share it with Meta." This future-proofs your attribution as third-party cookies die and privacy restrictions tighten.

For more on building AI-powered marketing intelligence that goes beyond platform attribution, see our guide on how DOJO AI helps challenger brands correlate marketing activities to revenue outcomes.

The 2026 Meta Ads Attribution Checklist for Performance Marketers

Copy this checklist and audit your account monthly.

Monthly Maintenance Tasks

• Check Event Match Quality scores in Events Manager. Goal: "Good" or "Great" for all key conversion events.

• Validate Pixel + CAPI deduplication. Compare total conversions to sum of Pixel-only + CAPI-only events. If the sum is higher than total, you're double-counting.

• Compare Ads Manager conversions to CRM data. Spot-check accuracy. Pull 10 recent CRM conversions and verify they appear in Meta's reporting.

• Review attribution setting for each campaign objective. Make sure you're using 7-day click, 1-day view (unless you have a specific reason to use a different window).

• Monitor iOS vs. Android performance splits. If iOS performance is dramatically worse, you have a signal loss problem. Prioritize CAPI implementation.

Quarterly Strategic Reviews

• Test incremental lift with holdout groups. Run a conversion lift study in Meta Ads Manager. This shows true causal impact, not just attributed conversions.

• Audit first-party data capture points on your website. Where can you collect emails earlier? Can you add a lead magnet? Can you gate high-value content?

• Benchmark Meta attribution accuracy against industry standards. If you're seeing 40-50% discrepancies between Meta and CRM, that's on the high end. Investigate CAPI setup and EMQ scores.

• Evaluate need for third-party attribution platform. If you're spending $100K+/month across channels and attribution is critical, consider tools like Triple Whale, Northbeam, or Rockerbox.

Common Mistakes to Avoid

Mistake #1: Running Pixel-Only Tracking in 2026

Why it fails: You're losing 40-60% of conversions to ad blockers, iOS opt-outs, and browser restrictions. Meta's algorithm is optimizing on incomplete data. Your cost per acquisition looks terrible because Meta isn't seeing most of your conversions.

How to spot it: Check Events Manager. If you see "Browser" as the only event source (no "Server" events), you're running pixel-only.

Fix: Implement Conversion API immediately. This is the highest-priority fix on this list.

Mistake #2: Not Deduplicating Pixel + CAPI Events

The problem: You send the same conversion event from both Pixel and CAPI without proper deduplication. Meta counts it twice. Your dashboard shows 100 conversions. You actually got 50.

Why it matters: Inflated conversion numbers make your campaigns look profitable when they're not. You overspend based on fake data.

How to spot it: Go to Events Manager > Overview > Event Source. Compare total events to "Pixel" + "Server" breakdown. If Pixel + Server sum exceeds total, you're double-counting.

Fix: Use the event_id parameter. When you send the same event from Pixel and CAPI, give them identical event IDs. Meta deduplicates automatically. According to Meta's documentation, event ID should be a unique identifier for that specific event (e.g., order ID + timestamp).

Mistake #3: Ignoring Event Match Quality (EMQ)

The problem: You set up CAPI, but you're only sending the bare minimum data (event name and value). No email, no phone, no user identifiers. Meta can't match events to user profiles. Your EMQ score is "Poor."

Why it matters: Low EMQ means poor attribution matching. Meta sees a conversion happened but doesn't know which user or which ad drove it. The algorithm can't optimize.

How to spot it: Events Manager > Data Sources > Select your Pixel > Event Match Quality tab. Check the score for each event type.

Fix: Send hashed customer information parameters with every CAPI event:

• Email (hashed with SHA-256)

• Phone (hashed with SHA-256)

• First name, last name (hashed)

• City, state, zip code, country

• IP address

• User agent

• fbp and fbc cookies (if available from client-side)

More parameters = better matching = higher EMQ = better attribution.

Mistake #4: Treating Attribution Settings as "Set It and Forget It"

The problem: You set your attribution window to 7-day click back in 2022 and never revisited it. Your business has changed (longer sales cycle, new product lines, different customer behavior), but your attribution hasn't adapted.

Why it matters: If your average sales cycle is 14 days and you're using a 7-day attribution window, you're missing half your conversions. If your sales cycle is 2 days and you're using a 7-day window, you're over-attributing and giving credit to ads that didn't drive the conversion.

How to spot it: Pull a CRM report of time-to-conversion (days between first ad click and purchase). Compare to your attribution window. If median time-to-conversion is longer than your attribution window, you have a measurement gap.

Fix: Review attribution settings quarterly. Adjust per campaign objective. Short-cycle, impulse purchases? Use 1-day click. Long-cycle B2B deals? Use 7-day click and supplement with CRM attribution matching.

What's Next: The Future of Meta Ads Attribution

Short-Term (2026-2027)

More iOS restrictions are coming. Apple's privacy-first stance is only getting stronger. Expect iOS 19 and beyond to further limit tracking. Link Tracking Protection will expand to more contexts. Opt-in rates for App Tracking Transparency will stay low.

CAPI will become table stakes. Right now, CAPI adoption is around 50-60% of Meta advertisers (rough industry estimate). By 2027, it will be 80%+. Advertisers who don't implement server-side tracking will be at a massive disadvantage.

Meta's "Incremental Attribution" metric will roll out broadly. Meta has been testing incrementality measurement (conversion lift studies, geo-split tests) as a complement to last-click attribution. According to Meta for Business blog posts in 2025, this metric is becoming more prominent. It measures true causal impact, not just correlation.

Long-Term (2028+)

Privacy Sandbox will evolve. Google's replacement for third-party cookies (the Privacy Sandbox initiative) is still being tested. By 2028, we'll know if Topics API, FLEDGE, and other privacy-preserving technologies actually work for advertisers. Early signs are mixed.

AI-powered probabilistic attribution will replace deterministic tracking. When you can't track users with cookies and pixels, you use AI models to infer attribution based on patterns. Meta is already investing heavily in this. Expect attribution to shift from "we know this person clicked this ad and bought" to "our model estimates that this ad influenced X% of purchases."

Modeled attribution will be the industry standard. Just like how Google Analytics moved from last-click to data-driven attribution models, Meta will rely more on machine learning to estimate impact when direct tracking isn't possible.

Strategic takeaway: Attribution will never be as "perfect" as it was in 2015-2020. Winners will be marketers who combine platform data (Meta, Google) with first-party customer data and business intelligence to make informed decisions despite imperfect measurement.

For more on how AI is reshaping marketing attribution beyond platform dashboards, see our analysis of revenue correlation modeling.

How DOJO AI Helps Challenger Brands Navigate Attribution Chaos

The Attribution Problem for Lean Marketing Teams

You don't have a data engineering team to build custom attribution models. You can't afford $50K/year attribution platforms. But you need accurate data to make $500K+ annual budget decisions.

The gap: Meta's dashboard shows incomplete data. Your CRM shows some conversions. Google Analytics shows others. Shopify shows revenue. LinkedIn Ads has its own conversion numbers. Nothing matches. You have five sources of truth and zero confidence in any of them.

DOJO AI's Approach

Unified data layer. DOJO AI connects Meta, Google Ads, LinkedIn, CRM, and website analytics into one intelligence layer. All your marketing data in one place.

AI-powered correlation analysis. Instead of relying solely on last-click attribution, DOJO AI identifies patterns between marketing activities and revenue outcomes. Which channels truly drive conversions? Which campaigns have delayed impact that platform dashboards miss?

First-party data enrichment. DOJO AI helps you capture and leverage first-party data to improve attribution accuracy across all platforms. Better matching = better measurement.

Performance monitoring with automated alerts. When attribution discrepancies spike (Meta reports 30% fewer conversions than your CRM for 3 days straight), DOJO AI alerts you immediately. Catch measurement issues before they cost you budget.

Cross-channel insights Meta's interface doesn't show. Does your Meta retargeting campaign actually drive conversions, or does it just take credit for purchases that would have happened anyway? DOJO AI helps you answer questions like this with incrementality analysis and cohort comparisons.

This is how challenger brands get enterprise-grade marketing intelligence without enterprise complexity or cost. You compete with larger competitors by being smarter, not just spending more.

See how DOJO AI's Marketing Operating System solves attribution challenges for performance marketers at challenger brands. Start your free trial.

Conclusion: Attribution Is Broken, But Fixable

Here's what you need to remember:

1. Meta ads attribution has deteriorated 40-60% since iOS 14.5. January 2026's deprecated attribution windows accelerated the decline.

2. Running CAPI + Pixel together is now mandatory. Pixel-only tracking is broken. Server-side tracking captures conversions that browsers and iOS block.

3. Adjust attribution windows to match reality. 7-day click, 1-day view is the current standard. Longer windows are gone. If you have a long sales cycle, supplement Meta attribution with CRM matching.

4. Build first-party data infrastructure for long-term resilience. Capture emails earlier. Use Advanced Matching. Send hashed customer data to Meta. Track Event Match Quality. Own your customer data.

5. Challenger brands need simplified solutions. You can't out-engineer enterprise competitors with data teams and $100K attribution platforms. Focus on the 20% of fixes (CAPI, first-party data, EMQ) that deliver 80% of accuracy improvement.

The attribution problem isn't going away. Privacy regulations will tighten. iOS restrictions will expand. Third-party cookies will die. But marketers who adapt to server-side tracking and first-party data strategies will have a massive advantage over those still relying on pixel-only setups in 2026.

The gap between advertisers who implement these fixes and those who don't is widening every quarter. Which side are you on?

Further Reading

• How to Export Meta Ads Data for 2026 Budget Planning - Complete guide to Meta Ads data extraction and analysis

• AI Marketing Attribution: Beyond Last-Click - How AI reveals true marketing impact beyond platform dashboards

• Beyond Attribution: AI Reveals Marketing-Lead Correlations - Revenue correlation modeling for challenger brands

• Meta Andromeda Explained: What Performance Marketers Need to Know - How Meta's new AI ad delivery system changes optimization strategy

• Facebook Ads Creative Strategy 2026 - Creative volume requirements in the Andromeda era

About the Author

This article was created by the DOJO AI team, drawing on official Meta developer documentation, iOS privacy update announcements, and analysis of industry-wide attribution challenges facing performance marketers in 2026.

Sources & References

• Meta for Developers: "Ads Insights API Metric Availability Updates" (October 16, 2025)

• Meta Business Help Center: "About Attribution Settings" (2025)

• Triple Whale: Industry analysis of CAPI adoption and attribution challenges (2025)

• Measured: "Meta attribution increasingly unreliable for iOS users" (November 2025)

• AdExchanger: "CAPI adoption may have plateaued" (September 2025)

• Jellyfish: Gentle Monster CAPI case study (2025)

• BrightVessel, WorkMagic: iOS 18 Link Tracking Protection analysis (2025)