{"id":3423,"date":"2026-05-18T10:29:53","date_gmt":"2026-05-18T10:29:53","guid":{"rendered":"https:\/\/hyros.com\/updates\/?p=3423"},"modified":"2026-08-03T04:30:16","modified_gmt":"2026-08-03T04:30:16","slug":"meta-conversions-api","status":"publish","type":"post","link":"https:\/\/hyros.com\/updates\/meta-conversions-api\/","title":{"rendered":"Facebook Conversions API (CAPI) Setup: Complete Guide"},"content":{"rendered":"\n

Here is what most Meta advertisers do not want to hear: if you are running only the Pixel in 2026, more than half your conversions are invisible to Meta’s algorithm. That is not a scare tactic. That is what happens when you rely on browser-side JavaScript to track a world full of ad blockers and iOS privacy changes. I originally built Hyros because I was scaling ad spend on Meta and the numbers I was seeing made no sense — campaigns that were clearly working looked like they were barely breaking even. The root cause was always the same. The data pipeline was broken. The Facebook Conversions API (CAPI) is Meta’s server-side tracking interface that sends conversion events — purchases, leads, add-to-carts — directly from an advertiser’s server to Meta’s servers, bypassing the browser entirely. CAPI was introduced in 2020 and became the Meta-recommended setup after iOS 14.5 App Tracking Transparency launched in April 2021, which caused the Meta Pixel to lose visibility on a large share of iPhone conversions. Running CAPI alongside the Pixel (what Meta calls a “redundant setup”) gives advertisers more complete data for optimization and reduces cost per action. According to Meta’s April 2026 announcement<\/a>, advertisers with a Conversions API setup for web events saw an average 17.8% lower cost per result compared to those without CAPI.<\/p>\n\n\n\n

TL;DR<\/h2>\n

The Meta Conversions API sends your conversion data server-to-server, sidestepping the browser problems (ad blockers, cookie expiration, iOS App Tracking Transparency) that have meaningfully degraded Pixel accuracy on iOS traffic. There are three ways to set it up: manual API integration where you write the HTTP POST requests yourself, Meta’s hosted CAPI Gateway that runs on AWS CloudFormation with minimal code, or a managed attribution platform like Hyros that handles CAPI as part of its server-side tracking layer. Regardless of which path you choose, event deduplication is non-negotiable — every CAPI event needs an event_id<\/code> that matches the corresponding Pixel event, or you will double-count conversions. Your Event Match Quality score should be above 6.0 at minimum and ideally above 8.0 for strong identity matching. After iOS 14.5 removed tracking consent from roughly 96% of US iPhone users initially, running CAPI alongside the Pixel is no longer a nice-to-have. It is the minimum standard for accurate Meta Ads measurement.<\/p><\/div>\n\n\n\n

What Is the Facebook Conversions API?<\/h2>\n\n\n\n

The Conversions API is a server-side data pipeline. That phrase matters because it describes exactly how it differs from the Pixel.<\/p>\n\n\n\n

The Meta Pixel is JavaScript code that runs in the user’s browser. When someone completes a purchase, the Pixel fires a browser-side event and sends it to Meta. The problem is that this event has to survive a minefield: ad blockers that prevent the script from loading, Safari’s Intelligent Tracking Prevention that caps cookies to 7 days (sometimes 24 hours), iOS App Tracking Transparency that blocks cross-app tracking altogether, and plain old page-load failures.<\/p>\n\n\n\n

CAPI bypasses all of that. When a conversion happens, your server sends the event data directly to Meta’s Graph API via an HTTPS POST request. The data never touches the browser. Ad blockers cannot intercept what they cannot see. Cookie restrictions do not apply because there is no cookie involved. And iOS privacy restrictions are irrelevant to a server-to-server data transfer.<\/p>\n\n\n\n

Each CAPI event includes hashed customer identifiers — email address, phone number, first name, last name — that Meta uses to match the event back to a specific user in its system. When it finds a match, and that user previously interacted with one of your ads, Meta attributes the conversion. The hashing uses SHA-256, which means you never send raw customer data to Meta; you send an irreversible hash.<\/p>\n\n\n\n

In May 2025, Meta retired the separate Offline Conversions API entirely. All offline conversion tracking now flows through CAPI. That consolidation made CAPI the single server-side entry point for every type of conversion data you want Meta to have.<\/p>\n\n\n\n

Why Did CAPI Become Essential After iOS 14.5?<\/h2>\n\n\n\n

Apple’s App Tracking Transparency framework launched in April 2021 and fundamentally changed digital ad measurement. ATT requires apps to ask users for explicit permission to track them across apps and websites. The result was immediate and dramatic.<\/p>\n\n\n\n

According to Flurry Analytics, which tracked behavior across over 1 million mobile apps, roughly 96% of US iPhone users initially opted out of tracking. That number has softened over time — current opt-in rates sit at approximately 35-37% according to Adjust’s Q2 2025 data — but the damage to browser-based tracking was structural.<\/p>\n\n\n\n

Meta felt the impact. During its Q4 2021 earnings call in February 2022, Meta CFO David Wehner projected approximately $10 billion in 2022 ad revenue loss from iOS ATT changes. That was Meta’s own money, not advertiser losses. The downstream effect on advertisers was arguably worse: the Pixel lost the ability to see conversions from the majority of iPhone users, and Meta’s attribution window permanently collapsed from 28-day click to 7-day click, 1-day view.<\/p>\n\n\n\n

The practical outcome for advertisers running Pixel-only tracking in 2026 is bleak. Stores using only the Pixel routinely lose visibility on a meaningful share of iOS conversions. Meta itself stopped publishing a single accuracy figure because the gap varies by traffic mix, consent rates, and ad-blocker prevalence. Your ROAS dashboard shows a fraction of what is actually happening.<\/p>\n\n\n\n

CAPI was Meta’s answer. By receiving conversion data server-side with hashed PII for identity matching, CAPI recovers a significant portion of what the Pixel lost. Advertisers running CAPI alongside the Pixel recover meaningful event volume the Pixel alone cannot capture. Per Meta’s April 2026 update<\/a>, web-event CAPI setups averaged 17.8% lower cost per result versus no-CAPI setups. Not because the ads got better — because the algorithm finally has enough data to optimize properly.<\/p>\n\n\n\n

For a deeper look at how iOS 14.5 reshaped the entire attribution landscape, see our guide on how iOS 14.5 changed ad attribution<\/a>.<\/p>\n\n\n\n

CAPI vs the Meta Pixel: How Do They Compare?<\/h2>\n\n\n\n
\"Circuit<\/figure>\n\n\n\n

The Pixel and CAPI are not competing technologies. They are complementary. But understanding where each one works — and where it fails — matters for setup decisions.<\/p>\n\n\n\n

Feature<\/th>Meta Pixel<\/th>Conversions API<\/th><\/tr><\/thead>
Data source<\/td>User’s browser<\/td>Your server<\/td><\/tr>
Cookie dependent<\/td>Yes — relies on JavaScript cookies<\/td>No — server-to-server<\/td><\/tr>
Ad blocker resistant<\/td>No — blocked by all major ad blockers<\/td>Yes — invisible to ad blockers<\/td><\/tr>
iOS ATT impact<\/td>Severe — loses most iPhone conversion data<\/td>Minimal — server-side data unaffected by ATT<\/td><\/tr>
Event deduplication<\/td>N\/A (sends to Meta directly)<\/td>Must include matching event_id<\/code><\/td><\/tr>
Setup difficulty<\/td>Low — paste a JavaScript snippet<\/td>Medium to high — requires server access or integration platform<\/td><\/tr>
Ongoing cost<\/td>Free<\/td>Free (manual), $10-$400\/mo (Gateway hosting), or included in platform pricing<\/td><\/tr>
Matching method<\/td>Cookie-based (fbp, fbc)<\/td>PII-based (hashed email, phone) + cookies<\/td><\/tr>
Real-time<\/td>Yes — fires on page load<\/td>Near real-time — depends on server implementation<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n

The important takeaway: the Pixel alone leaves measurable conversions on the table. Running both together is now Meta’s recommended setup. The Pixel catches what it can from the browser, CAPI catches what the browser cannot deliver.<\/p>\n\n\n\n

How Do You Set Up Facebook CAPI?<\/h2>\n\n\n\n
\"Traces<\/figure>\n\n\n\n

There are three paths. Manual API integration gives you full control but requires developer resources. Meta’s CAPI Gateway handles the infrastructure for you at a hosting cost. And managed platforms like Hyros handle CAPI as part of their broader attribution system.<\/p>\n\n\n\n

Regardless of which path you pick, the setup follows six core steps.<\/p>\n\n\n\n

Step 1: Create a dataset in Meta Events Manager<\/h3>\n\n\n\n

Meta renamed “Pixels” to “datasets” in its backend terminology, though the interface still uses both words interchangeably. Go to Meta Events Manager, select your business, and either create a new dataset or use your existing Pixel ID. The dataset ID is the same string you have been using as your Pixel ID — it is a 15-16 digit number.<\/p>\n\n\n\n

This dataset will receive events from both your Pixel (browser-side) and CAPI (server-side). Both channels feed into the same dataset.<\/p>\n\n\n\n

Step 2: Generate a CAPI access token<\/h3>\n\n\n\n

Go to Business Settings in your Meta Business Suite. Under Data Sources, find your dataset (Pixel). Open Settings and scroll to the Conversions API section. Click “Generate Access Token.”<\/p>\n\n\n\n

This token authenticates your server’s POST requests to Meta. Treat it like a password. It does not expire automatically, but you can revoke and regenerate it at any time. Store it in your server’s environment variables — never hardcode it in client-facing code.<\/p>\n\n\n\n

Step 3: Choose your integration path<\/h3>\n\n\n\n

Manual API integration.<\/strong> You write the HTTP POST requests yourself. The endpoint is:<\/p>\n\n\n\n

POST https:\/\/graph.facebook.com\/v19.0\/{DATASET_ID}\/events\n<\/code><\/pre>\n\n\n\n
Each request includes your access token and a JSON payload with the event data. Here is a minimal cURL example for a purchase event:<\/p>\n\n\n\n
curl -X POST \\\n  \"https:\/\/graph.facebook.com\/v19.0\/YOUR_PIXEL_ID\/events\" \\\n  -H \"Content-Type: application\/json\" \\\n  -d '{\n    \"data\": [{\n      \"event_name\": \"Purchase\",\n      \"event_time\": 1713100000,\n      \"event_id\": \"order_12345_1713100000\",\n      \"event_source_url\": \"https:\/\/yourstore.com\/thank-you\",\n      \"action_source\": \"website\",\n      \"user_data\": {\n        \"em\": [\"309a0a5c3e211326ae75b3e33b5e2794ea5a391ffb220657de92b6acbf12cdd7\"],\n        \"ph\": [\"254aa248acb47dd654ca3ea53f48c2c26d641550feae8e25841aeac7b2385e49\"],\n        \"client_ip_address\": \"123.45.67.89\",\n        \"client_user_agent\": \"Mozilla\/5.0...\",\n        \"fbc\": \"fb.1.1713099000.ABCdef123\",\n        \"fbp\": \"fb.1.1713098000.987654321\"\n      },\n      \"custom_data\": {\n        \"currency\": \"USD\",\n        \"value\": 149.99,\n        \"order_id\": \"12345\"\n      }\n    }],\n    \"access_token\": \"YOUR_ACCESS_TOKEN\"\n  }'\n<\/code><\/pre>\n\n\n\n
The em<\/code> and ph<\/code> fields are SHA-256 hashed. The fbc<\/code> and fbp<\/code> values come from the Meta cookies set by the Pixel in the user’s browser — you pass them through to CAPI unhashed.<\/p>\n\n\n\n
CAPI Gateway.<\/strong> Meta provides a pre-built infrastructure package that deploys on AWS via CloudFormation. You click through a setup wizard in Events Manager, authorize your AWS account, and Meta spins up the server infrastructure. Hosting costs run $10-$400 per month depending on event volume. The advantage is that you do not write API code. The downside is you are managing an AWS stack.<\/p>\n\n\n\n
Managed platform.<\/strong> Attribution platforms like Hyros handle CAPI integration as part of their tracking setup. You connect your Meta account, Hyros sends events through CAPI automatically, deduplication is built in, and you do not touch API code or cloud infrastructure. The cost rolls into your platform subscription — no separate line item, no AWS bill to manage.<\/p>\n\n\n\n
Step 4: Send a test event<\/h3>\n\n\n\n
Before going live, use Meta’s Test Events tool in Events Manager. Open your dataset, click “Test Events,” and find the test event code. Add this test code to your CAPI request payload as test_event_code<\/code>. Fire a test event and confirm it appears in the Test Events tab.<\/p>\n\n\n\n
Test events do not affect your live data or ad optimization. They exist purely for debugging. Remove the test code before pushing to production.<\/p>\n\n\n\n
Step 5: Implement event deduplication<\/h3>\n\n\n\n
This is where most CAPI implementations break. When you run CAPI alongside the Pixel, both systems send the same event to Meta. Without deduplication, Meta counts the purchase twice.<\/p>\n\n\n\n
Deduplication works through the event_id<\/code> field. You generate a unique identifier on the client side (typically a combination of order ID and timestamp) and attach it to both the Pixel event and the CAPI event. When Meta receives two events with the same event_id<\/code>, same event name, and same dataset ID within 48 hours, it keeps one and discards the duplicate.<\/p>\n\n\n\n
The most common mistake: generating different IDs for the Pixel event and the CAPI event. If the Pixel sends event_id: \"purchase_12345\"<\/code> and your server sends event_id: \"order_12345\"<\/code>, Meta treats them as two separate purchases. Your conversion count doubles, your CPA looks half of reality, and your optimization data is poisoned.<\/p>\n\n\n\n
Generate the event_id<\/code> once, on the client side. Pass it to the Pixel. Also pass it to your server (via a hidden form field, cookie, or data layer variable) so that CAPI sends the identical value.<\/p>\n\n\n\n
Step 6: Verify your Event Match Quality score<\/h3>\n\n\n\n
Event Match Quality (EMQ) measures how well Meta can match your server-side events to real Meta user profiles. Check it in Events Manager under the Event Match Quality column next to each event type.<\/p>\n\n\n\n
EMQ Score<\/th>What it means<\/th><\/tr><\/thead>
Below 6.0<\/td>Poor — significant matching gaps, CAPI is underperforming<\/td><\/tr>
6.0-7.0<\/td>Acceptable — basic matching working, room to improve<\/td><\/tr>
7.0-8.0<\/td>Good — strong identity resolution<\/td><\/tr>
Above 8.0<\/td>Excellent — near-complete matching<\/td><\/tr><\/tbody><\/table><\/figure>\n\n\n\n
To improve your EMQ score, send more matching parameters with each event:<\/p>\n\n\n\n