This is the single highest-leverage fix available in most Indian ad accounts, and it is also the one nobody wants to do because there are no screenshots to show for it.
What the Pixel actually is
A piece of JavaScript that fires from the visitor's browser when something happens — page loaded, form submitted, purchase completed — and reports it to Meta.
Meta uses those reports to learn who converts and find more people like them. Everything about campaign optimisation depends on that feedback loop.
Why the browser-side Pixel is not enough
Three things break it:
iOS privacy restrictions. App Tracking Transparency limits what can be attributed for a large share of users.
Ad blockers and browser protections. Safari's tracking prevention and standard blockers stop Pixel requests entirely.
Network failures. The visitor closes the tab, the connection drops, the script does not fire.
The result is that a meaningful share of real conversions never reach Meta. Estimates vary, but the practical range is 30–60% of conversions missing.
What Conversion API does differently
CAPI sends conversion data server to server — from your website's backend or a tag manager server container directly to Meta — bypassing the browser entirely.
Ad blockers cannot stop it. iOS restrictions do not apply the same way. The tab closing does not matter.
Running both together, with event deduplication so a single conversion is not counted twice, gives the algorithm a materially more complete picture.
What it is worth in practice
On an NGO donation account with Pixel installed but no CAPI, cost per result sat at ₹175.61≈ $1.81. After implementing dual tracking with mapped events, the scaled campaign reached ₹85.48≈ $0.88.
Creative improved somewhat. Audience testing helped. But the biggest contributor was that the algorithm stopped optimising on a distorted sample of who actually donated.
The events worth configuring
Do not track everything. Track the sequence that describes your funnel:
- PageView — baseline
- ViewContent — reached a key page
- InitiateCheckout or form start — declared intent
- Lead — form submitted
- Purchase — money moved, with value and currency
The value parameter matters. Without it, Meta optimises for conversion count and will happily find you many small conversions instead of fewer large ones.
Setup paths, briefly
Partner integrations (Shopify, WooCommerce, WordPress plugins) — easiest, least flexible.
Google Tag Manager server-side container — the standard for most businesses. Requires a server container, typically on Google Cloud.
Direct API implementation — a developer sends events from your backend. Most control, most work, and the most reliable for custom checkouts.
Whichever path, you need event deduplication configured, or a single conversion reported by both Pixel and CAPI gets double-counted and your data becomes worse rather than better.
How to verify it is actually working
- Events Manager → Test Events. Fire a real conversion and confirm it appears.
- Check the Event Match Quality score. Below about 5.0 means you are sending insufficient customer parameters — add email, phone, and external ID where you have consent.
- Look for deduplication warnings. If Meta reports duplicate events, your event IDs are not matching.
- Compare against your payment gateway. This is the real test. If Razorpay says 11,246 payments and Meta says 6,000, you have a tracking gap, not a campaign problem.
That last check is the one most people never run, and it is the only one that reflects reality.
The most common mistakes
- Installing CAPI without deduplication, then trusting inflated numbers
- Tracking form views as conversions instead of submissions
- Omitting the value parameter, so all purchases look equal
- Never sending offline conversions back, so the platform never learns which leads actually became customers
- Assuming the plugin worked and never testing it
The order of work
Get tracking right before you spend meaningfully. Sending traffic through a broken measurement layer means every optimisation decision after that point inherits the error — and you will spend months adjusting creative to fix a data problem.
If you want the whole stack checked, that is where our free growth audit begins.