Conversion API

Platforms
6 min read
Updated June 13, 2026

Why it matters

Browser tracking alone misses conversions, degrades with privacy changes, and weakens identity match. Conversion APIs send events from infrastructure you control, with hashed identifiers, click IDs, and custom parameters, so platforms can attribute outcomes and tune delivery.

For teams moving beyond top-funnel proxies, Conversion API is how value signals reach ad networks: purchase amounts, custom values, or user-level pLTV scores. Weak setup (low match rate, missing deduplication, wrong timing) is a leading cause of value-based bidding underperformance despite strong models.

Each network names and schemas its API differently, but the pattern is consistent: HTTPS events, event time, value, currency, user data, and dedup keys when paired with client-side tags.

Conversion API

Conversion API is the activation layer, not the model:

  1. Input: First-party data and revenue history in your data warehouse.
  2. Model: Churney produces user-level pLTV and signal transformation rules.
  3. Deliver: Send predictive events directly to ad networks via Meta CAPI, Google server-side conversion paths, TikTok Events API, or MMP/app analytics paths for mobile.
  4. Optimize: Campaigns use value or ROAS objectives; platform learning adjusts delivery.
  5. Prove: Holdout tests and experiment readout at maturity window vs BAU.

The data warehouse feeds modeling; Conversion API pipes carry the optimized signal to platforms.

Category variants

NetworkImplementationpLTV relevance
MetaMeta Conversions API (CAPI)Primary pipe for NPVO and custom value events; EMQ diagnostics.
Google AdsEnhanced Conversions / offline conversion upload via Google Ads APIValue rules and conversion actions for tROAS campaigns.
TikTokTikTok Events APIServer events for value optimization and match quality.
Mobile appMMP postbacks may forward to networksOften paired with CAPI for redundancy on iOS.

Common mistakes

  1. Tags only, no server path. Incomplete capture and weak match over time.
  2. Treating all networks as one schema. Each API has distinct fields and eligibility rules.
  3. Skipping deduplication. Duplicate events when browser and server both fire unpaired IDs.
  4. Ignoring match rate. Low identifier completeness limits optimization on value events.
  5. Late event_time. Delays hurt value-optimized campaigns beyond a few hours.
  6. Confusing pipe with model. Conversion API delivers signals; data warehouse trains pLTV.

Advertiser lens

RoleWhat they askWhat good looks like
Head of Performance / UAAre we server-side on all major spend?Coverage map by network, event volume, and value optimization eligibility.
VP Growth / CMOIs engineering lift worth it?Documented match uplift and path to pLTV campaigns.
Marketing Analytics / Data ScienceAre values calibrated per network?Schema docs, value distributions, experiment design vs BAU.
Data EngineeringWho owns endpoints and PII hashing?Network-specific playbooks, monitoring, and on-call rotation.

FAQ

What is Conversion API?

A server-side method for sending conversion and customer action events from your systems to ad platforms, as opposed to browser-only tracking.

How is generic Conversion API different from Meta CAPI?

Meta CAPI is Meta's branded implementation. "Conversion API" is the cross-platform category; always specify the network when implementing.

Do I need Conversion API for pLTV?

For web and many app flows, yes. Server-side delivery with strong match is the standard path for sending pLTV values reliably.

Can one integration serve all platforms?

Unlikely. Each network has distinct endpoints, parameters, and dedup rules. Shared orchestration can feed multiple APIs from one model output.

What identifiers improve match rate?

Hashed email and phone, click IDs, browser IDs, and consistent external IDs from first-party data. Freshness and completeness matter.

Does Conversion API replace incrementality measurement?

No. It improves signal delivery; holdout tests and geo experiments still prove causal lift.

Who implements Conversion API?

Marketing defines events; engineering or partners implement endpoints; analytics validates match and readout.

Not the same as

TermDifference
Meta Conversions API (CAPI)Meta-specific product; Conversion API is the generic category.
Pixel / tagClient-side tracking; complementary, not replacement.
Data warehouse exportAnalytics storage; not an ad network delivery pipe.
MMP SDKIn-app measurement layer; may forward to Conversion APIs.