Skip to main content
This page explains every stage between a Datalyr event and an accepted ad-platform conversion.

​
The five stages

A failure at any stage leaves the earlier stages intact, which is why the two screens can disagree. Events can hold an event Conversions never shows. Conversions can hold a row the ad platform rejected.

​
What a rule holds

The editor has two steps. Step 1 sets the trigger and the destination. Step 2 sets the mapping and the conditions.

​
Property conditions

A rule with conditions sends only the trigger events whose properties match every condition. The rule does not send an event that fails a condition. It writes no row for that event. Other rules on the same trigger still send it. A property is a path into the event, such as checkout, order.channel, or geo_country_code. Paths resolve the same way as value_path. Comparison rules:
  • Text compares without case and without leading or trailing spaces. Stripe equals stripe.
  • A missing property matches only is not, is not one of, and is not set.
  • >, ≥, < and ≤ need a number. £12.50 and 1,299.00 are numbers. 2026-10-09 and SKU-12 are not, and do not match.
A property that holds an object or a list matches only is set. Limits:
  • Up to 5 conditions per rule. All of them must match.
  • No conditions on user_properties.
  • No conditions on personal fields, such as email, phone, IP address, name, or address.
  • No conditions on identifiers, such as user_id and visitor_id.
  • No conditions on a Stripe trial_started rule.
If stored conditions are not valid, the rule sends nothing and writes the skip reason property_conditions_invalid.

​
How the value resolves

Datalyr tries each source in order and stops at the first number it finds.
  1. The rule’s value_path, read against the event.
  2. The rule’s value_static.
  3. The event’s own value, revenue, total, or amount.
  4. Zero.
A parameter_mappings entry that targets value overrides all four.
Meta, TikTok, and Snapchat reject a purchase worth zero. We skip that delivery with the reason zero_value_purchase rather than send a request that always fails.

​
How identity resolves

We look for a click ID on the event itself first. If there isn’t one, we search history. Datalyr also fills hashed match keys from the stored user profile and from the event’s geo_city, geo_region, and geo_country_code. An event carrying consent.marketing: false, consent.sale: false, or consent.ad_user_data: false ships with no user-level identifier at all.

​
Deduplication

Every platform receives the browser-minted event_id as its dedup key. The browser pixel fires with the same value, so the pair collapses into one conversion. Datalyr also takes a 300-second Redis claim per event and rule. A queue redelivery inside that window is skipped instead of resent.
Do not enable two rules for the same trigger event and the same destination asset. Each rule delivers separately with a different rule ID, and the ad platform counts the conversion twice.

​
Statuses

​
Skip reasons

​
Retries

The dead-letter queue runs three rounds. After that the row reads dropped. Google conversions never enter the queue, because Google validates the upload later through its own diagnostics poll. Two circuit breakers protect a broken setup. Five straight 5xx responses pause one workspace and platform for 60 seconds. Ten auth or config failures pause one rule for one hour.

​
Verify

  1. Open Conversions. Select manage rules. Open your rule.
  2. Select send test event.
  3. Read the rule’s matched, sent, and failed counts.
  4. Open the newest row under recent activity.
  5. Confirm the response body carries the ad platform’s own acceptance field.

​
When it does not work

​
Next

  • Conversions: read one delivery row end to end.
  • Meta Ads: the exact fields and hashing per platform.
  • Attribution: how a click ID reaches a server-side purchase.