imbahub
Menu +
EN
Talk to us
IMBAHUB / Knowledge & guides

Two records do not necessarily mean two enquiries.

Start with a controlled test event when a report appears to count too much. Follow the action, tag firing and receiving system. Do not change the counting rule to conceal a duplicate installation before understanding the cause.

One event can take several measurement paths
One fictional test enquiry
Direct tag
Separate import
Inspect the receiving actions

Two paths are a diagnostic hypothesis, not proof of double counting.

From suspicion to a documented cause

  1. Scope

    Choose the event, environment and report definition.

  2. Trace

    Compare triggers, requests and receiving actions.

  3. Retest

    Try the same event and a new enquiry after correction.

Check what the report is adding together

A form click, an accepted submission and a qualified enquiry are different steps. Several correct events become misleading when added together as new customers. Record event name, conversion action, data source and report column before assessing the total.

One business event may be sent directly to an advertising platform and also imported from analytics as another action. That differs from two network requests fired by one tag. Inspect both the website triggers and account definitions.

Follow a fictional submission

Send one permitted test enquiry in an agreed environment. Record the time and an internal test reference without personal information. Use tag management preview and relevant network requests to identify what triggers each signal.

Try validation failures, server errors, double clicks, retries and a reloaded confirmation where the environment allows it. Verify receipt separately. A network request does not establish that the platform counted a conversion or that an email arrived.

ObservationPossible causeNext test
Two tags after one accepted responseInstallation in code and tag managementInventory installation paths
Another signal on reloadA page view is treated as successUse the agreed event condition
One request, two report actionsDirect measurement and separate importCompare sources and purposes
Click counted despite server failureIncorrect triggering conditionTest rejection without a success signal

Separate counting rules from deduplication

Google Ads distinguishes its one-or-every conversion setting from transaction IDs used to avoid recounting the same transaction. A counting setting answers a reporting question; it does not repair incorrect events or automatically merge separate conversion actions.

Where an integration uses a transaction ID, it must follow the platform's requirements, remain stable for the same transaction and distinguish genuinely new transactions. Do not hardcode one shared ID. Google Ads, Analytics and other services need not share the same deduplication mechanism; assess each connection separately.

Correct the cause and repeat the test

Record the cause, change, code or container version and expected outcome. Remove a redundant installation only after understanding its role. Repeat the test and check that a genuinely new enquiry can still be recorded.

A correction does not make historical reporting errors accurate. Keep those limitations visible when comparing periods. Bring the record to ImbaHub if you need implementation help. Share account access through an agreed secure method, never passwords in a contact form.

Sources and further reading

Questions, answered

Do two requests mean two conversions?

Not necessarily. A platform can process or reject requests differently. Inspect both the technical journey and the actual record.

Does counting one solve every duplicate?

No. The setting does not repair duplicate tags, incorrect conditions or separate conversion actions.

Does testing need real customer details?

Normally not for checking tag triggers. Use permitted fictional data and a defined environment.

IMBAHUB

What would you like to make possible?

Keep a reproducible test case before changing tags or counting rules.

Talk to us
Product interface set to English in an isolated environment with fictional English demo content.

Product interface set to English in an isolated environment with fictional English demo content.