Analytics and Tracking

Custom Metric

Also called Custom definition, numeric parameter

A numeric event parameter registered in GA4 so it can be totalled and averaged like a built-in measure.

Quick facts: Custom Metric

Category
Analytics and Tracking
Also called
Custom definition, numeric parameter
Level
Intermediate
Affects
Report totals, averages, value-based reporting, definition allowance
Where to see it
GA4 (Admin, Custom definitions), DebugView, Explorations
In this article4
  1. How a custom metric works
  2. Why custom metrics matter
  3. Common mistakes with custom metrics
  4. How to act on it

How a custom metric works

Sending a number with an event is not enough to report on it. GA4 keeps a register of custom definitions, and a numeric parameter only becomes something you can total, average or chart once it is registered there as a custom metric. Until then it is collected but effectively invisible outside the debugging views.

Registration happens in Admin, under custom definitions. You name the metric, point it at the parameter, choose a scope — event level for most cases, item level for a value attached to an individual product — and set a unit of measurement, which may be a plain number, a currency, a distance or a length of time. The unit matters: it decides how the figure is formatted and, for currency, how it is treated in reports.

Why custom metrics matter

They let the reports carry numbers that belong to your business rather than only the ones Google supplies. A quote value on an enquiry form, the number of items in a basket, the weight of a shipment, the length of a booking, the size of a loan application — none of these exist as built-in metrics, and all of them change how you read the rest of the report.

The gain is usually in weighting. Counting enquiries treats every enquiry alike; totalling the quoted value shows which campaigns bring the larger ones. That is often the difference between an account that looks flat and an account that is quietly improving, and it is the sort of thing worth building into any reporting dashboard from the start.

Common mistakes with custom metrics

The first is expecting history. Registration works forward only. A metric registered today will show nothing for the weeks the parameter was already being collected, so register it as soon as the parameter goes live rather than when someone finally asks for the report.

The second is registering something that is not really a measure. An order reference, a postcode or a customer number is made of digits but means nothing when summed or averaged; those belong in a custom dimension as text. The third is running out of room. A property allows only a limited set of registered definitions, and casually registering every numeric parameter fills that allowance with fields nobody reads.

How to act on it

Start from the report you want. Write the sentence you would like to say in a meeting — total quoted value by campaign, average basket size by landing page — and register only the metric that sentence needs. Give it a clear name and the correct unit, especially for money, and confirm the parameter is genuinely arriving in DebugView before you register anything.

Then leave it alone. Definitions are hard to reuse cleanly once archived, and a small, stable set of well-named metrics is far more useful a year later than a long list nobody can explain.

Do and do not

Do

  • Register the metric as soon as the parameter goes live
  • Set the correct unit, especially for currency values
  • Name it for the report sentence it will answer

Do not

  • Register reference numbers or codes as metrics
  • Expect registration to fill in earlier weeks
  • Fill the definition allowance with fields nobody reads

Questions people ask about this

Why is my custom metric showing no data for last month?

Because registration applies from the moment you create it, not retrospectively. GA4 does not go back and process the parameter values it collected before the definition existed. The values are not lost from the underlying export, but standard reports will start from the registration date. Register a metric as soon as the parameter starts being sent, not when the report is requested.

When should I use a custom metric instead of a custom dimension?

Use a metric when the value is a genuine measure that means something added up or averaged — a quoted amount, a quantity, a duration. Use a dimension when the value labels or groups something, even if it is written in digits, such as an order reference or a plan number. The test is simple: would a total of this field mean anything to a person?

Can I delete a custom metric I no longer need?

You can archive it, which frees the slot for a new definition. Archiving stops the metric appearing in reports and any report or exploration relying on it will break, so check what is using it first. Because the allowance is limited and definitions are awkward to recycle, it is worth being selective at the point of creation rather than clearing up later.

Related terms

Found this useful?

Share it, or ask an AI to summarise it

Back to the glossary

Knowing the term is the easy part

Applying it to your own site and budget is the work. Book a call and I will tell you what actually applies to you.