# Google Ads Conversion Counting: One vs. Every, Primary vs. Secondary

Audit Google Ads conversion actions by separating One vs. Every counting from Primary vs. Secondary optimization before inflated events train bidding.

- **Author:** [R. Garcia](https://offleashmarketer.com/authors/r-garcia/)
- **Published:** 2026-08-27
- **Updated:** 2026-08-27
- **Last reviewed:** 2026-08-27
- **Evidence level:** docs reviewed
- **Canonical URL:** https://offleashmarketer.com/blog/google-ads-conversion-counting-one-vs-every/

> **Theo's read:** One vs. Every controls how repeat events are counted; Primary vs. Secondary controls whether an action is eligible to guide bidding. They solve different problems.

## Quick diagnostic

- **Symptom:** Lead totals look unusually high, several funnel events sit in the Conversions column, or the bidding strategy appears to optimize for activity that sales does not recognize.
- **Likely break:** A repeatable lead event is set to Every, an observational event is Primary, duplicate implementations record the same outcome, or a custom goal quietly includes a Secondary action.
- **First check:** Segment campaign reporting by conversion action, then document the Count setting, optimization status, selected campaign goal, source, trigger, and repeat rate for each action.

One of the most important findings in an early Google Ads audit was not a keyword or bid. It was the conversion-action table.

The account had a booked demo, a free trial, an “enter credit card” event, an older tracking implementation, a GA4 implementation, and a 30-second page timer. Several were positioned as meaningful conversions. Some appeared to describe the same business milestone. One engagement event had been allowed into the Primary layer.

That creates two separate questions marketers often collapse into one:

1. Should repeated events count once or every time?
2. Should this event guide bidding or remain observational?

Google Ads answers those questions with two different settings.

## One vs. Every Controls Repetition

The **Count** setting determines how Google Ads counts multiple conversions for the same conversion action after an ad interaction.

- **Every** counts every recorded conversion for that action after the interaction.
- **One** counts one conversion for that action per ad click.

[Google's conversion-counting documentation](https://support.google.com/google-ads/answer/3438531) recommends Every when each sale adds value and One when the goal is whether a unique lead was generated.

That leads to a practical default:

- purchases, paid subscriptions, and separately valuable transactions often use **Every**;
- lead forms, demo requests, and similar acquisition events often use **One**.

It is a default, not a universal law. If one click legitimately creates two separately valuable consultation requests, Every may reflect the business. If a buyer refreshes a thank-you page and fires the same purchase again, Every exposes a broken implementation rather than real additional revenue.

## Primary vs. Secondary Controls Optimization

The **Action optimization** setting answers a different question.

- **Primary** actions can appear in the Conversions column and influence bidding when the campaign uses the goal containing that action.
- **Secondary** actions are normally observation-only and appear in All conversions.

Google documents an important exception: if a Secondary action is placed in a custom goal and that custom goal is selected by a campaign, the action can still be used for bidding. See [Google's explanation of conversion goals](https://support.google.com/google-ads/answer/10995103).

This means “Secondary” is not enough evidence by itself. An audit must trace all four layers:

**Event → conversion action → conversion goal → campaign goal selection**

Only then can you say what bidding is learning from.

## Use a Four-Cell Decision Table

The two settings create four common combinations.

| Count | Optimization | Common Use | Main Risk |
| --- | --- | --- | --- |
| One | Primary | Lead or demo request used for bidding | Repeat people across different clicks can still be counted again |
| Every | Primary | Purchases or separately valuable transactions | Duplicate fires directly inflate bidding evidence |
| One | Secondary | Observational lead-stage event | It may be too far from revenue to evaluate quality |
| Every | Secondary | Repeat engagement or diagnostic activity | All conversions can look large even though bidding ignores it |

The table is a starting point. The event's business definition and technical trigger decide the correct setting.

## Why Lead Accounts Commonly Overcount

Lead-generation sites often fire events from actions that can repeat without creating another prospect:

- refreshing or revisiting a thank-you page;
- submitting a form more than once;
- opening a scheduling tool repeatedly;
- clicking a contact link and later submitting a form;
- triggering both a native Google Ads tag and an imported GA4 event;
- logging into an existing customer account;
- reaching a timer or scroll threshold;
- firing the browser event before server-side validation succeeds.

Setting a lead action to One can reduce repetition after a click, but it cannot prove uniqueness across different clicks, devices, cookies, or conversion actions. It also cannot distinguish a prospect from an existing customer.

That requires a downstream identifier and CRM reconciliation. It also requires care when comparing platform totals with analytics acquisition reports because a scope change can alter the population behind the metric.

## Start With the Conversion-Action Segment

Do not begin by changing the headline conversion total.

Segment campaign reporting by conversion action. For each action, document:

- exact name and category;
- source: website tag, GA4, call, CRM import, app, or another system;
- trigger condition;
- Count setting;
- repeat rate;
- Primary or Secondary status;
- standard or custom goal membership;
- campaigns using that goal;
- value and currency;
- attribution model and windows;
- first recorded date and recent volume;
- equivalent events in other sources.

Google defines repeat rate as the number that Every would record divided by what One would record among interactions that converted. A high repeat rate is not automatically an error, but it is a reason to understand the event.

## Decide Whether the Event Represents Value or Progress

A 30-second timer may help describe engagement. It rarely represents the same business outcome as a qualified lead, paid account, or purchase.

Ask four questions:

- Did the person create new economic value?
- Did the person reach a stage the business can verify?
- Can the event occur without the intended customer outcome?
- Would you deliberately ask the bidding system to find more people who perform this action?

That final question is the sharpest. If the answer is no, the event probably should not sit in the Primary optimization path.

Micro-conversions still have uses. They can help diagnose pages, build audiences where permitted, or show funnel progression. Keep them visible as observations without presenting them as acquired customers.

## Reconcile Duplicate Sources Before Choosing a Winner

In the old audit, an older tag appeared to count more free trials than the GA4 import. The tempting response is to select the larger or smaller number based on which story feels more plausible.

Instead, run a record-level test:

1. Submit a controlled test using a unique identifier.
2. Preserve the network, tag-manager, data-layer, GA4, Google Ads, and CRM evidence.
3. Confirm which actions fire and when.
4. Repeat without a new lead: refresh, revisit, log in, or repeat the action.
5. Compare reporting after the normal processing delay.
6. Trace several real, sanitized records through the CRM.

The goal is not to make Google Ads and GA4 totals identical. Their attribution and processing can differ. The goal is to prove what each event represents and eliminate technical duplicates.

## Change Bidding Inputs Carefully

Removing an inflated Primary action can make conversion volume fall immediately. That does not mean the correction harmed the business. It means the old numerator included activity the new definition excludes.

Google recommends managing material conversion-goal changes with attention to Smart Bidding because changing actions, sources, or optimization status changes the evidence the strategy receives. See [Google's current guidance on changing conversion goals](https://support.google.com/google-ads/answer/14571185).

Before changing the action:

- export the old configuration;
- annotate the change date;
- estimate the volume and value being removed;
- choose a stable replacement business action;
- avoid simultaneous budget, target, and structural changes;
- allow for conversion lag and bidding adaptation;
- report the definition break explicitly.

Do not backfill a clean historical trend across incompatible definitions.

## My Rule for Lead Conversion Counting

For a lead action, I start with **One** unless multiple events after the same click create separately valuable prospects. I make it **Primary** only when the business wants bidding to pursue it and can verify that the trigger represents the intended milestone.

Then I test the exceptions.

One does not equal one person. Primary does not equal important in every context. Secondary does not guarantee observation-only when custom goals are involved. The labels are useful only after the event, goal, campaign, and downstream record are connected.

The next audit in this sequence should cover conversion windows, because a correct event can still tell a misleading story when its credit window does not match the buying cycle.

## Before you act

- [ ] Export conversion actions and campaign goal settings before changing them
- [ ] Reconcile sample event IDs or customer records against the CRM or order system
- [ ] Record the current bidding strategy and allow for learning disruption when goals change

## Theo's challenge

**Changing Every to One can hide repeat behavior instead of fixing it.**

One may produce a more useful lead count, but it does not deduplicate people across clicks, devices, sources, or separate conversion actions.

- Are repeat events legitimate purchases, duplicate leads, multiple products, or a broken trigger?
- Does the CRM have a stable identifier that can reveal unique people and downstream outcomes?
- Will changing the goal alter bidding abruptly enough to require a staged transition?