When a GA4 report gives a conversion to your ad campaign, it is because the click fell inside the key event lookback window. According to the Google Analytics attribution settings help page (Opens in a new tab), acquisition key events (first_open, first_visit) default to 30 days with 7 days as the alternative, and every other key event defaults to 90 days with 30 or 60 days as alternatives. If you buy ads and read the results yourself, check that window and the report date range before you judge a campaign, because the same sale can be credited differently in two reports built with different settings.
What a lookback window settles
A shopper clicks an ad on Monday, comes back through a search on Saturday and buys. One sale, two touches. The lookback window sets how far back from the purchase GA4 may look for touches that can receive credit. A touch outside the window is not considered. A touch inside it competes for credit under the attribution model you selected.
Example scenario: An aquarium shop runs a three-week display campaign for freshwater starter tanks. A tank is a considered purchase, because the buyer reads about water chemistry, compares sizes and asks a friend. The shop closes its report on the last day of the campaign and sees a thin conversion count. Weeks later the same campaign shows more conversions, because sales that came after the campaign ended were still linked back to ad touches inside the window. The early report was not wrong, it was early.
The window options in plain terms
Think of the window as a number of days measured backward from the conversion. GA4 offers it in two sets, depending on the kind of key event.
Table: lookback window by key event type
| Key event type | Default | Other options |
|---|---|---|
| Acquisition events (first_open, first_visit) | 30 days | 7 days |
| All other key events | 90 days | 30 days, 60 days |
Google's help pages can change; check the current page before you rely on a specific option.
Acquisition events (first_open, first_visit)
These events mark a user's first arrival. Choosing 7 days narrows the credit to touches close to that arrival. Staying at 30 days keeps earlier touches eligible. Practical suggestion: pick 7 days only if your campaign aims at impulse arrivals, such as a weekend promotion, and keep 30 days when people research first.
Purchases, sign-ups and the rest
A purchase, a form submission or a registration follows the 90-day default unless you change it. Choosing 60 or 30 days shortens the lookback. Shorter windows credit the touches nearest the conversion. Longer windows let the first campaign touch stay in the picture.
Marking events as key changes the result
The window applies only to key events, so the list of key events shapes the report. The aquarium shop might mark a purchase, a newsletter sign-up and a store-pickup request. A newsletter sign-up inflates the conversion count without representing revenue. Practical suggestion: in a campaign report, show revenue-related key events separately from supporting ones.
- Purchase: the direct campaign goal
- Add to cart: a step showing the decision is advancing
- Newsletter sign-up: preparation for a later sale, not reported as the campaign result
Because acquisition events and other key events use different windows, do not compare them on the same row without a note. An acquisition event is measured over 30 or 7 days, and a purchase over 90, 60 or 30 days, so the two are not counting the same stretch of time.
Three controls on the Attribution settings screen
The help page places key event attribution settings in Admin, under Data display, in Events, on the Attribution settings screen. If you use the interface in another language, the labels may look different. Three controls sit together on that screen.
Table: GA4 key event attribution controls
| Control | What it decides | Choices according to the help page |
|---|---|---|
| Reporting attribution model | How credit is shared between touches | Data-driven and rules-based models. First click, linear, time decay and position-based models are no longer available as of November 2023 |
| Channels that can receive credit | Which channel set is eligible | Google paid channels or Paid and organic channels |
| Key event lookback window | How far back from the key event GA4 looks | Acquisition events: 30 days default, 7 days option. Other key events: 90 days default, 30 and 60 days options |
The page adds that these conversion attribution settings apply to Google Analytics and Google Ads conversion reports and to campaign optimization. Lookback windows only matter for events marked as key events, so set up the events first. For a step-by-step setup, see GA4 key event setup for forms, purchases and landing goals.
Who can claim credit: Google paid channels or paid and organic
The Channels that can receive credit control has two options. With Google paid channels, Analytics conversion reports show Google Ads conversions. With Paid and organic channels, conversions from organic sources are included for a broader cross-channel view. The page says this setting applies going forward to all linked Google Ads accounts.
Two limits to remember
- It applies only to web conversions. App conversions always use Google paid channels.
- When it is selected, the conversion window and counting method are managed in Google Analytics.
One report read under both settings
A paid-only view treats Google ads as the eligible channel set. A paid and organic view can place the ad click and the organic return in the same journey, which changes who is credited. Write the selected channel set into the report note. For the organic side of this question, see SEO assisted conversions and attribution models; that article looks at organic search contribution, while this one concentrates on window choice for ad campaigns.
What happens to past data when you change a setting
The help page separates two behaviors. Changing the reporting attribution model applies to historical and future data. Changes to the lookback window apply going forward. This is the key distinction for a post campaign report.
Table: effect of a setting change
| Change | Past data | New data |
|---|---|---|
| Reporting attribution model | Affected | Affected |
| Lookback window | Not affected | Affected |
| Channels that can receive credit (Paid and organic channels) | Applied going forward, per the help page | Affected |
If you need past data recalculated, a model change does that and a window change does not. Moving from 90 days to 30 days therefore does not rewrite last quarter. When this month's report and last month's report were built with different windows, note the change date before you compare them.
Leave the window alone mid-campaign
A window change creates a break in the data series on the day you make it. Changing the setting halfway through a campaign means the first half and the second half are measured under different rules. Practical suggestion: select the window before launch and leave it until the campaign report is final.
- Record the current settings in writing or with a screenshot.
- Schedule any change outside the campaign date range.
- Add the change date to the report note.
- After the change, do not compare the first week against the week before it as if the rules were the same.
A model change deserves even more care, because it rewrites history. After someone changes the model, previously shared reports may no longer match what GA4 now displays. Write the change date and the reason next to the old report.
Who may edit, and when data begins
- Changing attribution settings requires the Marketer role or higher at the property level.
- Data-driven attribution starts on November 1, 2021 and rules-based models start on June 14, 2021. Date ranges before a model's start date show partial data.
- The conversion management feature may not be available for every property.
If you revisit an old campaign, check those start dates before drawing a conclusion. A gap in a pre-2021 range can mean the model had not started yet, and not that measurement failed.
Mobile app sales need their own note. If the shop has an app, web conversions follow the Paid and organic channels choice while app conversions always follow Google paid channels. Combining them in one channel table mixes two credit rules. Practical suggestion: keep web and app conversions on separate rows and show the total only as information.
A decision rule for choosing the window
Practical suggestion: choose the window from the longest reasonable decision time you see in your own sales, not from an average. The help page does not say which window fits which kind of product, so the table here is a working method for your team.
Decision table: choosing a window
| Situation | Suggested approach | Reason |
|---|---|---|
| Low-value, quick purchase | 30 days for other key events | The decision is made within days, so old touches did not cause the sale |
| Expensive item that needs comparison | Keep the 90-day default | Weeks can pass between first touch and purchase |
| You measure new-user acquisition | 30 days on acquisition events, 7 days if needed | The first visit sits close to the last touch |
| Short seasonal campaign | Move the report date forward by the window length | A report closed on the final day misses late conversions |
The rule in use
- Estimate the gap between first touch and purchase in recent sales; if you have only a few observations, write it down as a range.
- Compare the long end of that range with the window options.
- Choose the shortest option that still covers the range.
- Write the choice and its date in the campaign calendar.
Planning the report calendar around the window
The longer the window, the longer after a campaign ends that conversions can still be linked back to it. Two reports solve this: a preliminary read when the campaign ends and a final read once the window has elapsed. Present the first as a preliminary number and take the second after the selected window has passed.
Table: a reporting calendar for the aquarium shop
| Stage | Action | Label on the report |
|---|---|---|
| Before launch | Record model, channel set and window | Settings record |
| Last day of the campaign | First summary of orders and channel split | Preliminary read |
| Weeks after the end | Refresh with the same settings and watch the change | Interim read |
| When the window has elapsed | Final pull for the campaign dates | Final read |
| After the final read | Note the gap between preliminary and final for the next campaign | Learning note |
Over a few campaigns this teaches you how late your ads show their effect. A large gap means the next report should close later. A small gap means the preliminary read is safe to use.
Diagnosing two reports that disagree
When two reports for the same campaign show different conversion counts, work through the causes from the cheapest check to the costliest. A settings difference is easier to confirm than a data error, so settings come first.
- Confirm both reports cover the same date range.
- Confirm both use the same channel set, Google paid channels or Paid and organic channels.
- Check whether the reporting attribution model changed between the two reports.
- Check whether the lookback window changed, and separate which report used which window.
- Compare the list of key events, because one report may count an extra event.
- If a gap remains, read the platform report's own definition; the gap may come from the source and not from GA4.
When the gap is genuine
If every setting matches and the numbers still differ, the cause sits outside the attribution screen. Treat that remaining gap as a measurement question for your analytics owner and record it in the report note instead of forcing the numbers to agree.
Worked example: aquarium starter kits and fish food
Example scenario: The aquarium shop sells two things through the same campaign landing page. Fish food is a quick, repeat purchase. A starter tank is a considered purchase. The property uses one set of attribution settings, so both products are measured with the same window. The marketer reads the report in three steps.
- Confirm the report shows the Paid and organic channels choice, because many tank buyers return through search.
- Note the 90-day window for purchases and the 30-day window for the first visit event.
- Split the purchase rows by product group, so quick fish food orders do not hide the slow tank orders.
With that note, the preliminary read for fish food can be trusted soon after the campaign, while the tank read waits for the window. If the shop compared this year's tank campaign with last year's, it would first confirm that both ranges were built with the same window. If the lookback window was changed in between, the comparison is not fair and the two periods should sit side by side with an explanation.
Table: how the aquarium shop reads each product group
| Product group | Decision speed | How to read the report |
|---|---|---|
| Fish food and water conditioner | Fast | Conversions land close to the click, so a short window gives the same picture |
| Filters and heaters | Medium | Shoppers compare specifications and visit more than once, so a preliminary read can fall short |
| Starter tanks and stands | Slow | The gap between first ad touch and purchase can approach the edge of the window, so wait for the final read |
Because the property has one set of attribution settings, anchor the window on the slowest product group and explain the others with a report note. Otherwise quick products look strong and expensive ones look as if the ad contributed nothing.
Sharing a report with a client, agency or publisher
Disputes start when three parties look at different numbers. Practical suggestion: end every campaign report with a four-line settings note.
- Reporting attribution model and selected channel set
- Lookback window for acquisition events and for other key events
- Date the window was last changed
- Date the report was pulled and whether it is a preliminary or a final read
The note lets someone who opens the report three months later read the numbers correctly. For the same habit applied to sponsored articles, see sponsored post reporting with UTM, GA4 and Search Console.
Each party's number answers a different question. The advertiser's GA4 number follows the Attribution settings chosen in the property. The agency's platform number follows the platform's own rules. A publisher's report focuses on impressions and clicks and carries no sales count. Writing down which number answers which question saves time in a review meeting.
Checklist before sending a post-campaign report
- Do the report start and end dates match the campaign calendar?
- Does the range reach back before the 2021 model start dates?
- Is the selected channel set written in the note?
- Are the windows for acquisition events and other key events known?
- Was the window changed during the campaign, and if so is the date recorded?
- Is the gap between preliminary and final read shown on its own row?
- Are app conversions, if any, separated from web conversions?
- Does the person who changed the settings have the Marketer role or higher?
Reading mistakes to avoid
Treating the window as the campaign length
A campaign that ran 14 days does not have a 14-day window. The window is a fixed setting chosen on the Attribution settings screen and is independent of the campaign calendar.
Confusing the date range with the window
The report date range selects which conversion dates enter the table. The window decides which touches can be credited for them. They work independently.
Expecting the platform report to match
An ad platform counts with its own definition. When you compare it with GA4 for the same period, a different number is expected. The window is only one part of the gap; consent settings, click counting and session definitions can contribute as well.
Reading a setting change as business performance
After you change a setting, the channel table can move. A model change recalculates history, so you will not see a sudden break in the chart, but old and new reports will disagree. A window change applies from the change date, so a gradual shift can appear afterward. Blaming that shift on the shop's real sales is mistaking a setting effect for a business effect.
Equating conversion count with revenue
The window decides which channel receives a conversion, not what the conversion is worth. A fish food order and a tank order count as the same key event. When you judge return on ad spend, look at server-confirmed order value, not at the event count.
A misreading: a shorter window deletes sales
Shortening the lookback window does not remove the order from the shop. It changes only whether an older touch can still be credited. Example scenario: the aquarium shop moves other key events from 90 days to 30, and starter tank orders whose first ad touch falls outside the new window drop out of the campaign row. The window change does not affect past data, so the old period and the new period were not built under the same rule.
Decision rule: when a count falls, check the date of the setting change before you check performance. Practical suggestion: compare only two periods that were produced under the same window. If you must show two periods built under different windows, print the window used under each one.
Frequently Asked Questions
What are the lookback window lengths in GA4?
According to the Google Analytics help page, acquisition key events (first_open, first_visit) default to 30 days with a 7-day option. All other key events default to 90 days with 30 and 60-day options.
Does changing the window rewrite old reports?
No. The help page says changes to the lookback window apply going forward. A change to the reporting attribution model applies to historical and future data.
Can an organic channel take credit for an ad-driven conversion?
With Paid and organic channels selected in Channels that can receive credit, conversions from organic sources are included. With Google paid channels, Analytics conversion reports show Google Ads conversions. The paid and organic option applies only to web conversions.
Are first click and linear models still selectable?
The help page states that first click, linear, time decay and position-based models are no longer available as of November 2023. It names data-driven and rules-based models.
Which roles may change attribution settings?
A user needs the Marketer role or higher at the property level. The conversion management feature may not be available for every property.
Should I pull the report the day the campaign ends?
Practical suggestion: pull a preliminary read at the end and label it, then prepare the final read once the selected window has passed. Conversions inside the window can still be linked to the campaign later.
Conclusion
Whether GA4 credits an ad conversion to your campaign depends on three settings together: the reporting attribution model, the channels that can receive credit and the key event lookback window. Acquisition events default to 30 days with a 7-day option, and other key events default to 90 days with 30 and 60-day options. A model change recalculates history, while a window change applies only going forward.
When you build the post campaign report, write the settings into the note, separate the preliminary and final reads, and record every change date. If you are planning placements and want creative, landing page and measurement set up together, you can explore ad spaces.
Post a comment
Please log in to leave a comment. Login