Attribution Window Settings Audit for Small Teams

TL;DR: Audit attribution windows by outcome, interaction type, and reporting date. Keep bidding settings separate from comparison reports, record approved exceptions, and use CRM outcomes to judge whether a budget decision is justified.

Platform settings reviewed September 8, 2026. The publication date is December 7, 2025; the dated updates below describe later platform changes.

What is an attribution window, and what should this audit decide?

An attribution window is the time during which an interaction can qualify for credit toward a later outcome. A settings audit records those rules across your ad accounts, analytics, and customer relationship management system, or CRM. Its output is an approved measurement policy: which outcome counts, which interactions qualify, and which report supports each decision.

The first decision is whether two reports count comparable opportunities. Matching a number of days helps only when the outcome, interaction type, and date basis also match. A purchase after a click, a lead after a video view, and a deal closed after a sales call are different observations.

Name the shared worksheet Attribution Window Settings Audit: Align Google Ads, Meta, GA4, and Your CRM. For a shared-drive search label, use “attribution window settings google ads vs meta”. Give one person responsibility for keeping it current. Tie the work to a real decision, such as approving next month's paid-search budget, using the same discipline as an automation ROI calculation.

Use the audit in these situations:

E-commerce: the same order appears in search and social reports, and the owner wants to understand each platform's claim before reallocating spend.
Home services: calls arrive quickly, but scheduled jobs close later; the team needs separate lead and revenue windows.
B2B services: a qualified lead converts into a deal after several weeks, beyond the short window used to compare lead acquisition.
Subscription businesses: new-customer purchases and renewals share an event name even though they support different budget decisions.
Agency handoffs: a new operator inherits mixed settings, saved reports, and undocumented exceptions across accounts.

Do not start by changing every platform to seven days. First distinguish the window, which determines eligibility, from the attribution model, which divides credit among eligible interactions. Our GA4 model comparison guide covers the latter decision.

How do you check the Google Ads attribution window?

Check the window on each conversion action: open Goals → Conversions → Summary, select the action, and inspect Edit settings. Record click-through, engaged-view, and view-through windows separately, along with the conversion source and whether the action is used for bidding. Capture the current values before saving anything.

Google's conversion-window documentation provides these reference points:

Google Ads default click-through window is 30 days when not customized; available duration depends on conversion source.
Google Ads lists default engaged-view and view-through windows of 3 days and 1 day, respectively; audit each actual action.

A default is a starting reference, not evidence of your account's configuration. Native website actions, imported outcomes, calls, and app actions need separate register rows. Google also states that Smart Bidding uses the chosen conversion window, so a window edit can change the signal used to optimize bids. Treat that edit as an operating change with an owner.

There is another control to label correctly. The lookback selector inside an attribution report is separate from the conversion-action window. Use Path metrics to examine conversion delay, recording whether it measures from the first or last interaction. A chart already limited by a short collection window cannot prove that later outcomes never occur.

For outcome-date comparisons, Google documents conversion-time columns, including Conversions (by conv. time) and All conv. (by conv. time). Select the intended action and metric scope rather than comparing an unlabeled account total. Record the account time zone and export time as well.

How should you audit Meta attribution settings?

Meta attribution settings define which ad interactions and time spans can receive conversion credit; record the setting shown for each relevant ad set and outcome. Keep the ad set's delivery setting separate from any comparison columns used in reporting. A report view is evidence for a decision, not automatic approval to edit delivery.

In Ads Manager, inspect the selected ad set's conversion or performance settings and copy the attribution label exactly as displayed. Where Compare attribution settings is available in the columns controls, save separate click and view comparisons. Record unavailable options as unavailable; account type, objective, and rollout can affect what you see.

For example, an account may show 7-day click alongside 1-day view. That means a short view-based claim belongs in a different category from a click-based claim. Do not assume that one combined result means every conversion followed a site visit, or that the same options exist for every campaign. Avoid adding overlapping window columns: a seven-day click total generally includes the first day too.

Definitions can change even if the displayed number of days stays the same. In a first-hand announcement about Meta's attribution update, Alex Schultz described a move to link-click eligibility for website and in-store click-through conversions, with social actions grouped under engage-through attribution. The announcement described a rollout; it does not verify the state of your account.

Save the interaction definition and date with every baseline. If an older export includes engagement under a click label, annotate that break before comparing it with a newer export. Do not equate Meta's engage-through label with Google's engaged-view definition merely because both mention engagement.

Which GA4 lookback window are you changing?

GA4 has separate settings for key-event reporting and conversions shared with advertising tools. A seven-day option can be valid for an acquisition key event or a supported conversion setting without being valid for every key-event report. Identify the object and screen before choosing a duration.

For key-event reporting, inspect Admin → Data display → Events → Attribution settings. Google's key-event lookback guide states: GA4 key-event lookback defaults are 30 days for first_open/first_visit (7-day alternative) and 90 days for other key events (30/60 alternatives).

Here, first_open concerns first app use and first_visit concerns a first website visit. A purchase or lead key event belongs in the other group. Record the group explicitly instead of writing a single “GA4 window” row.

There is a later change to distinguish from that property setting. According to Google's release notes: On August 11, 2026, Analytics conversion management added custom click-through integer windows of 1-90 days and engaged-view windows of 1-30 days. The configuration path is Advertising → Conversion management → more options → Settings, with management also available through the linked Google Ads interface.

GA4 setting being audited	What to record	Main mistake to avoid
Key-event lookback	Acquisition versus other event group, duration, property	Assuming a conversion-management option rewrites this property setting
Conversion window	Exact conversion, click or engaged view, duration, linked account	Treating the shared conversion as a separate independent signal in every system
Attribution model	Model and report dimension scope	Calling a credit-allocation change a window change
Eligible channels	Google paid or paid and organic, where applicable	Expecting a Google-only view to equal a cross-channel report

Google's settings guide notes that conversion-management availability can vary by property. It also distinguishes event-scoped model changes from session and first-user dimensions. If a control is absent, record the limitation and use a clearly labeled export comparison; do not claim that a native report uses a window it cannot select.

A seven-step window audit and approval workflow

Build one settings register, one report definition, and one change record before touching live optimization. The minimum useful audit names the outcome, preserves the baseline, and explains how the team will judge the change. The following sequence is an operating recommendation, not a platform requirement.

1. Pick one business outcome

Start with completed purchases, qualified leads, or closed-won deals. Write down exclusions such as test records, duplicate orders, renewals, cancellations, and refunds. A website form event and a CRM qualified lead can both be useful, but they need separate rows and names.

2. Copy the current settings into a register

Create one row per action, ad set, key-event group, or CRM report. Save screenshots or exported settings with the audit date. The sample entries below illustrate fields to capture; they are not recommended defaults.

Register field	Example entry	Why it matters
System and object ID	Google Ads, purchase action 123	Makes the setting findable without guessing from names
Outcome and counting rule	Completed order, one per order ID	Separates orders from event fires
Eligible interaction and window	Click, 30 days	Defines the claim's time limit
View or engagement treatment	Separate report column	Prevents a view claim from masquerading as a site visit
Date basis and time zone	Outcome date, America/New_York	Gives exports a shared calendar boundary
Attribution model and channels	Named model, eligible channels	Separates credit rules from elapsed time
Setting versus report view	Live optimization / comparison only	Prevents an analyst's view from silently changing bids
CRM outcome anchor	Lead created / qualified / closed won	Shows which stage ends the measured delay
Owner and evidence	Paid-media lead, dated screenshot	Creates a checkable baseline
Approved exception	Longer revenue cohort; review date	Keeps an intentional mismatch from becoming an incident

3. Measure the delay from the right starting point

A cohort is a group of records with a shared starting period, such as leads created in January. For consented, joined records, calculate elapsed time from an eligible interaction to the selected outcome. Keep click-to-lead, lead-to-qualified, and lead-to-close delays separate. Choose a cohort old enough to observe the intended window plus normal reporting or import delay; recent clicks have not had their full chance to convert.

Mark missing timestamps as unknown. Do not replace them with upload time, infer them from the current CRM stage, or treat missing observations as same-day conversions. If identifiers or events fail basic checks, use the conversion tracking QA worksheet before interpreting the lag distribution.

4. Define the CRM reporting policy

The CRM does not have one universal ad-attribution window. Define a report's outcome date, eligible touchpoints, joins, and revenue basis; distinguish closed-won amount from collected cash. Preserve source history rather than overwriting it to make an ad dashboard agree.

For example, HubSpot's report documentation uses contact-create dates for contact attribution, deal-create dates for deal attribution, and deal-close dates for revenue attribution. A date filter selects outcomes; it does not by itself establish a click lookback rule. If you want a custom seven-day click cohort, calculate that eligibility from available timestamps in your worksheet and label it as your rule.

5. Save a comparable reporting view

Choose one outcome, interaction definition, date basis, and time zone. Set a maturity cutoff: the latest starting date whose records have had enough time to reach the outcome. Where supported, compare click-only windows without editing live settings. Where the tools cannot express the same rule, retain each platform’s own report and compare only the joined records whose eligibility you can calculate.

For example, a click on day 0 followed by a lead on day 5 belongs inside a seven-day click comparison. A lead on day 10 does not. A day-5 lead that closes on day 40 remains a valid lead outcome and a later CRM sale; the close does not retroactively become a seven-day purchase. Use actual timestamps for boundary cases, and label uncertain platform boundary semantics instead of rounding them into a false match.

Once the definitions are set, use the revenue reconciliation worksheet for remaining transaction-level differences. Platform totals still need not equal each other or the CRM because identity coverage, modeling, and eligible channels can differ.

6. Approve a setting change separately

Write the current value, proposed value, affected objects, expected signal change, owner, approver, and review date. Name the budget decision this supports. Keep model, counting-rule, and campaign-budget changes separate so the team can interpret what happened.

Use an explicit status: keep, change with approval, or blocked by missing evidence. If the expected loss of eligible outcomes would leave too little bidding signal, pause the change and review the outcome choice. A longer CRM sales cycle may justify retaining a qualified-lead signal while reporting revenue over a longer cohort.

7. Verify, monitor, and roll back prospectively

Read back the saved settings and confirm the intended objects changed. Compare the next mature cohort with the baseline, annotate expected eligibility differences, and investigate unexplained shifts. Set the review cadence around your conversion cycle, with an extra check after imports, campaign handoffs, or platform definition changes.

If an export fails or looks incomplete, stop the comparison and retry it before approving a change. If two operators propose edits, designate one owner and reject stale approvals against a changed baseline. A rollback restores the approved setting for future measurement; preserve the intervening period's exports and annotations rather than promising to recreate lost history.

Operator composite: a budget argument becomes a policy

This illustrative operator composite shows a settings audit at a small home-services company; it is not a public customer claim. All counts, costs, and time savings below are invented planning inputs. The useful result is an explainable reporting policy, not proof that a new window creates customers.

The company spends an assumed $12,000 per month across search and social. Its manager compares 110 Google Ads lead credits, 90 Meta lead credits, and 140 valid CRM leads. Each report carries a different date or interaction rule, yet the budget meeting treats 200 platform credits as 200 distinct leads. The numbers do not support that interpretation.

The operator uses Google Ads, Ads Manager, GA4, a CRM export, and a spreadsheet. They inventory each lead action, save the settings, and agree that the shared comparison ends at a valid lead submission. Closed jobs remain in a separate CRM revenue cohort because a booking can take several weeks.

The first worksheet is wrong: a developer supplied upload time as the outcome timestamp. That moves late imports into the current week. The operator stops the comparison, obtains the original lead-created timestamps, and marks records without usable evidence as unknown rather than inventing a date.

The corrected worksheet identifies 80 CRM leads with a qualifying Google click and 55 with a qualifying Meta click inside the illustrative seven-day rule. Twenty belong to both sets. The union is therefore 80 + 55 - 20 = 115 unique leads; the other 25 of the 140 CRM leads remain outside those observed sets. These are custom joined-record counts, not a promise that native platform reports will show 80 and 55.

The owner keeps the live bidding settings pending a separate test. The new report shows the 115-lead union, its 20-lead overlap, and the 25 other or unclassified leads, while retaining each platform's native view. No increase in leads or revenue is claimed. An apparent discrepancy has become a documented limitation the team can discuss without changing the underlying customer records.

Loaded labor means pay plus employer benefits and related employment costs. Assume the audit takes six hours at a loaded labor cost of $75 per hour: $450 upfront. If clearer reporting later saves two hours per month but review takes half an hour, the net capacity value is (2 - 0.5) × $75 = $112.50 per month. Simple payback is $450 ÷ $112.50 = 4 months. This estimate depends on actually reclaiming useful time; salaried capacity is not automatically cash savings.

Settings audit costs and ROI

Budget for staff time to collect settings, validate dates, and maintain the policy. A settings-only audit can use tools you already have, while repairs to tracking or CRM joins add separate work. The USD table below uses assumed hours and loaded labor rates to make the estimate reproducible; it is not vendor pricing or a service quote.

Work item	Illustrative hours	Assumed USD hourly cost	Illustrative cost
Inventory actions and save settings	1-2	$75-$125	$75-$250
Validate outcome dates and compare cohorts	2-4	$75-$125	$150-$500
Approve policy and document exceptions	1-2	$75-$125	$75-$250
Initial audit total	4-8	$75-$125	$300-$1,000
Recurring review, per cycle	0.5-1	$75-$125	$37.50-$125

These are planning ranges for a small, already functioning stack. Missing access, unreliable identifiers, a long sales cycle, or many conversion actions increase the work. Check current software pricing before upgrading: HubSpot's documentation places deal and revenue attribution reports in Marketing Hub Enterprise, but a settings register does not require buying that report builder.

Measure value as verified time reclaimed, fewer repeated investigations, or a separately tested improvement in a business decision. Do not count a larger attributed-revenue total as new revenue. Use the ROI calculator to test your labor assumptions, and the ROAS leak calculator to explore suspected wasted ad spend without treating disputed attribution as confirmed waste. ROAS means revenue divided by ad spend; its numerator still needs a clear definition.

When is a window audit not enough?

A window audit cannot repair missing events, broken identities, or an undefined business outcome. It also cannot prove that advertising caused a purchase that would not otherwise have happened. Fix collection problems first, and use an appropriate experiment when the decision depends on incremental sales.

Do not make a live change when the only available cohort is still immature. A business with very few completed outcomes may need a longer observation period rather than a more elaborate policy. Likewise, a one-channel team with a stable definition may need only a short register and quarterly review, not a new attribution platform.

Avoid these five mistakes:

Forcing one duration everywhere. Shared reporting definitions can coexist with documented platform-specific optimization settings.
Adding overlapping claims. A customer can qualify in two platforms or nested time windows; those totals are not unique customer counts.
Using a date filter as a lookback rule. “Deals closed this month” says nothing about how long before closing a click may qualify.
Changing windows, models, and budgets together. The resulting comparison cannot isolate the setting's effect.
Treating restored settings as restored history. Keep snapshots, effective dates, and exceptions for the period between changes.

FAQ

The short answers below resolve the terms and edge cases that most often derail a settings audit. Use them beside the register so each operator applies the same definition.

What does attribution window mean in a purchase example?

It means the allowed delay between an eligible interaction and the purchase being credited. If a shopper clicks and buys five days later, that interaction fits a seven-day click window. A matching window makes the interaction eligible; the attribution model still determines whether and how much credit it receives.

What is the Google Ads attribution window for an imported lead?

It is the configured eligibility period for that conversion action, subject to the source's supported settings. Record it separately from any import deadline: a timely upload and an eligible ad-to-outcome delay are different checks. Do not substitute the upload timestamp for the lead's actual occurrence time.

What is the attribution window in Ads Manager?

It is the time rule attached to the relevant conversion credit, such as a displayed click or view window. Read the ad set's actual setting and the report's selected columns together. For an audit, an exact copied label is more useful than an assumed account-wide default.

Should Google Ads, Meta, GA4, and the CRM use the same window?

Use a shared rule for a comparison only where its inputs and controls support that rule. Keep intentional differences documented for optimization and longer CRM outcomes. Forcing every screen to display the same duration can hide real business delays without making attribution identical.

Does changing an attribution window rewrite history?

Do not assume it does. Google Ads applies window changes going forward and does not recover previously excluded conversions simply because you lengthen the window. GA4 also describes key-event lookback changes as forward-looking; reporting-model changes have different behavior. Check each control's documented effect and retain exports.

Can GA4 use a seven-day lookback for purchases?

Yes, for a supported conversion in conversion management; the separate non-acquisition key-event setting has different choices. If two GA4 reports disagree, first copy their conversion or key-event names and setting paths into the register. Confirm that they measure the same outcome before investigating the totals.

How should the CRM handle a lead that closes after the ad window?

Keep the lead and its later close in the CRM with their real timestamps. Report the lead under the acquisition rule and the sale under a defined revenue cohort. Preserve the relationship between them without claiming the later sale fell inside a shorter click-to-purchase window.

How often should a small team repeat this audit?

Review the register before a material budget decision and after changing conversions, integrations, or campaign ownership. A monthly or quarterly routine can work for a stable account, depending on its sales cycle. Increase attention when a platform changes the meaning of an interaction, even if nobody edited the duration.

Answer clarity notes

These notes distinguish verified platform behavior from the planning policy and example used in this guide.

Dates: the article's publication stamp is December 7, 2025. Platform guidance was reviewed September 8, 2026. August 11, 2026 refers specifically to Google's conversion-management update; it does not describe the separate property key-event lookback controls.
Evidence: Google and HubSpot documentation supports the linked settings facts. The Meta update is sourced to Alex Schultz's first-hand announcement; it is not proof of rollout completion in an individual account. Record the controls and definitions actually available to you.
Costs and ROI: USD ranges are illustrative labor estimates, not vendor fees or That'sGonnaHelp quotes. The operator composite is not a public customer claim. Its counts, rates, savings, and four-month payback are assumptions, not measured customer results. These examples and ranges are not guarantees.
Recommendations: the register, approval statuses, comparison cohort, and review cadence are operating guidance. They do not establish a universal best attribution window or prove advertising incrementality.
Scope: this guide supports US SMB reporting and workflow decisions. It is not legal, financial, tax, privacy, compliance, or platform-policy advice. Keep access to customer-level exports limited to the people who need it.
Do not infer: matching durations does not equal matching credit. A custom joined-record comparison is not a native platform report, and a change in attributed revenue is not evidence of a change in real sales.

Sources

These primary sources support the specific platform behaviors linked in the article. The worksheet design and illustrative labor math are That'sGonnaHelp planning guidance.

That'sGonnaHelp can help your team document the settings, trace the CRM outcome dates, and leave a review process you can run before the next budget decision.