Google Analytics Adds Custom Conversion Windows: Match Reporting to Your Sales Cycle

A sale can happen weeks after the first campaign visit. If the reporting window is shorter than that journey, an earlier interaction may sit outside the period considered for conversion credit.

Google Analytics' August 11 update adds custom integer lookback windows for conversions: one to 90 days for click-through conversions and one to 30 days for engaged-view conversions. That gives teams more flexibility to align the window with the business question.

A more flexible window does not identify a buyer, create another sale or prove that an earlier campaign caused the purchase. It changes which eligible interactions can be considered under the reporting method.

Write the question before choosing the number

Start with the conversion you intend to evaluate.

A completed appointment request, a paid order and an accepted sales opportunity represent different points in a relationship. Decide which outcome matters and where its record comes from.

Then ask what the report should help you decide. Are you reviewing credit across campaigns, comparing the timing of inquiries or evaluating the cost of acquiring a particular group?

Do not pick a longer window simply because a channel receives a larger number afterward. A useful window should have a reason you can explain before seeing the preferred result.

Connecting website activity with a practical decision begins with that definition. The report should follow the business question, with its limits visible.

Separate three different windows

Keep the conversion lookback, the report's date range and your outcome-review period distinct.

A conversion lookback considers eligible interactions before the conversion. The report date range determines which period you are inspecting under that report's rules. An outcome-review period follows a defined group forward from entry to see what happens.

Those windows can overlap without being interchangeable.

For example, a team may review August conversions while considering earlier campaign interactions. A separate pilot may follow contacts acquired during August for 30 days each. Both can be useful, but they describe different groups and timelines.

Write the rules beside the result. “August performance” is too vague when one person means conversions in August and another means everyone first acquired in August.

Follow a fictional 35-day journey

Consider a hypothetical sequence, not a customer result.

On day zero, a person clicks Campaign A. On day 28, there is an eligible email interaction. On day 35, the business records the defined purchase.

Under a 30-day lookback, the day-zero interaction is too old to fall inside that window. The day-28 interaction is seven days before the purchase and can fall inside it.

Under a 45-day lookback, the day-zero interaction can also be eligible for consideration. That does not mean Campaign A must receive all or any particular amount of credit. The attribution method, available data and other applicable rules still determine how credit is assigned.

The purchase remains one purchase. The person's identity has not become more certain because the window grew.

Inspect your own time to outcome

Use a small set of appropriate records to understand the timing your business can actually establish.

For each, record the starting event, conversion event, elapsed time and any gap in the connection. Keep the entry rule consistent. A first known inquiry is not the same starting point as a first observed page visit.

Include incomplete and uncertain cases in the explanation. A sample containing only fast, completed deals can give the team a narrow view of the process.

Do not turn a small sample into a precise universal buying cycle. Use it to ask whether the current window fits the decision and where a fuller review is needed.

Review the integrations joining campaign and sales records. Confirm what the records establish before describing the elapsed time as a complete customer journey.

Keep click-through and engaged-view definitions visible

The August update provides different allowed ranges for click-through and engaged-view conversions. Do not combine them under an unexplained single number.

Record the conversion type, chosen lookback and reason for the choice. Check which configuration applies to the particular conversion and linked account you use.

Avoid assuming that a change in one setting updates every conversion or every connected report in the same way. Confirm the behavior you need to rely on before using the result in a budget decision.

If someone asks why two reports disagree, compare definitions and settings before declaring one wrong. A different eligible interaction period can produce a different view without changing the underlying business transaction.

Do not use the window to repair missing identity

A longer lookback can only work with the evidence available to the reporting method. It does not establish a person behind an anonymous event or make an unsupported cross-device association reliable.

Keep observed activity, associated contact records and attributed credit on separate lines. An old page visit that falls within 45 days is not automatically a confirmed touch by the person who later purchased.

Review consent and collection choices when documenting that evidence. Some parts of the journey may remain unavailable or unsuitable for the use you propose.

Preserve the unknown category. An attribution setting should not become a reason to invent a complete personal history.

Compare settings without changing the business result

Before editing a live reporting configuration, save the current definition, selected windows, date range and relevant results through the process your business uses.

Decide how you will evaluate the proposed change. Confirm how it affects future reporting and what, if anything, changes in historical views. Do not assume every old export will update or that every past report is recalculated.

Keep business-system outcomes available for reconciliation. The number of paid orders or signed agreements should not be quietly rewritten because the credit assigned to channels moved.

Use measurement features to ask a precise question about observed evidence, configured rules and the result they produce. A better-looking chart needs an explanation before it becomes a performance claim.

Keep attribution separate from causation

Credit under a reporting rule is not the same as proof of incremental effect.

In the fictional example, adding the day-zero interaction to the eligible period tells you it can be considered. It does not establish whether the purchase would have happened without Campaign A.

Use careful language in the client or management report. “The selected method credits this campaign” describes the reporting decision. “This campaign created every credited sale” asks the evidence to establish more.

A claim about additional sales caused by a campaign needs an evaluation designed for that question. Changing the lookback alone does not provide it.

Leave a record the next analyst can understand

Write a short configuration note: conversion definition, record source, report date range, conversion type, lookback, attribution method, reason for the choice and date of the change.

Name the owner who can explain the configuration. Keep the earlier version available when comparing reports across periods.

Then check a few representative journeys against the rule. Include a shorter case, a longer case and a case with missing information. Confirm which questions the setting resolves and which remain open.

The useful outcome is a report aligned with a clear sales-cycle question. Choose the window for that purpose, keep the identity limits intact and explain any change in credited results before acting on it.