In plain English
Measure marketing by separating observed activity, enquiries, qualified leads, opportunities and confirmed sales. Preserve campaign and source details, but label attribution as direct, modelled, reported or unknown. Compare costs and outcomes over a defined period, review missing data and use the combined evidence to choose the next test rather than claiming perfect credit.
What does attribution actually describe?
Attribution assigns credit to ads, clicks or other touchpoints before an important action. It is a reporting model applied to available data, not a complete reconstruction of every influence on the buyer. Customers may encounter referrals, reviews, search, advertising and offline conversations across devices. Consent choices, browser restrictions and manual records can also break the observable chain.
Google Analytics lets eligible properties choose attribution settings for key-event reporting and conversion reporting. Different scopes and models answer different questions. Keep the selected model, lookback settings and report dimensions documented. When a stakeholder asks which channel caused the sale, explain what the report observed and how credit was assigned before presenting the number.
Sources for this section: Select attribution settings, Scopes of traffic-source dimensions.
Which evidence should sit beside source reports?
Report platform activity, website actions, enquiries, qualified leads, opportunities and confirmed sales as separate layers. Add cost and capacity where the business can verify them. A campaign may create fewer enquiries but more suitable conversations, while another may create demand that later appears as direct or organic traffic. The decision needs the full path, not one platform's preferred metric.
Use customer-reported discovery as supporting context, not a perfect replacement for technical source data. Sales notes can explain referral relationships, prior awareness and offline contact that analytics misses. Keep contradictions rather than forcing a single answer. A source field, an attribution model and a customer's memory can each be informative while describing different parts of the journey.
How should campaign tagging and unknown values be handled?
Use a controlled naming convention for campaign parameters across email, paid social and other tagged links. Google Analytics documents source, medium and campaign parameters for identifying referring campaigns. Record the convention, owner and examples so case differences and ad hoc names do not fragment reports. Avoid overwriting platform auto-tagging without understanding the effect on attribution.
Unknown should remain a visible category. Do not redistribute unattributed sales merely to make a report total look complete. Track the proportion and causes of missing source, qualification and outcome data, then assign owners to correct recoverable gaps. A report that exposes uncertainty helps improve operations; a report that hides it creates false confidence and weakens the next budget decision.
Sources for this section: Collect campaign data with custom URLs.
How does imperfect evidence still support a useful decision?
Write the decision question before reviewing the dashboard. The business may need to decide whether to continue a campaign, repair tracking, change the offer, improve follow-up or test a different market. Identify the evidence each choice would require and the confidence currently available. This prevents an attractive but irrelevant metric from becoming the conclusion.
End the review with a decision, owner, expected observable signal and review date. When attribution is weak, choose a bounded test that can improve it: consistent campaign naming, a lead-source prompt, a qualification field or a controlled landing page. The purpose of measurement is not to eliminate every uncertainty. It is to make the next commercial action more accountable and learnable.
Sources for this section: About key events, Select attribution settings.
Official references
- Select attribution settings (Google Analytics Help)
- Scopes of traffic-source dimensions (Google Analytics Help)
- Collect campaign data with custom URLs (Google Analytics Help)
- About key events (Google Analytics Help)