Acquisition, UTMs, and attribution in plain English

Source, medium, and campaign do different jobs

Source names where the visit came from, such as google, bing, or newsletter. Medium describes the general type of marketing or link, such as organic, cpc, email, or referral. Campaign identifies your specific initiative, such as spring_launch or trial_week.

GA4 classifies many visits automatically, but your own links often need campaign parameters called UTMs. Add utm_source, utm_medium, and utm_campaign to a destination URL. Use utm_content or utm_term only when they answer a real question, such as which creative or paid keyword was involved.

https://example.com/pricing
  ?utm_source=newsletter
  &utm_medium=email
  &utm_campaign=april_product_update
  &utm_content=header_cta

Consistency is more important than cleverness

Email, email_newsletter, and Newsletter may split into separate rows or channels depending on how they are used. Lowercase values, underscores instead of spaces, and a shared naming guide make campaign reporting much easier to read.

Never add UTM parameters to ordinary internal links on your own site. Doing so can start a new attributed session and overwrite the campaign that originally brought the visitor. Use internal event parameters for internal promotions and navigation instead.

Attribution assigns credit; it does not explain everything

A user may discover you through social, return through search, and finally buy after an email. Attribution rules decide how credit is distributed across those touchpoints. GA4 also has user-scoped, session-scoped, and event-scoped acquisition dimensions, so similarly named fields can answer different questions.

Data-driven attribution uses observed paths and modeling to distribute credit, but no model can recover every untracked or cross-device interaction. Treat attribution as a useful decision aid. Compare models, examine the path, and avoid saying that one channel “caused” every assigned conversion.

Understand channels, referrals, direct traffic, and manual tagging

GA4 groups acquisition data into channels using available source, medium, campaign, and referral information. Automatic classification handles many common sources, while campaign parameters give you more control over links you create. The channel group is a derived label, so inconsistent source and medium values can produce surprising classifications.

Direct is not a person typing your URL by definition. It is a bucket for visits where usable acquisition information is absent or not attributable to another recognized source. Missing referral data, untagged links, redirects, browser behavior, and document-to-web journeys can all contribute. Treat Direct as a signal to investigate measurement as well as behavior.

Referral exclusions are appropriate for specific technical cases, such as some payment gateways, but they should not be used to hide unwanted analysis rows. An exclusion changes how returning referrals are handled. Document the business reason and test the full journey before adding one.

Create a UTM governance system instead of a one-off URL

A useful naming guide defines allowed values, ownership, examples, and forbidden patterns. Keep utm_campaign stable across every placement in the same initiative. Use utm_content for meaningful creative or placement differences, and use utm_term where its role in your search or audience workflow is clearly defined.

Do not put personal information, internal campaign codes nobody understands, or constantly changing values into campaign parameters. A campaign name should remain readable months later. Store the final URLs in a shared campaign register so reporting can be reconciled with spend and launch dates.

Test the final rendered link, not just the builder output. Link shorteners, social platforms, email service providers, and redirects can add, remove, or alter parameters. Confirm that the landing page preserves the expected query string and that your site does not accidentally leak sensitive values into analytics page locations.

Choose the acquisition scope before making a channel claim

First-user source / medium describes the acquisition associated with the user’s first measured visit. Session source / medium describes the acquisition associated with the current session. Event-scoped attribution fields support event-level credit in the relevant reporting context. These fields can legitimately disagree.

Imagine a user acquired by paid social who returns through email and purchases. A first-user report highlights paid social’s role in acquisition. A session report highlights the email visit. An event attribution view distributes or assigns credit according to the selected reporting logic. None of these views makes the others false.

Write the scope into your chart title. “Key events by session source / medium” is clearer than “Conversions by channel.” When stakeholders compare dashboards, check that they are not unknowingly comparing first-user acquisition with session acquisition.

Evaluate a campaign with a complete evidence pack

A campaign report should connect delivery, arrival, behavior, and outcome. At minimum, record the tagged URLs, launch dates, audience, spend source if applicable, landing page, sessions, key actions, and any measurement limitations. Looking only at GA4 sessions can miss whether the campaign was actually delivered as planned.

Compare both volume and efficiency. A campaign with fewer sessions may produce more qualified leads per session, while a high-volume campaign may create little downstream value. Use absolute counts to understand scale and rates to understand efficiency, then inspect the landing page and audience mix for alternative explanations.

Attribution can help allocate credit, but incrementality requires a different standard of evidence. If the business question is “What extra sales did this campaign cause?”, consider experiments, holdouts, or other causal designs where feasible. Do not answer an incrementality question with an attribution table alone.

Worked example: evaluate a webinar campaign across email and social

A training company promotes one webinar through a customer newsletter and two paid social creatives. All links share utm_campaign=autumn_ga4_webinar, while source, medium, and content identify the sender, channel type, and creative. The team uses one campaign register containing final URLs, launch dates, audience, spend, and landing page.

Social generates more sessions, but email generates more registrations per session. Rather than declaring email universally better, the analyst checks audience intent: existing subscribers may already trust the company. Paid social may introduce new users who return later through another channel, so first-user and return-session analysis adds useful context.

If leadership asks which channel caused incremental registrations, attribution is not enough. A feasible holdout or experiment may be needed. The campaign report separates delivery, observed behavior, assigned credit, and causal uncertainty.

Attribution interpretation checks before budget recommendations

Attribution outputs are sensitive to identity, available touchpoints, consent, lookback rules, and conversion definition. A channel with little assigned credit is not necessarily a channel with no influence, and a channel with substantial credit is not automatically incremental.

A responsible recommendation explains which model was used, which interactions were observable, and what decision the model can reasonably support. For high-stakes budget allocation, compare evidence rather than switching models until one confirms the preferred story.