Email attribution connects a send, click, or journey touch to a later business result. ESP dashboards, product analytics, CRM reports, and billing systems often count the same result differently. The job is to pick a model, instrument links consistently, and explain which number is trusted for each decision.
last updated 2026-05-074 sections
section 01
Attribution models
No model is neutral. Last-click favors campaigns close to purchase. First-click favors acquisition. Linear and time-decay models spread credit but are harder to explain. Lifecycle teams should use one reporting model and label exceptions.
model
fit
risk
Last-click
Short conversion paths and campaign reporting.
Undervalues earlier lifecycle touches.
First-click
Acquisition source analysis.
Overcredits first contact.
Linear
Longer journeys with several touches.
Can overcredit low-value sends.
Time decay
Revenue journeys with many contacts.
Harder for teams to audit manually.
section 02
UTM naming
UTM tags should identify source, medium, campaign, content, and journey where needed. Naming should be boring and stable so reports can group campaigns without manual cleanup.
field
example
rule
utm_source
loops or customerio
Use the sending platform or owned property.
utm_medium
email
Keep constant for email.
utm_campaign
trial_conversion_q2
Name the program, not the subject line.
utm_content
variant_a_cta_top
Use for variant or placement.
section 03
Reconciling reports
The ESP knows delivery and clicks. Product analytics knows sessions and activation. Billing knows revenue. CRM knows account ownership. A reporting layer should join them instead of forcing one tool to answer every question.
okStore provider message ID and campaign ID with the user or event record.
okUse consistent account IDs for B2B reporting.
okFilter unsubscribe and preference-center clicks from conversion reports.
okDefine the revenue window before reporting lift.
section 04
Counting errors
Common errors include counting bot clicks, mixing user-level and account-level revenue, attributing renewal revenue to low-intent newsletters, and comparing ESP clicks to billing conversions without a join key.
okExclude known link scanners where tooling supports it.
okReport user and account attribution separately.
okMark transactional email separately from marketing journeys.
okKeep a plain-English note next to every revenue report explaining the model.