What does a campaign taxonomy audit reveal about attribution gaps?

An audit method for finding the links and labels that make campaign attribution incomplete or misleading.

What does a campaign taxonomy audit reveal about attribution gaps?

Question and scope

This protocol asks how much of a campaign's traffic can be attributed reliably using the team's own taxonomy. It examines the links the team published, the parameters those links carried, and the records that arrived in the CRM. It does not attempt to model multi-touch attribution or compare vendor platforms. The scope is the gap between what the team intended to measure and what the data can actually show.

The unit of analysis is one live campaign link. A link is in scope if it was published during the review window and points to a page the team owns. Links created by partners or affiliates may be included when the team controls the destination. The audit should record the link, its parameters, the landing page, and the resulting CRM record.

Define the taxonomy fields

List the fields the team requires on every campaign link, such as source, medium, campaign, and content. For each field, write the allowed values and the rule for choosing one. A field without an allowed-value list is an invitation to improvise. The audit should test against the written standard, not against the auditor's preference.

Record which fields are required and which are optional. A long list of optional fields often produces inconsistent data because some teams fill them and others do not. If a field is not used in a report, consider removing it. The taxonomy should be as small as the reporting questions allow.

Collect links from the campaign brief, the ad platform, the email builder, and the social scheduler. These sources often contain different versions of the same destination. Include links from at least two channels so the audit can compare conventions. A sample from one channel will not reveal cross-channel inconsistency.

Record the date the link was published and the person who built it when that information is available. Ownership helps when a link needs repair. If the source is unknown, mark it as unknown rather than guessing. The audit is a snapshot, and it should state the date of the snapshot clearly.

Test source and medium consistency

Check whether source and medium values follow the standard and whether they are used consistently across links. Look for capitalization differences, spaces, and synonyms that split one channel into several rows. A report that shows paid-social and paid_social as separate channels will undercount both. The fix is usually a value table, not a new analytics tool.

Compare the values with the platform's own naming. A link that says source=linkedin but medium=paid-search creates a contradiction that no report can resolve. Record the contradiction and the likely cause. The auditor should not silently correct the link without telling the owner.

Take a sample of clicks or submissions and trace them to the CRM record. Check whether the source and campaign fields match the link parameters. A mismatch may come from a redirect, a form that drops hidden fields, or a CRM integration that overwrites values. The link alone does not prove attribution works end to end.

Record the point where the value changed and the number of records affected. If the CRM overwrites the source with a default, the taxonomy may be correct and the integration wrong. The audit should distinguish a labeling problem from a data-flow problem. The repair belongs to the owner of that step.

Look for campaigns split across spellings

Group campaign names by similarity and count how many variants exist for the same initiative. Common variants include abbreviations, quarter labels, and channel names embedded in the campaign field. Each variant creates a separate row and makes the campaign look smaller than it is. List the variants and the canonical name.

Decide whether the variants can be merged in reporting without changing the source links. A mapping table is often safer than editing historical URLs. If the team edits live links, record the change and the date. Historical reports should remain reproducible.

Check the landing page against the promise

Open each sampled link and confirm the landing page matches the campaign message and the expected offer. A correct UTM on a generic homepage produces a confusing record even if the parameters are perfect. Record whether the page, the headline, and the form align with the campaign. This is a quality check, not a design review.

Check for redirects that drop parameters and for pages that load a different form than the one in the brief. A redirect can erase the campaign value before the form sees it. Record the final URL and the redirect chain. The owner should fix the redirect rather than adding a second set of parameters.

Count the links that cannot be assigned to a source, medium, and campaign using the standard. Report the count and the share of the sample, not only a percentage. A single unclassified link may be harmless, while a cluster suggests a missing convention. The count is a useful measure of taxonomy health.

Group the unclassified links by cause, such as missing campaign, unknown source, or conflicting medium. The grouping shows where to invest the repair effort. A long tail of one-off causes may not be worth fixing. Focus on the causes that affect the most links or the most valuable campaigns.

Separate missing data from wrong data

Missing data means a required field is absent, while wrong data means a field is present but inaccurate. The two need different repairs. Missing data is often a template problem, while wrong data is often a training or integration problem. Report them separately so the team does not apply one fix to both.

Check a small sample of wrong values against the campaign brief to confirm the error. Do not assume a value is wrong because it looks unusual. A campaign may legitimately use a partner name as the source. Record the evidence behind each finding. The audit should be defensible to the person who built the link.

Limits and sources of error

The audit sees the links the team knows about, and it may miss links created in ad platforms or by partners. Browser privacy settings and cookie consent can prevent some clicks from reaching the CRM, which limits any link-based measure. The audit cannot distinguish a lost click from a mislabeled one without additional evidence. Record the access limits and the known blind spots beside the result so the reader does not overstate coverage.

The sample may overrepresent recent campaigns or the channels the auditor can access. State the sampling method and the access limits. The result is a description of the sample, not a complete census. A team should repeat the audit before drawing conclusions about trends.

Decide the repair order

Rank the findings by the number of records affected and the value of the campaign. Fix the links that feed the most important reports first. Add a value table and a link template to prevent the same errors from returning. Repairing a few high-impact links is better than rewriting every historical URL.

Assign an owner and a date to each repair, and record the change in the taxonomy reference. Tell the campaign builders what changed and why. A repair without a communication will be undone by the next launch. The audit should end with a short, owned list.

Set a recurring audit

Repeat the audit after each major launch and once a quarter for steady campaigns. Use the same sample method so the results can be compared. Track the unclassified share and the number of campaign variants over time. A falling trend is evidence that the standard is working.

Keep the audit lightweight enough to finish in a few hours. A large audit that happens once is less useful than a small one that happens regularly. Store the results beside the taxonomy reference so the team can see the history. The goal is a reliable record, not a perfect one. Add the audit date and the auditor to the report so a later reader knows which version of the taxonomy was tested. Note the channels that were out of scope so the next audit can decide whether to include them.

Sources

  1. Advertising and Marketing Basics, Federal Trade Commission. Provides a reference for truthful and consistent marketing claims.
  2. CAN-SPAM Act: A Compliance Guide for Business, Federal Trade Commission. Outlines compliance considerations for commercial email campaigns.
  3. Business Trends and Outlook Survey, U.S. Census Bureau. Supplies business conditions context for campaign measurement.

Related GTM research