First, count the variants
Export one quarter of campaign data from every system that holds it: campaign names from each ad platform, campaign, source and medium values from the analytics property, and the campaign field from the CRM. Put them in one sheet and, for each real campaign, list every spelling it appears under. The number of distinct values divided by the number of real campaigns is a simple measure of the drift, and it gives a baseline to improve against.
Then read the variants for patterns. Differences in capital letters, hyphens and underscores point to typing habits. Values that are structured but in a different order point to a platform or a team with its own pattern. Values that look nothing like the rest usually come from an agency, a freelancer or a sending tool configured once and forgotten. Note for each pattern who created the campaigns concerned. The causes below are ordered roughly as these patterns usually appear.
If the main visible damage is visits landing in an "unassigned" or catch-all channel, the medium values are the issue, and the unassigned traffic diagnosis reads that symptom inside the analytics property. This page addresses the operational causes behind all of the drift.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Ad platforms, analytics and CRM each describe the same campaign differently | Whether a written list of campaign attributes and allowed values exists | No shared taxonomy across platforms and teams | Agree one short taxonomy with an owner and a change log |
| Typos, extra spaces and one-off values in campaign and source fields | How links are built: by hand, from a template or from a builder | Links typed by hand | Build every tagged link from a controlled builder |
| The same value appears in several capitalisations or with different separators | Variant counts that differ only in case or punctuation | Case, separators and spelling drifting | Adopt and enforce the case and separator rules |
| One platform's campaigns sort and split differently from the rest | Name structure by platform | Each ad platform named to its own pattern | Map every platform's pattern to the shared fields |
| Values that follow no internal pattern, clustered by supplier | Who created the campaigns with unparseable names | Agencies and freelancers using their own conventions | Put the taxonomy in every scope and brief |
| The convention exists on paper, but nobody checks it | Date of the last naming audit and the named owner | Nobody owns or audits the taxonomy | Name an owner; schedule audits with fix-by dates |
Fix, verify and prevent each cause
No taxonomy shared across platforms and teams
Without an agreed list of what a campaign name must say, each system ends up describing campaigns in its own terms, and joining them needs a person with a lookup table. The data operations and taxonomies guide treats the taxonomy as the shared language that lets a report join spend, sessions and pipeline.
Fix: agree a short list of attributes that reporting actually groups by, with the allowed values for each, and record where each attribute appears in every system. Keep it in one document with an owner and a change log.
Verify: take one new campaign and confirm that its values in the ad platform, the analytics property and the CRM can be joined without a mapping.
Prevent: add new values only through the owner, and remove attributes no report uses.
Links typed by hand
Every hand-typed link is a fresh chance for a misspelling, a stray space or a value nobody has used before. Google's documentation on URL builders describes the campaign parameters a link can carry and which are required for campaign data to be recorded, so a missing or mistyped field is not a cosmetic problem: it changes what the property receives. The UTM link request SOP routes link requests through one builder and a register.
Fix: retire hand-built links; issue every tagged link from a builder with drop-downs for each controlled field, and log each one.
Verify: the next month's export shows new campaign values only from the allowed list.
Prevent: make "link built from the builder" a launch check and reissue existing links from the register rather than rebuilding them.
Case, separators and spelling left to habit
Analytics tools treat values that differ only in case or punctuation as different values, so the same channel or campaign splits into several rows. The UTM and naming conventions guide sets the case, separator and character rules, and they are not repeated here.
Fix: adopt those rules, build them into the builder so it cannot produce a non-conforming value, and map existing variants to their correct value in reporting.
Verify: the variant count for values that differ only in case or separators falls to nothing for new campaigns.
Prevent: include a scan for case and separator variants in the monthly audit and trace each back to the person or tool that created it.
Each ad platform named to its own pattern
Platform specialists name campaigns in the structure that suits the platform's interface, so search, social and video campaigns end up with different field orders, abbreviations and separators. The Google Ads naming conventions guide shows how to align a platform's names with tracking values, including inserting the campaign name into the tracking template rather than typing it.
Fix: map each platform's naming fields to the shared taxonomy, keep platform-specific detail in labels or later fields, and use the platform's own substitution features where they exist so the tag carries the name automatically.
Verify: a report can split spend from every platform by the shared attributes without manual matching.
Prevent: require new campaign types and new platforms to have their naming pattern approved before the first campaign launches.
Agencies and freelancers with their own conventions
External partners usually arrive with their own naming habits, and unless the business's taxonomy is part of the scope, they will use them. The briefing and scope guidance in selecting, briefing and reviewing agencies is where expectations like this belong.
Fix: send the taxonomy and builder to every partner, add naming compliance to the scope and the onboarding checklist, and ask them to rename non-conforming campaigns at the next safe moment.
Verify: partner-created campaigns pass the next audit at the same rate as in-house ones.
Prevent: include naming compliance in partner reviews alongside results.
Nobody owns or audits the taxonomy
A convention without an owner drifts back to habit within months, because nobody has the authority to ask for a fix or to add a needed value properly. The centralised vs decentralised operations comparison places taxonomy and naming rules with the central operations function even where teams run their own campaigns.
Fix: name a role as owner, give it authority to approve values and request fixes, and schedule a monthly audit that lists new names that do not parse.
Verify: each audit produces a short record of exceptions found and fixed, and the exceptions fall month on month.
Prevent: report the audit result alongside the monthly pack, so drift is seen by the people who depend on the data.
A prevention routine for campaign naming
Consistent naming is kept, not achieved. At the point of creation, every campaign name and tagged link comes from the builder, using values from the taxonomy, and new values are requested from the owner rather than invented. At launch, the check compares the live names and tags with the work item and opens one link per platform to confirm the parameters arrive. Each month, the owner exports new campaign values from every system, flags anything that does not parse, and sends each exception back to its creator with a fix-by date. Each quarter, the owner reviews the taxonomy itself: attributes no report uses are removed, informal values that keep appearing are either added properly or their cause is fixed, and the variant count is compared with the baseline. When a new partner, platform or team starts, onboarding includes the taxonomy, the builder and the audit. Report the variant count alongside the headline KPIA metric tied directly to an objective. set, because every measure cut by campaign or channel, and every AttributionAssigning conversion credit to touchpoints. model applied to them, depends on the names joining up.
A travel company counts its campaign variants
A travel company exported one quarter of data and found 340 distinct campaign values across 2 ad platforms, the analytics property and the CRM, mapping to 96 real campaigns. Of the variants, 41% differed only in case or separator, 23% came from one social agency using its own structure, and most of the rest were hand-typed email campaign values. The marketing operations lead agreed a taxonomy of 5 attributes, put a builder in front of everyone creating links, added the taxonomy to the agency's scope and started a monthly audit. Three months later new campaigns produced 104 values for 92 real campaigns, and the monthly report no longer needed a day of matching.
Common mistakes
Cleaning the report every month instead of fixing where names are created. Writing a convention without a builder. Adding every attribute anyone can think of. Banning informal values without asking why they appeared. Renaming live campaigns without listing what depends on their names.
