Marketing Data Dictionary
A marketing data dictionary is a documented reference that defines every field and metric used in marketing reporting: what each one means, how it is calculated, where it comes from, and which values are valid. It exists so that words like session, lead, and conversion carry one agreed meaning across analytics, ad platforms, and the CRM. Without one, two teams pull the same metric from different systems and neither number can be defended.
What a Marketing Data Dictionary Contains
One row per field or metric, and eight columns per row.
| Column | What it records | Example |
|---|---|---|
| Field name | Exact technical name as stored | utm_campaign |
| Business definition | Plain-language meaning, one sentence | The campaign that drove the click |
| Data type | String, integer, date, boolean, currency | String |
| Allowed values | Closed list or format rule | Lowercase, hyphen-separated |
| Source system | Where the value originates | GA4 |
| Calculation | Formula, for derived metrics | conversions / sessions |
| Owner | Who approves changes | Marketing ops lead |
| Effective date | When it took effect | 2026-01-15 |
The calculation column is what makes it useful. Two teams can agree conversion rate means “how often visitors convert” and still report different numbers, because one divides by sessions and the other by users.
Both dimension and metric fields belong here. Dimensions come from your campaign taxonomy, which decides which dimensions exist and which values are legal; the dictionary records what each means and where it is stored.
Marketing Data Dictionary Example
A working example for a small stack:
| Term | Definition | Source | Calculation |
|---|---|---|---|
| Session | A visit, ended by 30 minutes of inactivity | GA4 | Not derived |
| Users | Active users: an engaged session or a first session | GA4 | Not derived |
| Key event | A GA4 event flagged as a business outcome | GA4 | Not derived |
| MQL | Lead scoring 50+ that requested contact | HubSpot | Not derived |
| Conversion rate | Key events divided by sessions, same date range | Derived | key_events / sessions |
| ROAS | Revenue from paid campaigns divided by spend | Derived | revenue / ad_spend |
Three entries in that table exist because the platform definitions moved:
- GA4 standard reports label the Active users metric as “Users”. Total users is a separate metric. A dashboard that says “Users” without stating which one is ambiguous.
- Google renamed GA4 conversions to key events in March 2024. Google Ads kept the word conversion for its own metric, so the two platforms now use different words for related but distinct things.
- Universal Analytics started a new session when the campaign source changed mid-visit. GA4 does not, which is why any definition written before 2023 needs an effective date.
Data Dictionary vs Data Catalog
A business glossary is the authority for what a term means, a data dictionary for what a field holds, and a data catalog for what exists and where.
| Artifact | Scope | Built by | Marketing example |
|---|---|---|---|
| Business glossary | Business concepts, no technical detail | Business and data teams | “Qualified lead”, company-wide |
| Data dictionary | Fields and metrics: type, source, formula | Marketing ops and analytics | Every column in the campaign reporting table |
| Data catalog | An inventory of datasets and assets, with lineage | Data engineering | Every table the GA4 BigQuery export writes |
The distinction that matters is scope. A catalog tells you a table called campaign_performance exists and which pipeline populates it. The dictionary tells you its conversion_rate column divides by sessions, not users. Most marketing teams hit definition disagreements long before they struggle to locate data.
How to Build a Marketing Data Dictionary
- Start with the metrics in your reports. Open the dashboards leadership already reads and list every metric on them. Do not try to document the entire stack.
- Write the formula, not just the meaning. For every derived metric, record the numerator, the denominator, and the date range rule.
- Name the source system for each field. Where two systems hold the same metric, mark one as authoritative.
- Attach allowed values to every dimension. Free-text fields drift. linkutm’s rules feature restricts each field to its approved list at link creation, which keeps real data matching the documented values.
- Assign one owner per row. An unowned definition is one nobody can change or defend.
- Stamp an effective date and keep old versions. Historical numbers were produced under the old rule and should not be silently restated.
Review the dictionary whenever a platform ships a naming change or a new channel is added, not on a fixed calendar.
Common Data Dictionary Mistakes
- Defining terms without formulas. “Engagement rate: how engaged visitors are” is not a definition. It cannot be reproduced.
- Documenting every field in the warehouse. A 400-row dictionary nobody maintains is worse than a 30-row one that is current.
- No effective dates. Without them, a reader cannot tell whether a 2024 report used the current definition.
- Treating it as a one-time project. Platform vendors change definitions without asking. GA4’s conversion rename is the obvious case.
- Confusing definition drift with measurement gaps. When two platforms disagree, check the definitions first. Counting units, attribution windows, and collection loss cause a data discrepancy even when both sides define the metric identically.
Frequently Asked Questions
What is a data dictionary?
A data dictionary is a reference document that describes every field in a dataset: its name, meaning, data type, permitted values, source, and owner. In marketing it covers both dimensions (campaign, channel, market) and metrics (sessions, conversion rate, ROAS). Its purpose is reproducibility, so two people calculating the same metric arrive at the same number.
What is the difference between a data dictionary and a data catalog?
A data dictionary documents fields and metrics in depth: type, formula, valid values, and owner. A data catalog inventories which datasets exist across systems, with their lineage, and is usually populated automatically. The dictionary answers “what does this column mean”, the catalog answers “where does this data live”. Most organizations build a dictionary first.
What should a data dictionary template include?
Eight columns cover most marketing use cases: field name, business definition, data type, allowed values, source system, calculation, owner, and effective date. Calculation and owner are the columns teams most often omit and most often need. Start with the metrics already on your dashboards.
Why do marketing teams need documented data definitions?
Because the same word means different things in different tools. GA4 reports Active users under the label “Users”, Google Ads counts conversions differently from GA4 key events, and CRM lead stages are defined per company. Written definitions turn those differences into known facts rather than recurring arguments.
To keep tagged links matching your documented field values, build them with the free UTM builder at linkutm.