What Is Metadata? Definition, Types, and Real Examples
Metadata is data that describes other data. Here are the three main types, real examples, and why it decides whether your reporting works.
Metadata is data that describes other data. A photograph’s metadata records when it was taken, on what camera and at what resolution; a marketing campaign’s metadata records which brand it belongs to, which channel it ran on and which audience it targeted.
There are three main types. Descriptive metadata identifies content: titles, keywords, campaign names. Structural metadata says how pieces fit together, such as which asset belongs to which campaign. Administrative metadata governs use: rights, ownership, expiry. In marketing the descriptive layer is where money is won or lost, because it is what analytics tools group your reporting by.
What is metadata?
The definition is settled; what trips people up is that metadata is a relative category, not a fixed one. Merriam-Webster gives the standard formulation — data that provides information about other data (Merriam-Webster, “metadata”, accessed 2026-09-10) — and every serious source agrees with it. The word dates to the 1960s and entered general dictionaries as computing made the distinction unavoidable.
So the definition is easy. What makes metadata confusing in practice is that it is relative rather than absolute. The same field can be data in one system and metadata in another. A campaign name is metadata when you are analyzing click-through rates, because it describes the thing being measured. It is data when you are auditing the naming convention itself. Nothing about the field changes; what changes is what you are asking about.
This is why “is X metadata?” is usually the wrong question. The useful question is what the metadata is for, because that determines who creates it, when, and how much consistency it needs.
The three main types of metadata
Metadata divides into descriptive, structural and administrative. The taxonomy comes from library and information science, where it was developed to catalog collections, and it transfers cleanly to digital content (Carnegie Mellon University Libraries, “Metadata Guide”, accessed 2026-09-10).
| Type | What it does | Library example | Marketing example |
|---|---|---|---|
| Descriptive | Identifies and describes content so it can be found | Title, author, subject headings | Campaign name, channel, audience, brand |
| Structural | Records how parts relate to a whole | Chapter order within a volume | Which creative belongs to which campaign |
| Administrative | Governs handling, rights and lifecycle | Acquisition date, copyright status | Usage rights, expiry date, approving team |
Administrative metadata is sometimes split further into rights, preservation and technical subtypes. For most working purposes the three-way split is enough.
The proportions differ sharply between the library case and the marketing case. A catalog is mostly descriptive and administrative, created once by a trained cataloger. Marketing content is heavily descriptive, created continuously by many people under deadline, most of whom have never seen a metadata standard and are not trying to build one.
Examples of metadata
The clearest examples are a photo’s EXIF data, a document’s author and date, and a campaign’s channel and audience fields.
- A photograph. EXIF metadata embedded by the camera records the date, device, exposure, lens and often GPS coordinates. None of it is visible in the image (Wikipedia, “Metadata”, accessed 2026-09-10).
- A document. Author, created and modified dates, word count, revision history. Visible in file properties, invisible on the page.
- A web page. The title tag, meta description and structured-data markup that tell search engines and social platforms what the page is. This is the layer people mean by web metadata: it never appears in the visible page, but it determines the headline in a search result, the preview card in a shared link, and increasingly whether a generative engine can identify the page’s subject at all.
- An email. Sender, recipient, timestamp and routing headers, all separate from the message body.
- A marketing campaign. Channel, audience, region, brand, creative variant and the tracking parameters that carry those values into analytics.
The first four are generated by software. The fifth is typed by a person, and that is the entire difference.
Photo and image metadata, in depth: Photo metadata explained — EXIF, IPTC and what travels with an image file.
Metadata for marketing content
Marketing metadata describes campaigns, creative and channels, and unlike a library catalog it is written by dozens of people, continuously.
This is the case the standard definitions do not cover. Every ranking explainer on this topic treats metadata as a cataloging artifact: something applied deliberately, by someone whose job is to apply it, to an object that already exists and will not change. Marketing metadata is none of those things. It is applied in passing, by people whose job is to launch a campaign, to an object that is being created at that moment and may be duplicated twelve times before the week is out.
Three consequences follow.
The failure mode is inconsistency, not absence. A missing catalog record is obvious. Marketing metadata is almost never missing, because the platform requires the field. It is filled in differently by different people, which looks like completeness and behaves like corruption. Every audit that counts populated fields will report this data as healthy, which is why the problem usually goes undetected until somebody queries across two systems and the join fails.
Volume makes manual correction unsustainable. Manual tagging and metadata entry at scale is one of the most common problems enterprise teams raise with us, across 49 accounts. It is rarely framed as a tagging problem. It is framed as an analyst spending their week reconciling names instead of doing analysis.
Creative and asset metadata has its own gravity. Governance and discoverability of metadata on creative assets is a distinct, equally common problem, raised across 49 accounts, and it usually sits with a different team and a different system (a DAM) from campaign metadata. Organizations routinely solve one and not the other, then discover the two vocabularies do not join.
Why inconsistent metadata breaks reporting
Analytics tools group by exact string, so two spellings of one channel become two rows.
This is the mechanical reason metadata quality is a reporting problem rather than a tidiness problem. Grouping is a string-match operation. Paid Social, paid_social and paid social are three distinct values, and every tool that aggregates by them will produce three rows showing a third of the volume each.
Nobody loses a quarter because metadata was missing. They lose it because the same thing was described three different ways.
The damage compounds in a way that is easy to miss. A fragmented channel does not look like an error; it looks like underperformance. Teams reallocate budget away from channels that appear weak, when what is actually weak is the grouping. By the time someone traces it back, decisions have already been made on the fragmented numbers, and the historical record cannot be fully repaired because the original values are what they are.
Colgate-Palmolive’s global IT team described the position this puts an analytics function in.
“We find ourselves constantly fixing metadata in our reporting. It was manual and time consuming. Our reporting was always able to tell us how we did…but we still didn’t know how we were doing. Our analytics was limited.” — Mark Zomick, Sr Product Manager, Global Information Technology, Colgate-Palmolive
How to find metadata
In a file, check properties or EXIF; in a marketing platform, it is the campaign and creative fields.
- On a file (Windows): right-click → Properties → Details.
- On a file (macOS): right-click → Get Info, or open in Preview → Tools → Show Inspector for full EXIF.
- On a web page: view source and read the
<head>, or use a browser extension that surfaces meta tags and structured data. - In an image: a dedicated EXIF viewer shows camera, exposure and location fields the operating system hides.
- In a marketing platform: the campaign, ad set and creative naming fields, plus any custom fields your team has defined. These are the ones nobody thinks of as metadata, and they are the ones that reach your reporting.
The last case is worth stating plainly because it is the one people miss. If you are trying to find your marketing metadata, you are looking for the fields somebody fills in when they set a campaign live.
Who owns metadata standards
Metadata standards are owned where the metadata is created, not where it is reported.
This is the single most common organizational mistake on this topic. Because inconsistent metadata surfaces as a reporting problem, ownership drifts to the analytics or data team, who are the people it hurts and the people least able to fix it. They can observe the inconsistency and clean it after the fact. They cannot be present at the moment a campaign manager or an agency types a value.
The test for whether ownership actually sits in the right place is simple: ask who can stop a wrong value from being saved. If the answer is nobody, or “we catch it in the weekly report,” then the standard is owned by people who can only observe it. A field with a named owner who cannot reject a value is a field with a documented owner and no governance.
Ownership that works has three parts: a named owner for each field, an agreed list of permitted values, and a mechanism that applies the list at the point of entry rather than the point of reporting. The first two are governance work; the third is a systems question, and it is the one that decides whether the first two have any effect. How that is operated day to day — request flows, change control, who may add a value — is metadata management.
See standards applied where metadata is createdPermitted values enforced at the point of entry.Explore Claravine Data StandardsFrequently asked questions
What are the three main types of metadata?
Descriptive, structural and administrative. Descriptive identifies content, structural records how parts relate, administrative governs rights and lifecycle.
What are some examples of metadata?
A photo’s camera and date, a document’s author, a campaign’s channel and audience fields.
How do I find my metadata?
File properties or EXIF viewers for files; campaign and creative fields inside marketing platforms.
Is metadata the same as data?
No. Data is the content; metadata describes it. The distinction is relative — the same field can be metadata in one analysis and data in another, depending on what you are asking about.
Why does metadata matter in marketing?
Because reporting groups by it. Inconsistent metadata fragments the numbers before anyone reads them.
Sources
Outbound citations, named and dated:
- Merriam-Webster, “metadata” (accessed 2026-09-10) — the general-dictionary definition, which is the #1 ranking result for this query and sets the definitional baseline the page must be correct against.
- Carnegie Mellon University Libraries, “Metadata Guide” (accessed 2026-09-10) — the descriptive / structural / administrative taxonomy and its origin in information science.
- Wikipedia, “Metadata” (accessed 2026-09-10) — EXIF as the canonical embedded-file example.
