Short answer. In careful usage, labelling assigns a class or value to a whole record, while annotation is the broader term covering labels plus structure: where the signal is, what attributes it has, when it occurs and how it relates to something else. In commercial usage the two words are interchangeable and vendors use them inconsistently, so neither tells you what you are buying. Compare the output schema, geometry, attribute set, edge-case rules and review standard instead.
Key takeaways
- Labelling usually means a whole-record decision drawn from a fixed set of options; annotation usually means a decision about part of a record, or one that carries extra structure such as a span, polygon, timestamp, attribute, relation or rationale.
- Vendors use the two words interchangeably, and the same provider will often use both on the same page, so the term on a quote is not a specification and cannot be compared across proposals.
- The one place the distinction carries real information is the billable unit: record-level labels are priced per record, while sub-record annotation is priced per object, per span or per hour and its cost is driven by density.
- A specification that lists the output schema, geometry, taxonomy, attributes, edge-case rules, density expectation, quality standard and delivery format makes the vocabulary question disappear.
- Lifewood Data Technology scopes work from the output schema rather than the service name, with two independent review passes held to a 95%+ accuracy SLA across 50+ languages.
Why are the two terms used interchangeably?
Both words describe adding supervision to raw data, and every labelling task is trivially also an annotation task, so a team marking images as cat or dog is annotating them. The distinction that survives across most usage is one of granularity rather than kind.
Two providers describing the same workflow will call it different things, and the same provider will use both words on the same page. This is not evasion — it is genuine terminological drift across modalities, tools and company histories. It becomes a commercial problem only when a buyer uses the word as though it defined the deliverable, which is when quotes stop being comparable and scope arguments start. Even cloud platforms blur the line: AWS describes a labelling workflow in which several workers' responses, called annotations, are consolidated into a single label.
- Labelling tends to describe a decision about a whole record, drawn from a fixed set of options.
- Annotation tends to describe a decision about a part of a record, or a decision carrying additional structure — a span, a polygon, a timestamp, an attribute, a relation, a rationale.
The American spelling labeling and the British labelling are the same word, and neither carries a technical distinction. Vendors in the same market use both. The table below sets the two terms against the same criteria so the difference can be read at a glance.
| Criterion | Data labelling | Data annotation |
|---|---|---|
| Unit of decision | Whole record | Part of a record, or a record plus structure |
| Typical output | A class or value from a fixed set | Geometry, spans, timestamps, attributes, relations or rationales |
| Modalities where the word dominates | Tabular; document classification | Image, video, entity extraction, speech timing |
| Billable unit | Per record | Per object, per span or per hour |
| Main cost driver | Record count | Density: objects or entities per asset |
| Quality question | Is the label correct against a gold set | Do independent annotators agree on geometry, attributes and judgement |
Where the terms diverge in practice is by modality, and largely for historical reasons:
| Modality | Word usually used | Why |
|---|---|---|
| Image and video | Annotation | The output almost always includes location, not just category |
| Speech and audio | Transcription, or annotation | Output is a text stream plus timings, not a class |
| Text | Both, roughly evenly | Document classification is labelling; entity extraction is annotation |
| Tabular | Labelling | Record-level targets by construction |
| LLM output review | Evaluation, rating, or annotation | The target is a judgement, not a property of the input |
Where does the difference between annotation and labelling actually matter?
There is one place the difference is not cosmetic, and it is the reason it keeps mattering commercially: the two imply different billable units. A record-level label is priced per record, while a sub-record annotation is priced per object, per span or per hour.
The per-record price is stable because every record takes roughly the same time. The per-object price is driven by density — how many objects are in the image, how many entities are in the document — which varies enormously between assets that look identical in a file listing.
This is why a vendor who quotes per image without asking about your average objects per image has priced an assumption rather than your work. Anyone comparing quotes should read how to compare annotation vendor quotes and annotation pricing models before normalising anything, because the unit is where most of the variance between quotes lives. The 10 best human-in-the-loop AI companies for data annotation list shows how differently providers describe the same per-object work.
What should a specification say instead of the term?
Replace the term with eight fields. If all eight are answered, no one needs to agree on vocabulary; if any are blank, the word on the contract will be filled in differently by each party.
- Output schema. The literal structure of one delivered item — field names, types, permitted values. Not a description of it: the structure.
- Geometry or granularity. Whole record, span, box, polygon, mask, keypoint, cuboid, timestamp range, or a rank over candidates.
- Taxonomy. The class list, with an inclusion and exclusion test per class rather than a one-line description.
- Attributes. The optional properties attached to an instance — occlusion, truncation, visibility, confidence, dialect, speaker role. Attributes are how you record a real distinction that annotators cannot apply consistently as a class.
- Edge-case rules. What happens with partially visible objects, overlapping speech, mixed-language text, ambiguous intent — and, critically, what an annotator does when genuinely unsure. A skip-and-flag path with adjudication beats a forced guess, which manufactures noise and hides it.
- Density expectation. Average and range of objects, entities or turns per asset. This is the number that determines cost.
- Quality standard. The metric, the threshold, the overlap rate for double-labelled work, and who adjudicates disagreement.
- Format and delivery. File format, coordinate convention, encoding, split structure, and what accompanies the data — guideline version, agreement figures, provenance.
Two rules of thumb sit behind that list. Do not collect detail the model will never use — every additional field costs money and adds an opportunity for inconsistency. And prototype the schema on a small, deliberately diverse sample before scaling: if trained reviewers cannot apply the instructions consistently, the taxonomy needs revision, and finding that out after fifty items is a morning's work rather than a re-adjudication project. The NIST AI Risk Management Framework makes the same point from the governance side, recommending that data collection and labelling decisions be documented across the AI lifecycle. How unresolved schema ambiguity turns into model error is covered in what data annotation is and how label errors reach the model.
Which does your project actually need?
Work backwards from what the model must predict. The prediction target determines the granularity, and the granularity determines the billable unit.
| If the model must... | You need | Priced by |
|---|---|---|
| Choose one category for a whole record | Record-level labelling | Per record |
| Find something and say where it is | Localisation annotation | Per object |
| Track something across time | Temporal annotation with persistent identity | Per object per sequence, or per hour |
| Relate two things to each other | Relation annotation | Per document, usually per hour |
| Judge whether an output is good | Rubric scoring or preference ranking | Per comparison |
| Explain why a judgement was made | Annotation with rationales | Per item, at the highest rate |
The bottom two rows are where most enterprise spend has moved, and they are also where the word "labelling" is least useful — nobody is assigning a class to anything. A rater is applying a rubric to open-ended output, and the quality question is whether independent raters agree, not whether a label is correct. Writing a rubric that raters can agree on is its own discipline, covered in how to write a preference rubric raters agree on, and the choice between evaluation formats is covered in what to buy: RLHF, SFT or distillation.
How do you compare vendors when their vocabulary differs?
Ignore the headline term and compare the operating method. Six requests settle it, and none of them require the vendor to use your words.
- A sample delivered file from a comparable project, so the schema is visible rather than described.
- The reviewer instructions for that project, including the edge-case section — which is the part that grows during a project and is the highest-value artefact it produces.
- The quality method: overlap rate, metric, threshold, adjudication path, gold-set injection.
- The pre-labelling position: what is automated, at what confidence, and whether unassisted control batches are kept.
- The escalation route for uncertain and out-of-scope items.
- Turnaround on a guideline change, because production annotation is iterative and the responsiveness matters more than the initial rate.
Red flag: a proposal that answers your specification by restating your terminology back to you. It means the vendor has agreed to the word rather than to the deliverable, and the disagreement has been deferred to the first disputed batch rather than avoided. Independent AI data validation of a delivered sample against your own gold set is the quickest way to find out whether the agreement was real.
How does Lifewood scope annotation and labelling work?
Lifewood scopes work from the output schema rather than from the service name. Geometry, taxonomy with inclusion and exclusion tests, attribute set, edge-case rulings and adjudication path are agreed before the first batch, then versioned as rulings accumulate.
The quality standard is stated per task type rather than as a single figure, with two independent human review passes held to a 95%+ accuracy SLA and a 95%+ inter-annotator agreement threshold measured against a customer-approved gold set. Scope spans classification and record-level labelling through to segmentation, tracking, transcription, relation extraction and LLM response evaluation, delivered through the AI data services line.
That scope runs across 100+ languages and 40+ delivery centres across 30+ countries with 56,000+ registered contributors — which matters for the tasks where the judgement depends on hearing or reading the material as a native speaker would. The AI-data heritage runs to 2004, giving the company over two decades of production annotation across every granularity in the tables above.