Identity & Addressability
Part of Advertising identity and addressability
Contextual targeting versus identity-based targeting
Compare contextual and identity-based targeting by the decision each can support, its coverage limits and the evidence to request.
Contextual targeting selects an ad opportunity using information about its content or placement. Identity-based targeting uses an eligible identifier linked to an intended audience. Choose by what must be true when the ad is served: it appears in a relevant environment, it reaches a defined audience, or both. Combining the conditions may reduce eligible opportunities.
What each approach can establish
| Campaign question | Contextual approach | Identity-based approach |
|---|---|---|
| Is the placement relevant to the message? | Classify the content or placement at a useful level. | An audience match alone does not answer this. |
| Is this an existing customer? | Content alone cannot establish that. | An eligible customer-to-media match may help, subject to link quality. |
| Has this person seen the ad elsewhere? | Content alone cannot establish cross-property exposure. | Only exposures the system can link and observe can be counted together. |
| Can the targeting decision avoid a cross-context person ID? | Potentially, depending on the implementation. | The decision needs an available identifier or match. |
A category assigned to a whole site does not prove what a particular page or video contains. Ask which content unit was classified and how unclassified placements are handled.
Identity-based buying has a different coverage limit. A customer record may match a media account yet produce few eligible impressions in the chosen inventory.
A platform audience count is not a delivery forecast. Check eligible records, usable supply and delivered impressions separately.
Choose by the brief
Consider a fictional Australian outdoor retailer promoting rain jackets. If it wants to appear beside relevant weather or hiking material, placement classification may answer the targeting question. If it wants a distinct message for eligible previous buyers, it needs a usable customer-audience match.
If it also wants a frequency limit across publishers, context can define suitable placements but cannot recognise the same individual across them. Identity may permit some deduplication, limited to impressions and devices the chosen system can link and observe. Specify whether the limit is counted by person, household, device or browser.
A combined rule may suit a brief that requires both audience and surroundings. Ask for the eligible intersection before booking. Two percentages calculated from different populations cannot be multiplied into a reliable delivery estimate.
Key Considerations in Targeting Strategy
- Coverage limit for identity-based buyingA customer record may match a media account yet produce few eligible impressions.
- Platform audience count vs delivery forecastCheck eligible records, usable supply and delivered impressions separately.
- Frequency limit counting optionsPerson, household, device or browser — specify clearly.
Evaluate the proposed routes
- For context, identify the content unit, classification method and unclassified inventory.
- For identity, identify the input audience, match denominator, identity level and usable inventory.
- Where a comparison is feasible, align creative, dates and geography and document differences in supply.
- Review delivered volume and outcomes with unknown or unobservable populations shown separately. An outcome difference may reflect delivery or measurement access as well as targeting.
Neither label settles privacy considerations. A contextual buy may still pass device or other data through ad technology. An identity buy may involve personal information. Review the actual data flow before describing either configuration as privacy safe.
Privacy Implications of Targeting Approaches
- ProsNeither label settles privacy considerations. A contextual buy may still pass device or other data through ad tech.
- ConsAn identity buy may involve personal information. Review actual data flow before describing either as privacy safe.
