-
Notifications
You must be signed in to change notification settings - Fork 3
Description
Context
Data Science are building an automated pipeline that assigns IAB Content Taxonomy tags to Guardian articles.
Commercial could use these tags for contextual targeting via Prebid / GAM.
Before implementation we need to determine how these tags move through the stack.
Pipeline under consideration:
flowchart TD
A[CAPI<br/>IAB Categories]
B[dotcom-rendering<br/>frontendData]
C[Commercial Bundle<br/>Targeting config]
D[Prebid / GAM]
E[Demand Partners]
A --> B --> C --> D --> E
Questions to answer
1. Data contract
Where will IAB tags appear in the CAPI response?
Example possibilities:
- content.metadata.iabTags
- content.tags
Questions:
- structure of tags (string / id / tier)
- number of tags per article
- primary vs relevant tags
2. CAPI → Frontend pipeline
Verify whether tags are already exposed in frontendData.
Possible path:
CAPI → DCR → window.guardian.config.page
If not, identify required change.
3. Commercial integration
Determine how tags should be exposed to the commercial bundle.
Possible location:
window.guardian.config.page.iabTags
Ensure availability before ad initialisation.
4. Vendor targeting
Determine how tags should be passed to demand partners.
Options:
Prebid ORTB2
pbjs.setConfig({
ortb2: {
site: {
content: {
cat: ["IAB1"]
}
}
}
})GAM key-values
Example:
iab=sports_soccer
5. Experimentation
Determine whether tagging should initially be rolled out behind:
- feature flag
- A/B test
Metrics:
- CPM
- Fill rate
- Bid density
Potential follow-up work
These will likely become separate implementation issues once the spike confirms the design.
- Expose IAB tags in CAPI / frontendData
- Surface tags to the commercial bundle
- Integrate tags into Prebid configuration
- Determine GAM targeting strategy
- Implement experiment / feature flag
Expected output of this spike
- confirmed data schema
- defined pipeline from CAPI → Commercial
- recommendation for vendor integration
- list of implementation issues