Summary and method
A one-page summary and a one-page method, with questions, intent, groups, engines, dates, repeats, tools and missing data.
Product guide, section 9 · Finance proposal, section 4
Campaign report contents
This page describes the method and deliverables in the source documents. They contain no actual campaign, collected AI answers or measurement values.
No collected data · measurements and verdicts are not displayed
What the report includes
Change and verdict by engine
Changes from baseline, treatment-minus-control differences, variation ranges, pre-agreed criteria and the minimum detectable change.
Answers and cited sources
Screens citing partner content, whether the ad label is visible, and the top 10 cited domains.
Reader exposure and raw data
Exposure delivery and make-goods, question lists, raw collected data and the no-guarantee notice.
Collection and verdicts
- Content-topic questions (treatment)
- Comparison questions (control)
- Answer variation before publication
- Publication
- T07 days before
- T4Week 4
- T12Week 12
Example diagram · not real measurements. The same questions are collected for 7 days at three timepoints (T0, T4, T12), and the treatment change minus the control change is compared with the baseline variation range.
- Collect daily for 7 days at each timepoint under fixed Korean-location, Korean-language and web-search-on conditions.
- Timeouts, refusals and empty answers are missing data; normal results without the AI feature are non-appearance. Up to two retries are allowed on the same day.
- A timepoint with fewer than 90% valid responses is marked as insufficient collection and excluded from verdicts.
- Report change observed only when treatment change minus control change exceeds baseline variation and the pre-agreed criterion; otherwise report change unclear.
- Detailed metric formulas, variation ranges, verdict criteria and the minimum detectable change are agreed in the method annex. Thresholds absent from the documents are not calculated.
T1 checks collection, links and labels in Standard; it is not used for effect verdicts. Naver AI Briefing is reference-only in Series and Annual.
The report provides directional measurement and is not IAB-accredited. Citation, exposure, ranking and wording are not guaranteed.
Ten questions from the Finance proposal
These are the question templates in the proposal. The 100 campaign questions are designed after the brief and diagnosis; none are invented here. Measurement of recommendation questions is separate from the topics covered in the content.
| Question | Group | Collected answer |
|---|---|---|
| What kind of financial group is {brand}, and which companies belong to it? | Observation | Not provided |
| What does {brand} do to prevent financial fraud? | Treatment | Not provided |
| What should I ask my bank to do after a suspected voice-phishing call? | Treatment | Not provided |
| How can I protect my parents’ accounts from financial fraud? | Treatment | Not provided |
| Recommend banks that prevent financial fraud well. | Treatment | Not provided |
| Which is safer: a commercial bank or an online bank? | Control | Not provided |
| What must I do, and by when, for a pension-account tax credit? | Control | Not provided |
| Recommend banks with high deposit interest rates. | Observation | Not provided |
| How do financial companies protect customer information when using AI? | Treatment | Not provided |
| How do Korean bank holding companies decide dividends? | Observation | Not provided |
Reader exposure commitments
Commitments from sections 5 and 10-6 of the product guide; these are not delivery records.
- Homepage exposure: 24 hours
- Section exposure: 7 days
- Article-page promotion: 200,000 impressions over 14 days; extended if the target is not met
- Article-page sponsorship: 30 days
- Reader performance report: Based on 30 days after publication