Who this decision is for
- Marketing managers at overseas firms who see different numbers for the same channel in Naver, Google and their own analytics and do not know which to trust.
- Finance or executive sponsors who ask which Korean channel deserves more budget and expect a single clean answer.
- Teams running both search and relationship-led activity in Korea, where much of the influence never shows up in click data.
- Agencies reporting Korean channel results to a foreign client who need to explain gaps without sounding evasive.
The decision to make
The decision is how much weight to give each source of evidence when deciding where to place effort. Platform reports favor the platform that produced them, analytics tools miss visits that arrive without tracking, and buyers who found you through a Korean partner or a search on Naver may later type your address directly. No single view is complete.
The trade-off is precision against usefulness. Chasing a perfect attribution model consumes time and often ends in false confidence. We recommend a modest approach: agree on definitions, gather three kinds of evidence, note their known biases, and reserve major budget shifts for cases where the signals agree, treating small differences as noise.
Options compared
| Option | Best when | Watch out for |
|---|---|---|
| Last-click reporting | The program is small, the buying decision is short, and a simple, consistent view is enough to compare channels roughly. | Undervalues discovery channels such as search content, and overcredits the final direct or branded visit. |
| Self-reported source question | Inquiries are few but valuable, and a required how-did-you-hear field on the form can capture influence that clicks miss. | Answers are subjective and sometimes vague; use fixed options plus a free text field and read them qualitatively. |
| Blended review of platform, analytics and self-reported data | Several Korean channels operate at once and leadership needs to judge them together every quarter. | Requires discipline and explanation; the three views will disagree, and the report must explain why rather than force agreement. |
Risks to manage
- False precision. A tidy percentage split across channels implies certainty that does not exist. Present ranges and directions, and label modelled figures as estimates, never as facts.
- Untagged links. Campaign links without consistent parameters land in a direct or unassigned bucket. Set a naming convention for tracking parameters and check it before each campaign.
- Platform reports taken at face value. Each platform's dashboard credits itself generously. Cross-check against on-site analytics and inquiries before moving budget based on a single dashboard.
- Offline and messaging influence ignored. Phone calls, messaging chats and referrals rarely appear in reports. Add a source question in every inquiry path and ask sales to record it.
Evidence to gather first
- Platform reports from Naver and Google tools covering the same period, with definitions of each metric, so differences can be explained rather than merely observed.
- On-site analytics showing traffic sources and inquiry events, with a check that tracking parameters are applied consistently to campaign links.
- Self-reported source answers collected from inquiries or from sales conversations, kept in a form that can be counted and read.
- The sales team's notes on how Korean buyers describe their discovery, especially referrals, events and partner introductions that no tool records.
Recommended process
- Define channels. SWIN and the client agree the channel list, naming and definitions so that every report uses the same categories.
- Fix the tagging. We set a link-tagging convention, audit current campaign links, and add a source question to each inquiry route.
- Gather three views. Each period, SWIN collects platform reports, analytics data and self-reported answers, and records known biases for each.
- Reconcile and explain. We compare the views, explain disagreements in plain language, and mark conclusions as strong, tentative or unsupported.
- Advise on effort. SWIN recommends modest shifts of effort where the evidence agrees, and suggests small tests where it does not.
What SWIN can deliver
- A channel definition sheet with naming conventions for campaign link parameters.
- A revised inquiry form specification including a self-reported source question with fixed options.
- A period report comparing platform, analytics and self-reported evidence with stated limitations.
- A confidence rating for each conclusion, marked strong, tentative or unsupported.
- A short recommendation note on effort shifts and tests worth running next.