Why a higher cost per click did not settle this campaign decision

A cheaper click is not a sufficient decision rule

In one clinic’s international-customer campaign, average cost per click rose from KRW 1,884 to KRW 3,307. Reversing the campaign on that evidence alone could have missed an important change. Advertising spend was approximately KRW 3.4 million and KRW 3.55 million in the two comparison months, while reported revenue changed from KRW 27.38 million to KRW 62.33 million.

The useful decision is which customer action the campaign should pursue and when its financial outcome can be assessed. Click costs appear quickly. An inquiry may take longer to become a completed visit and payment. Treating indicators that run on different schedules as equally final can destabilize the next budget decision.

Bidding was not the only change between these months; the advertising structure changed too. The later revenue therefore does not establish an isolated causal effect from one bidding setting. It does provide a concrete opportunity to examine which signals a decision needed to consider together.

Attach a reporting status to each number

The earlier month recorded 97 inquiries, KRW 27.38 million in confirmed revenue, and approximately 8x ROAS. The later report recorded KRW 62.33 million and provisional ROAS of 17.6x, based on 22 completed visits, with six people still to visit. Inquiries, visits, and revenue cannot be used as interchangeable outcomes.

The approximately 128% revenue increase is the difference between the two reported revenue figures divided by the earlier figure. It excludes anticipated revenue from the six pending visits. Reported ROAS relates revenue to advertising spend; it does not deduct treatment costs, payroll, agency fees, or other costs to establish net profit. Rounded spend also limits arithmetic precision.

Bidding and five service groups changed together

The project began with a click-focused approach, then moved from Maximize Clicks to Maximize Conversions. Advertising was organized into five service groups, each with search terms, copy, and destination pages. This created a structure for examining the service a prospect sought and the explanation they encountered, beyond the overall account average.

Different services can involve different purchase conditions. Timing may dominate one decision while scope dominates another. Separate groups can reveal those differences. They do not automatically improve performance, and excessive fragmentation of small samples can make temporary zero-conversion results look like a lasting absence of demand.

Maximize Conversions requires conversion tracking and aims to obtain as many configured conversions as possible within the budget. This project’s click-first sequence should not become a fixed waiting period for every account. Another account’s starting strategy should reflect tracking readiness, the meaning of its configured actions, and the available budget.¹

Applying the lesson: inspect the action behind the label

A conversion label inside an ad account does not establish a completed clinic visit. A contact-button click, a valid inquiry, and a confirmed appointment are different events. Inspect which action the account actually counts and whether increasing it supports the operating objective. A familiar label can conceal a measurement definition that no longer fits the decision.

Primary conversion actions feed the Google Ads Conversions column and bidding. Reporting based on ad-click dates also differs from reporting by the time a conversion occurred. Comparing advertising reports with payments can therefore place the same transaction in different months. Align those definitions before treating a discrepancy as a tracking error.²

A useful review places the action name beside its actual trigger, duplicate-handling method, and the next status checked by the consultation team. Where an action cannot be connected to revenue, state that limit. The purpose is to explain what bidding has been asked to increase, rather than make the account’s conversion total larger.

Keep pending appointments in a separate state

The project record describes an earlier provisional ROAS changing after later visits were completed. International customers may book before traveling, leaving time between the initial inquiry and payment. Adding new inquiries and visits from earlier inquiries as if both were new outcomes creates a risk of double counting across reports.

Retain both the inquiry date and the completed-visit or payment date. Keep pending bookings apart from confirmed revenue, then update the same inquiry when its status changes. A revenue total without dates is harder to match to the inquiries that produced it. The completed status needs a timestamp as well as an amount.

That distinction also prevents premature judgment of the later period. Incomplete outcomes may justify further observation before a major budget change. A confirmed measurement error calls for correction immediately. Limited data and incorrectly defined data are different problems and need different responses.

Make the next decision on comparable outcomes

A higher CPC does not, by itself, require a cut. Higher revenue does not, by itself, justify unrestricted expansion. Examine whether service-level inquiries progressed to bookings and visits, whether revenue depended on a few large transactions, and how many outcomes remain pending. Apply the same discipline when comparing countries.

Expansion becomes more defensible when action definitions and reporting dates are stable and useful downstream outcomes recur. Misclassified consultations or possible duplicate revenue require a measurement repair. A small sample with many pending visits calls for more time. Distinguishing those situations matters more than the bidding strategy’s name.

The repeatable decision is whether there is evidence to accept a more expensive click, and when that evidence becomes sufficiently complete to act on. Answering both questions gives the next budget decision a firmer basis than reacting to a change in the click price alone.