An overseas marketing team may keep a familiar metric name while changing what it counts. A report called priority-product conversions can refer to one product range in September and a narrower range in October. The chart remains easy to read, but its continuity is cosmetic. A change in the number does not, by itself, describe a change in buyer behaviour.

TikTok's custom conversion creation guide, updated in October 2026, makes an important distinction: names and descriptions can be edited, but rules cannot. A different rule requires a new custom conversion. That is a useful reason to manage measurement definitions explicitly, especially when a local agency and a Chinese supplier share reports but do not share daily access to the same account.

Define the event set before naming the metric

A custom conversion filters events already being sent from a connected source. It does not create information the underlying event never contained. Before proposing a more selective definition, inspect real examples of the event and the parameters available to the rule. Confirm that a product identifier, currency or other chosen parameter has the meaning the team expects.

For a product-specific campaign, a stable product field can be more useful than an incidental word in a URL. However, that is an implementation decision to validate, not a universal prescription. A site might send the same field differently across markets or omit it on one form. The first question is whether the candidate rule includes and excludes the intended examples under the actual event design.

TikTok documents AND logic when multiple rules are added. All conditions must be met. Adding another condition therefore changes the eligible set; it does not automatically improve the commercial quality of the people represented. Missing parameter values can exclude events that the business would otherwise want to count. Calling the result qualified demand cannot repair that mismatch.

Ask the supplier's sales team what distinction they need to observe, then identify which parts are genuinely available in platform events. A request for a catalogue, for example, may be observable without establishing a purchasing timetable. Keep the event definition modest enough to be evidenced and retain later sales qualification in its own records.

Preserve the cutover, not just the current label

When the business changes the conditions, retain the old object's identifier, rule and reporting period. Record the replacement's intended purpose and activation point. A renamed object still follows its original rule; a newly created object can represent a different set. A dashboard should not silently treat those actions as equivalent merely because the human-readable names match.

If both definitions can be observed over a suitable overlap period, compare their matched events before interpreting their totals. This helps reveal whether the new condition excludes the expected cases or merely exposes a data-quality issue. Do not assume that creating the new object reconstructs all historical reporting under the replacement rule. Establish the coverage from the data actually available.

Where no overlap exists, show a break in the series. That may look less tidy than a single uninterrupted line, but it prevents a false performance comparison. A monthly review can describe the reason for the change and report each period under its own definition without manufacturing a percentage improvement across incompatible measurements.

This is separate from overlapping custom conversion rules within one period. Overlap asks whether one event belongs to several sets. Versioning asks whether a set has changed across time. Both require the rule to remain visible, but they are not interchangeable explanations for a reporting difference.

Separate measurement changes from delivery changes

Creating a replacement custom conversion does not establish that every existing campaign has selected it. If the team intends to optimise towards the new object, inspect the effective campaign setting and plan that change deliberately. A reporting definition and an optimisation target are related, but they affect different parts of the operating decision.

Changing both at once can make interpretation harder. The observed result may reflect the altered event set, a different delivery target or both. Where practical, validate the new definition first, then decide how and when it should be used for optimisation. If simultaneous changes are necessary, retain that limitation in the review instead of assigning every movement to creative quality.

For a distributed export team, the handover can stay short: what the old definition counted, what the new one counts, when the switch occurred and which comparisons are no longer valid. The overseas marketer and China-based commercial manager can then decide whether to investigate event data, change creative or wait for a comparable reporting period.

Sources