Xiaohongshu influencer campaigns: connect content performance to sales and costs
Reconcile Xiaohongshu content metrics, linked orders, refunds and campaign costs without confusing attribution with incremental sales.
A Xiaohongshu post with strong views and saves has attracted attention, but the brand still needs to know which orders can be connected to the campaign, what was refunded and how much the collaboration cost. A shopper might see a creator's post, search for the brand later and purchase elsewhere. Neither post views nor a rise in total sales proves that a particular piece of content caused the sale.
For a finance team assessing a China ecommerce campaign, separate content performance, orders linked under a documented rule, and the operating contribution of those orders. Keep unlinked activity visible rather than treating it as zero.
Separate the evidence into three layers
| Layer | Question | Records to reconcile |
|---|---|---|
| Content | Which posts were viewed and engaged with? | Creator agreements, content reports and their reporting period |
| Order linkage | Which orders qualify under the chosen rule? | Official reports actually available to the account, authorised order and refund data, and any valid campaign identifier or matching evidence |
| Contribution | What remains after refunds, product cost, fulfilment and campaign costs? | Product costs, settlement statements, creator and production fees, actual promotion spend and contracts |
The official Xiaohongshu Pugongying creator-collaboration page describes content and conversion reporting. Available fields, access, attribution windows and cross-platform capabilities depend on the merchant's tools, permissions and current rules. A post interaction is not automatically a store order or net profit.
Where a permitted campaign identifier or platform-provided order association is available, record the campaign scope, matching rule, observation window and deduplication method. If orders cannot be linked reliably, show attribution as not available; do not assign every order in the same period to the campaign. When multiple channels claim the same order, define a consistent allocation or deduplication rule before adding reports together.
Calculate contribution for a linked order batch
Suppose a brand runs a Xiaohongshu creator campaign and selects orders linked to it under a pre-agreed, auditable rule. For each order, it reviews completed refunds for 30 days after the transaction. The 30-day window is an internal comparison assumption, not a universal platform refund or attribution rule. Amounts below are RMB millions, excluding VAT, on a consistent cost basis.
| Item | RMB m | Treatment |
|---|---|---|
| Orders linked under the agreed rule | 24.0 | Cancellations and duplicate orders excluded |
| Less refunds completed in the 30-day window | 3.6 | Same order batch |
| Sales after completed refunds | 20.4 | 24.0 − 3.6 |
| Less product cost | 12.24 | Adjusted for actual returned goods |
| Less fulfilment | 1.02 | Attributable to the batch |
| Less platform transaction fees | 0.612 | Assumed not already deducted from sales |
| Less campaign and promotion cost | 2.8 | Creator fees 1.8 + production 0.3 + actual promotion spend 0.5 + other collaboration costs 0.2 |
| Contribution from linked orders | 3.728 | 20.4 − 12.24 − 1.02 − 0.612 − 2.8 |
The example assigns all RMB 2.8 million of the campaign's cost to this linked batch. Some of that spend may also have influenced orders that cannot be linked, so label the allocation explicitly. Do not deduct transaction fees again if sales already reflect them. Likewise distinguish an advertising account top-up, reimbursement to an agency and actual promotion consumption so that the same cost is not counted twice.
The RMB 3.728 million is an operating contribution for orders associated by a stated rule; it is not proven incremental profit caused by the content. Some customers may have purchased anyway, and others influenced by the posts may not be linkable. Later refunds may change the result. Company-wide overhead and tax are excluded, so this is not company net profit.
Put the attribution rule next to the numbers
- Scope: Identify the creator posts, stores, products and order period. Exclude orders from other campaigns unless the allocation rule explicitly includes them.
- Rule: Record the source report, matching conditions, observation window and how orders involving multiple channels are deduplicated.
- Differences: Platform reports, store orders, refund records and bank receipts may use different dates and amounts. Reconcile them rather than treating any one as revenue or cash.
- Unknowns: State what remains unavailable, including unlinked cross-channel orders, longer-term influence, open after-sales cases and costs not separately captured.
- Data access: Use only information the company is entitled to process. Aggregated reporting and necessary order reconciliation do not require copying customers' private communications.
Google Analytics' description of attribution models illustrates how conversion credit can depend on the chosen model; it does not mean Xiaohongshu applies that same model. An order linked by a rule does not by itself establish causation.
Decide what to do with the next budget
Compare campaigns only when the scope and accounting are aligned: the intended audience reached, linked sales and refunds, full collaboration costs, and concentration in a few products. For a future round, consider a comparison designed in advance with similar audiences or content and a specified observation period. With few observations, or simultaneous price and promotion changes, collect more evidence before declaring incremental impact.
If reliable order association is unavailable, assess the content reach and interactions that can be verified, and track store sales and costs separately rather than inventing a return on spend. See how to review a creator collaboration's contribution, the profit analysis example, or contact CaiGeek, an AI financial assistant for small and medium-sized businesses.
Move the work forward with CaiGeek.
Start with your question, supported by business context, data and memory.