Renoir IntelligenceGuide · Marketplaces and creators

Marketplaces and creators

TikTok Shop samples: why a 10-fold return does not mean profit

Sample ROI relates sales to a specific cost base. To decide on further shipments, a brand must understand the denominator, account for other spending and identify where the operation is stalling.

Renoir EditorialPublished on 5 min read

Updated on

A brand distributes samples, sees a 10-fold multiple in its report and considers increasing shipments. The decision seems simple until a question arises that the indicator alone cannot answer: what did it cost to generate those sales? In Renoir’s view, evaluating samples requires connecting commercial figures to the operation that made the content possible, without treating sales as money available to reinvest.

The denominator changes what the result means

In TikTok Shop’s US documentation dated October 2, 2025, Sample ROI for free samples divides GMV by the estimated cost of the products sent as samples. For refundable samples, it divides GMV by the amount refunded to the creator. The cost estimate can be configured. These definitions belong to the US documentation; they do not confirm availability or equivalent rules in Brazil.

The consequence matters: two indicators called ROI can use different bases. A multiple from one model should not be compared directly with the other as though they measured the same expense. Before building a ranking, write beside each figure what was divided by what. Also record whether the cost is estimated or established from the brand’s records.

If an estimate was too low and is revised upward, the multiple falls even if sales do not change. That does not demonstrate worse creator performance. It demonstrates a revision to the calculation base. Keeping the assumption used for each reading prevents a campaign from being rewarded because its cost was underestimated.

R$2,000 in sales and a calculation still unfinished

Consider an entirely hypothetical scenario: a group of free samples has a product cost of R$200 and generates R$2,000 in attributed GMV. Division yields a multiple of 10. The figure describes a relationship between sales and that cost; it does not show what remained. Shipping the samples, for example, may incur a cost without being part of that denominator.

To develop the calculation, assume that, only in this example, R$2,000 also equals the revenue being considered, with no difference between the two amounts. Assume R$900 in cost of products sold, R$200 in commissions, R$100 in order shipping paid by the brand, R$200 for products sent as samples and R$100 to ship those samples. R$500 would remain before other expenses.

This simplified balance is not net profit: it excludes, for example, staff, tools and taxes. Nor does it predict a real campaign’s result. Its purpose is to show why a 10-fold multiple is not a margin. Sold products and samples are separate units; the two shipping expenses cover separate shipments. If a spreadsheet already includes any of these costs, it must not be deducted again.

Compare groups that can answer the same question

In Renoir’s proposed analysis, each batch should retain a recognizable link between the product, creators, sample costs and attributed sales. Comparing the entire store’s GMV with the cost of one small shipment creates a mismatch. If available data cannot distinguish that origin, state the limitation rather than complete the calculation with invented attribution.

To evaluate two batches, also record price, offer and time since receipt. A newly delivered group may not yet have published content; another may have had weeks of exposure. That difference calls for observation, not an immediate judgment about quality. Comparisons between products with different costs and margins also require the broader calculation: the highest multiple cannot settle the choice alone.

The next shipment depends on where work stalled

A weak result first calls for an operational check. How many products arrived? How many creators started production? How many published? These questions form a proposed tracking method, not a promise that the dashboard offers such features. If samples are delayed, demanding more videos or recruiting more profiles does not fix transportation. Additional spending may simply enlarge the backlog of problems.

If products arrived and content was published, the investigation changes. Examine the fit between item, audience, demonstration and offer. Content may explain usage well but face an uncompetitive commercial offer; another piece may reach people without clarifying the benefit. These are hypotheses to check against available material and data. The multiple identifies a result but cannot establish its cause alone.

Expand the test when both the calculation and execution support it

Our recommendation is to authorize the next batch with an explicit hypothesis: retain a particular product and creator profile, correct the delay or test another demonstration. Decide beforehand how much can be allocated to the test and which expenses will enter the assessment. No universal multiple is presented here as a success signal. Room for investment depends on each operation’s structure.

Revisit the decision if actual costs exceed estimates, the new group takes longer to publish or the offer changes. Increasing shipments expands an operational bet; it does not automatically multiply past results. A useful reading of sample performance ends with a bounded next test and visible costs, rather than merely a large number in a report.

Sources

RENOIR INTELLIGENCE

News and analysis of Renoir.