Skip to content
About Log in Sign up Deutsch

Guides · For brands and retailers

How do I set success criteria and a stop rule before an influencer test?

Put success criteria and a stop rule on one sheet before launch, and have marketing, finance and the managing director sign it: budget cap, counting rule, minimum volume, pass and stop thresholds, decision-maker, follow-on budget. In Germany, a consumer order should count only once the 14-day withdrawal period is over (§ 355 BGB). Even 100 orders vary by about 20% through chance alone. The sheet works for single bookings, an affiliate programme, an agency and wyrld Boost, where budget and KPIs are fixed in advance anyway.

By the wyrld team · Last updated: · 8 min read

With wyrld Boost, many creators feature your products at the same time on Instagram, TikTok or YouTube. Your budget is fixed up front, and the creators are paid from it by results.

Key points

  • If you expect 8 orders, chance alone puts the result anywhere between 3 and 14; at 100 orders, between 81 and 120 (Poisson distribution, about 95% of cases).
  • In Germany, a consumer order should count only after the 14-day withdrawal period (§ 355 BGB), which for goods starts on receipt (§ 356(2) BGB).
  • Pass and stop thresholds must sit further apart than their chance ranges: in the example, 120 and 60 orders on a €12,000 cap (assumed figures).
  • Our advice: minimum volume first, then the cap and duration that make it possible, then the thresholds. Get it signed before the first creator is approached.
On this page
  1. The test agreement, ready to sign
  2. Why 8 orders prove very little
  3. Setting minimum volume and thresholds so chance doesn’t decide
  4. Counting rule in the test agreement: an order counts only after the withdrawal period
  5. The test with wyrld Boost: which fields the campaign fills and which stay yours

The test agreement, ready to sign

The test agreement is an internal document between marketing, finance and the managing director, not a contract with a partner. It records what counts as failure before anyone has seen a result. A threshold set after the test bends to fit the outcome. A sceptical management team is more likely to approve a pilot if the loss is capped and a failed test does not reopen the whole debate.

The ten fields below are filled in with assumed figures. They are not market benchmarks, so replace them with your own. Where basket value, margin and target cost come from belongs in the business case. How to set up the test itself is covered in our guide to testing on a small budget.

The template deliberately leaves out purchases made without a link or code. That makes the test strict: the figure is a floor, not an estimate. If you want to include sales on Amazon or in stores, define the method in the counting rule beforehand.

Creator pilot test agreement (example figures assumed)

Creator pilot test agreement, [brand], as of [date]

1. Goal: We are testing if creator videos sell [product group] at no more than €100 per kept order.
2. Budget cap: €12,000 net, all costs included (creators, samples, shipping, partner, agency). Anything more needs a new signature.
3. Duration: 6 weeks of posts from [date] to [date], then a 5-week run-off. Cut-off date: [date].
4. Counting rule: An order counts if a link or code ties it to a test creator and, 14 days after receipt, it has been neither withdrawn nor returned. Purchases by creators or staff do not count.
5. Minimum volume: 100 kept orders and 8 creators who have posted. Below that, the test is inconclusive, however good the cost per order looks.
6. Pass threshold: no more than €100 per kept order, so 120 orders or more on the full cap.
7. Stop threshold: €200 or more per kept order, so 60 orders or fewer on the full cap. In that case there is no follow-on budget.
8. Interim check: end of week 3. We review posts, labelling and stock. The thresholds are not judged at this point.
9. Decision-maker: [name, managing director]. Finance supplies the figure, marketing the proposal. Decision no later than 5 working days after the cut-off date.
10. Follow-on budget: On a pass, €30,000 is released for the next wave without a new proposal. Between the thresholds, field 9 decides on one repeat of up to €12,000. On a stop, €0.

Signatures: [marketing] [finance] [managing director], [date]

Why 8 orders prove very little

Average ordersChance rangeSwingWhat you can conclude
83 to 14about ±70%Something is selling, nothing more
2012 to 29about ±45%A rough direction, no comparing creators
5037 to 64about ±28%A verdict only if far from the threshold
10081 to 120about ±20%A pass or a stop can be justified

Eight orders prove very little, because small counts swing widely even when nothing has changed. The chance range in the table is not a forecast. It shows that if the true average is 20 orders, an identical test will land between 12 and 29 in about 95 out of 100 runs, with the same creators, product and price. The figures use the Poisson distribution, which NIST describes as modelling the number of events in a given time interval. Its standard deviation is the square root of the mean, so 10 at 100 orders.

The table shows the minimum uncertainty, because it assumes orders come in independently of each other. With creators that is only roughly true, since a single video can trigger many orders at once. If one creator brings in half the orders, the test tells you more about that video than about the channel.

The same goes for comparing creators. Eight orders for one and 14 for another is not a proven difference, because 14 is still within the chance range of 8.

Setting minimum volume and thresholds so chance doesn’t decide

The minimum volume follows from how much variation you can live with. As a rule of thumb for the 95% range, divide 4 by the permitted swing squared. Allow ±20% and you get 4 ÷ 0.04 = 100 orders; at ±10% it would be 400. If budget and duration cannot get you there, the test is too small for the question. Then scale back what the test is meant to prove rather than stretching a small result. Our order of work would be: minimum volume first, then the cap and duration that make it possible, and the thresholds last.

Pass and stop thresholds must sit further apart than their chance ranges. In the example with a €12,000 cap, the pass is at 120 orders (range 99 to 142) and the stop at 60 (range 45 to 76). The ranges do not touch. Put the stop at 100 (range 81 to 120) and the same campaign could pass one time and fail the next.

The minimum volume protects you against a false success. If the cap is not used up, 40 orders on €4,000 of costs can hit €100 per order. Below 100 orders that is not a pass, just a test with no conclusion.

Between the thresholds lies a grey zone, 61 to 119 orders in the example. It needs its own rule, or every middling result turns into an extension. In the template, that means exactly one repeat that changes one lever.

The stop threshold only governs your follow-on budget, in other words money you have not yet spent. It does not end bookings already running; those are covered by your contract with each partner.

Counting rule in the test agreement: an order counts only after the withdrawal period

For online purchases by consumers in Germany, an order should count only once the withdrawal period (Widerrufsfrist) is over, because until then it can be cancelled without giving a reason. Consumers have 14 days (§ 355(2) BGB), and for goods the period runs from receipt, not from the order date (§ 356(2) no. 1 BGB).

If the goods in one order are delivered separately, the period starts only with the last item (§ 356(2) no. 1(b) BGB). If the consumer was not properly informed about the right of withdrawal, the period does not start, and the right then expires no later than 12 months and 14 days after receipt (§ 356(3) and (4) BGB). Sending the withdrawal in time is enough to meet the deadline (§ 355(1) BGB). The consumer must send the goods back within 14 days of the withdrawal, and here too sending them in time is enough (§ 357(1), § 355(3) BGB).

That is where the five-week run-off in the template comes from. Delivery time, the 14-day period, up to 14 days for the return and arrival at the warehouse add up to about five weeks after the last order. Evaluate earlier and you are counting orders that may still come back. Three cases should therefore be named in the counting rule:

  • Longer voluntary returns: if you offer 30 days, use that period for counting too.
  • Goods with no right of withdrawal: under § 312g(2) BGB this applies, for example, to goods that perish quickly. Count these from delivery and deduct cancellations.
  • Sales to businesses: the right of withdrawal under § 312g(1) BGB belongs to consumers. Show these sales separately so the deadlines stay clear.

The test with wyrld Boost: which fields the campaign fills and which stay yours

With wyrld Boost, budget, products, KPIs and time frame are fixed before launch, which fills fields 1 to 3. The cap covers the full package price plus your own costs. We set out the minimum scope of the campaign in your offer. Counting after the withdrawal period is easiest when people buy in your shop. The minimum volume in field 5 and fields 6 to 10 stay your own figures, because we do not promise reach or revenue. If you need the pass threshold promised, this is the wrong route.

wyrld Boost at a glance

What it is
The campaign service from wyrld, a social commerce platform based in Hamburg, Germany. Many creators feature your products at the same time, so you do not have to book each one. People buy in your own shop, on Amazon or through price comparison: you decide.
Price
You set the budget. Campaigns usually start from €10,000 plus VAT. Creators are paid from the budget according to the KPIs they reach, such as purchases, clicks or watch time. The package price also covers selection, payouts, reporting and a short consultation.
Creators and videos
wyrld selects suitable creators, and each creator decides whether to take part. Creators label campaign videos as advertising. The videos run on the creators’ channels. The right to reuse them yourself is not part of the package.
How it runs
Depending on the campaign, you approve the videos before they go live, or the creators post from your brief. We set out the minimum scope of the campaign in your offer. We do not promise reach or revenue. After the campaign you can see which creator delivered which results.
Who it is for
wyrld Boost is open to brands of every size and in all categories, including brands from abroad that want to sell in Germany, Austria or Switzerland, provided the product and its advertising are lawful. Agencies book as normal customers.
Brand and shop subscription
Separately, this shows you every day which creators feature your products and, where the shop reports purchases, who is selling. Price on request.
FAQ

Frequently asked questions.

Can I change the thresholds during the test?

Only with a new signature from everyone involved and a stated reason, for example a product selling out. Otherwise the threshold bends to fit the result, and the agreement loses its value.

How many orders do I need for a reliable result?

That depends on how much variation you can live with. As a rule of thumb, divide 4 by the permitted swing squared: ±20% gives 100 orders, ±30% about 44. At 8 orders the chance range runs from 3 to 14, so there is very little you can prove.

Does the test agreement work for a wyrld Boost campaign?

Yes. With wyrld Boost, budget, products, KPIs and time frame are fixed before launch, and after the campaign you can see which creator delivered which results. You set the pass and stop thresholds internally. They govern your follow-on budget and are not a promise from wyrld on reach or revenue.

All questions on one page →

Last updated:

Sources

The figures in the example are assumptions, not market benchmarks. This page explains test agreements and withdrawal periods in Germany in general terms and is not legal advice. For what applies to your shop, speak to your legal adviser. Accessed on :

Related questions.

Sell through creators.