Most agencies get paid when the files ship. Kaliber puts its creative fee behind what happens after launch — measured against a verified baseline from your own Meta account, judged on the complete batch, refunded in full if it fails.
If the complete Kaliber creative batch fails the test we agreed before production, you choose a full refund of the creative fee or one regenerated batch under the same guarantee. If that regenerated batch also fails a valid test, the fee is refunded automatically.
The whole page is here for anyone who wants to read it. If you don't — these five carry the point.
Your baseline comes from Meta's native account data over 30–60 days — not a spreadsheet prepared for the claim.
Metric, baseline, deployment protocol and data threshold are agreed in writing before we start. No tolerance band is applied to the result.
Meta allocates delivery. No per-asset impression threshold. Your team or agency manages the ads.
The complete batch is judged as one economic result. A lucky low-spend winner doesn't rescue a failing portfolio.
You elect cash or one regenerated batch. If the regeneration also fails, the creative fee is refunded automatically.
How the guarantee is protected
The test itself has no tolerance band. Meet or beat your verified baseline and the batch passes. Miss it by any margin and you are refunded. There is no 5%, 10% or 15% buffer working against you.
Kaliber is protected the way an underwriter is protected — before the risk is accepted, not inside the payout test:
· Selection. We assess the account first and decline coverage where a fair test is not possible. You can still buy the creative without the guarantee.
· A verified baseline. Drawn from 30–60 days of your own Meta data across all comparable campaigns — not your best week.
· Conditions the test must run under. All 25 launched, tracking intact, no mid-test changes to the offer, landing page or audience, and enough data to read the result.
· An inconclusive test is not a failure. If the data can't support a fair reading, we extend or re-run rather than declare a result.
Kaliber bears creative risk. You bear execution risk. Creative that doesn't beat your baseline is covered in full. A test that wasn't run as agreed is not.
Every guarantee is a signed document with a start and an end. Cover begins the day the Guarantee Schedule is signed and ends at a longstop date recorded on it — by default 60 days after your test period closes. Claims are made within 10 business days of the determination report, and never after the longstop. One signed Schedule covers one batch and one test.
Before Kaliber makes the first ad, we use Meta's native account data to agree what success and failure mean. Every field below goes into your signed Guarantee Schedule:
Your baseline is calculated from native Meta account data — not a spreadsheet prepared for the claim. It is spend-weighted across 100% of comparable qualifying campaigns during a 30–60 day reference period.
Comparable means the same campaign objective, attribution setting, audience type and valid tracking. It is not your weakest historical creative and it is not forced to be your strongest — it is representative.
Neither party may exclude qualifying data after seeing the test result. Anomalies are declared and excluded before signature.
Every covered creative in the batch must be launched and remain eligible for delivery under the agreed protocol. Meta may distribute impressions unevenly — that is normal and does not invalidate the test. There is no per-asset impression threshold.
Your team or existing agency manages the ads. Kaliber does not replace your media partner. During the test:
The complete batch is judged as one performance portfolio, using the primary metric you chose.
The result is compared to the locked baseline directly. Meet or beat it and the batch passes; miss it by any margin and it fails. A super-winner that absorbs meaningful spend can carry the portfolio to a pass — because that is what actually improved your business. One low-spend asset technically beating baseline does not automatically rescue a failing portfolio.
If the planned test period is materially different from the reference period because of a major promotion, holiday, launch or predictable seasonal shift, we will use a concurrent control, a matched seasonal period, a refreshed baseline or a later test window. If none produce a fair comparison, we decline the guarantee before you commit.
Coverage is offered only where we can construct one. Common reasons Kaliber declines the guarantee (you can still buy the creative service without it):
The guarantee covers the creative fee only. It does not cover media spend, landing-page work, media-management fees, or any other services. Refunds are paid in the invoice currency within 10 business days of the election or final resolution.
Your in-house media team or existing media agency. Kaliber does not need to replace your media partner. The named media manager follows the launch protocol and gives the batch a fair opportunity to receive delivery.
Meta's native account data — an unaltered ad-level export or scoped read-only account access — is the contractual source of truth for pass/fail. Any analytics platform used to organise or present the data does not override it.
Individual winners are useful creative intelligence, but winner count does not trigger the refund. The combined economic performance of the complete batch determines whether the guarantee passes or fails.
If that creative receives meaningful delivery and improves the combined portfolio enough to pass the agreed test, the batch passes. The guarantee protects the business outcome — not a quota of green cells in a report.
Uneven delivery is expected. Every covered creative must be launched and remain eligible, but there is no per-asset impression threshold. The test becomes conclusive when the complete portfolio reaches the agreed spend and conversion thresholds.
Not without written agreement. Editing a covered creative changes what is being tested. If an asset is rejected because of Kaliber's creative, Kaliber replaces it and the test clock does not begin for that asset.
Material changes can make the baseline incomparable. Disclose them before they occur. Depending on the impact, the parties may continue, correct and extend, exclude affected data, restart, refresh the baseline or end coverage if a valid test is no longer possible.
Natural fatigue during the test is part of real portfolio performance. It is not an excluded circumstance.
Choose a full cash refund of the creative fee or one regenerated creative batch. A regenerated batch carries the same guarantee. A second valid failure results in an automatic cash refund.
No. It covers only the creative fee identified in the Guarantee Schedule.
No. Kaliber declines the guarantee where the account lacks sufficient Meta data, conversion volume, tracking integrity, comparable seasonality, delivery budget or creative freedom to run a fair test. The engagement may still be purchased without guarantee coverage.