An advertising test should answer a specific question: can this offer reach suitable new customers at an acceptable cost? Choose one product group, one campaign environment and one primary outcome. Followers or video views do not prove orders are economical. This plan uses fictional amounts to make the decision transparent.
Start with the offer and contribution
Check price, availability, delivery promise and product page before creating ads. Calculate contribution before acquisition: revenue excluding VAT minus goods, delivery, payments, returns allowance and handling labour. The remainder must cover acquisition and overheads; it is not an unlimited advertising budget.
Configure measurement carefully
Compare campaign reports with actual orders and refunds. Check duplicate purchase events, currency and order value. Server-side transmission can provide another measurement route, but does not guarantee better performance or bypass privacy rules. Tracking cookies require valid consent. Send only data supported by appropriate legal grounds and clear information. Meta describes preparation, testing and deduplication for direct integration. Include that technical work in time and costs rather than treating it as a compulsory first step for every small shop.
Create a compact test matrix
Compare a photo and a demonstration video of the same benefit. Keep price, offer, audience settings and landing page constant and record duration and budget for each. Check how the platform allocates traffic; unequal reach or audiences weaken a simple comparison. Use real product imagery and clear advertising disclosure.
Set caps and review timing in advance
Fictional trial: €150 media, €50 materials and four hours at €25 for creation and review. Total trial cost is €300 including labour. Schedule an interim technical check and final review after the relevant returns period. Do not change everything daily; record each change and its reason.
Use responses to improve product information
Ad responses can expose missing information. Add a size comparison or demonstration when repeated questions show a need. Paid exposure does not purchase a direct Google ranking boost. Accessible pages and helpful, accurate information remain important for organic search.
Assess costs including labour
If the trial produces ten new buyers, acquisition costs are €30 per customer including the labour above. With €22 contribution per order before acquisition, the shortfall is €8 before overheads. Platform ROAS may hide this because VAT-inclusive revenue and attributed orders use different definitions. Use your own clearly stated definitions.
Stop for errors or a reached cap
Pause for a broken link, incorrect price or measurement error. Record the fix and retest product, checkout and measurement before restarting. If the setup works but contribution is insufficient, choose one substantive adjustment or stop within your loss limit. Keep the old and new trial results separate.
Choose one next step
Record what worked, remaining uncertainties and whether another trial fits your loss limit. Expand only when fulfilment and contribution justify it. For a broader starting point, read the first-customer plan.
Sources and checks
Primary sources checked on 26 October 2026: Autoriteit Persoonsgegevens: tracking cookies; Google: helpful content; Meta Blueprint: Conversions API direct integration.
