The short answer
Low ROAS is an outcome, not a diagnosis. Start by verifying the revenue and spend inputs, then locate which component changed. A campaign can look weaker because of measurement, traffic cost, conversion behavior, order value, or a combination of factors.
Check the report before the campaign
Confirm the currency, dates, attribution settings, conversion definition, and refund treatment. Inspect whether recent data has had time to mature. Check event values against actual orders if revenue looks unexpectedly high or low. A tracking correction and a customer-behavior change require different responses.
Decompose the ratio
Illustrative comparison: the initial period has a 2% order rate, $100 AOV, and $0.50 CPC, giving 4×. The later period has the same order rate and AOV but $0.80 CPC, giving 0.02 × $100 ÷ $0.80 = 2.5×. In this constructed example, the arithmetic change is entirely due to CPC. That does not establish why CPC rose.
Choose checks that match the changed input
- CPC: inspect traffic and placement mix, bids, and delivery conditions.
- Order rate: inspect audience intent, offer consistency, stock, checkout errors, and measurement.
- AOV: inspect discounts, product mix, and the treatment of revenue.
- Contribution: inspect fulfillment, fees, returns, and variable costs.
These are possible investigation paths, not automatic fixes. A landing-page redesign cannot be assumed to solve a change caused by more expensive traffic or a missing conversion event.
Compare meaningful segments
Separate new and returning customers or major product groups when reliable data permits. Avoid selecting tiny segments solely because one ratio is extreme. A few orders can produce an unstable average, and a recent segment may be less mature than the comparison.
Test one explanation
Write a hypothesis, the evidence supporting it, the proposed change, and the outcome that would count as improvement. Include contribution and volume alongside ROAS. Keep a record of changes and allow an appropriate observation period based on the business. If the evidence remains ambiguous, say so rather than attributing the problem to a familiar cause.
Locate the change before choosing a fix
- Gather: Compare report settings, CPC, orders per click, AOV and contribution between mature periods.
- Produce: Choose one plausible explanation and a measurable next check.
Decision boundary: A weak ratio alone does not establish that creative, targeting or the landing page caused it.
Next reading
Examples are illustrative, not client results or benchmarks. The calculations use the inputs shown; real-world decisions depend on your records and assumptions. Report a correction.