How Should I Use Feedback Like 'Confusing' or 'Looks Cheap' to Decide What to Change in an Ad?

Preserve the reaction before interpreting it

Record the exact comment, creative version, display size and viewing context. Avoid rewriting 'looks cheap' as 'the background color is wrong' unless the reviewer actually identified the background.

Ask a neutral follow-up such as 'What in the image gave you that impression?' or 'What were you expecting to understand?' Nielsen Norman Group's leading-question guidance cautions against supplying the desired interpretation in the question itself.

Separate an observed symptom from a proposed cause. 'I cannot tell which item is included' is a concrete comprehension symptom. 'The product needs a larger shadow' is a proposed visual intervention that may not address it.

Also compare the reaction with the authoritative product and offer records. If the advertisement shows the wrong variant, that is a factual defect requiring correction, not a matter of taste.

Code the symptom into a testable category

Use four provisional categories: product identity, information hierarchy, credibility and missing evidence. These are editorial categories, not a validated psychological scale.

An identity symptom means the viewer cannot name the advertised item or variant. A hierarchy symptom means the viewer notices a secondary element but misses the intended primary message. A credibility symptom concerns a specific visual inconsistency, such as a distorted label or implausible reflection. A missing-evidence symptom means the image asserts or suggests a benefit without showing the necessary supporting information.

One comment can have several plausible explanations. Record alternatives rather than assigning a single cause from a vague adjective. A small qualitative review can identify repair candidates, but not the prevalence of an opinion among customers.

Hypothetical ceramic-mug advertisement

Hypothetical illustration—not collected customer feedback: a fictional advertisement shows a blue ceramic mug against a dark background. A reviewer is imagined saying, 'It looks cheap.' The source record confirms the mug's shape and blue finish but provides no quality-performance claim.

The proposed neutral follow-up asks which visible detail produced that impression. Three possible answers would lead to different work: 'I cannot see the mug's rim' suggests an identity or visibility issue; 'the printed logo looks distorted' suggests a fidelity defect; 'the giant discount badge makes it feel like clearance' suggests message hierarchy.

The editor must not assume that all three occurred. Each is a separate hypothetical branch requiring its own evidence and possible revision.

Change one supported cause and repeat the question

For an identity problem, inspect the product's defining features and intended crop. For a fidelity defect, compare the candidate against the real approved source and repair the actual mismatch. For a hierarchy problem, revise emphasis while retaining the verified product and offer. For missing evidence, obtain the relevant source rather than inventing a persuasive demonstration.

Write the proposed intervention and expected observation before creating another version. If the rim becomes visible, the next question is whether viewers can identify the mug's form—not whether the new background is universally more premium.

Use a fresh unbriefed reviewer when first impressions matter. Record which prior feedback the revised creative was intended to address, and keep unrelated aesthetic suggestions separate.

Stop when the symptom has a defensible explanation and an actionable next check. If the reviewer cannot specify what caused the reaction, retain the comment as subjective feedback rather than inventing a technical diagnosis.

Record your own observations

The filled rows are illustrative examples. Use the blank row for your own files and observations.

Download the editable CSV worksheet

What the references establish

• Nielsen Norman Group — Avoid Leading Questions to Get Better Insights from Participants: Supports neutral probing without suggesting the reason for a participant's reaction.
• Nielsen Norman Group — User Interviews 101: Explains the limitations of self-reported reactions and the need to distinguish reported attitudes from observed behavior.