The halo effect
One strong impression spreads to unrelated judgements about the same thing.
A salient positive or negative trait colours ratings of traits that have nothing to do with it. People are typically unaware the spread happened and will explain their derived judgement on its own terms. It applies to products and brands as readily as to people.
How it shows up in software
A reputation for reliability makes an unrelated new feature feel safe. A famous security incident makes unrelated features feel careless. Inside a product, one excellent surface lifts the perceived quality of the rest, which is why a well-built onboarding flow flatters a mediocre core, and why a single ugly legacy screen drags on everything near it.
Using it well
- Fix or hide the one screen that is visibly worse than the rest before adding anything new. It is doing more damage than its traffic suggests.
- In research, separate the attribute you care about from the overall impression. Ask about the specific step before asking for an overall rating, not after.
- Recognise that a strong brand impression will carry your weaker surfaces for a while, and use that time to fix them rather than to ship more surfaces.
- When entering a new category, expect existing goodwill to transfer. Design for the expectations that come with it rather than for a blank slate.
Where it turns manipulative
- Trading on a reputation earned in one area to sell something unrelated and weaker, for example a trusted privacy stance used to sell an unaudited AI feature.
- Placing prestigious customer logos or press badges next to claims those customers never made.
- Using a single polished demo surface to raise expectations that the shipped product will not meet.
Where you have seen it
Apple
Packaging, unboxing and retail fixtures receive the same material attention as the device, so the impression is formed before the product is switched on.
Stripe
Documentation quality is treated as a primary surface, and its clarity sets expectations for the reliability of the API behind it.
What the research says
- Thorndike, 1920Well evidenced
Officers rating their soldiers produced trait ratings that correlated far more than the traits plausibly should, which Thorndike named a halo error.
Observational rating data from a century ago. The phenomenon has been replicated many times since; the original methodology is weak by modern standards.
- Nisbett and Wilson, 1977Well evidenced
Students watched the same lecturer behave warmly or coldly, then rated his appearance, mannerisms and accent. The warm version got higher ratings on all three, and students denied that his manner had influenced them.
The unawareness finding matters for research practice: asking users why they rated something highly will not recover the halo.
- Dion, Berscheid and Walster, 1972Mixed evidence
Physically attractive people were attributed better personalities and better life prospects by raters who knew nothing else about them.
The broad effect holds, but specific findings from this literature have been narrowed by later work and some published sizes are inflated.
Grades are a judgement about the evidence, not about how useful the idea is. Plenty of contested effects are still worth knowing, as long as you do not cite them as settled.