The first round of cognitive biases covered some of the most widely recognised - confirmation bias, anchoring, the availability heuristic.

This second set goes deeper into territory that designers encounter daily but rarely name: the mechanics of how effort is perceived, how attention gets stolen, and how the language we use to build trust either works or quietly undermines itself.

These seven biases sit closer to the surface of the user relationship - in the way feedback signals communicate care, in whether our design vocabulary reads as welcoming or suspicious, in the difference between a user who feels valued and one who feels processed.

Labour perception bias

When users see evidence that work is being done on their behalf - a loading indicator, a progress bar, a brief message explaining what the system is processing - they not only understand that their request is being handled, they value the result more than they would if it had appeared instantly.

This is labour perception bias: the displayed effort increases perceived quality.

The practical implications are worth sitting with. A product that surfaces instant results without any visible process risks being perceived as less thorough, less considered, even untrustworthy - not because the output is worse, but because the brain has come to associate effort with value.

Used responsibly, this means designing feedback states that honour the genuine work happening in the background; it doesn't mean manufacturing false delays to exploit the effect.

Banner blindness

Decades of exposure to web advertising have trained users to filter out anything that resembles an ad banner - visually, positionally, structurally.

This filtering is so deeply ingrained that it extends beyond actual advertisements: any element styled or positioned to look like a promotional unit gets treated as visual noise and dismissed before it's consciously read.

The consequence for designers is specific: if you need to surface something important - a system notice, a useful recommendation, timely contextual guidance - avoid the aesthetic vocabulary of advertising. Bright contrasting backgrounds, rounded containers sitting above content, isolated text in a colour that matches a call-to-action button: all of these trigger the avoidance response.

The element's actual usefulness is irrelevant; the styling determines whether it gets read.

Attentional bias

We focus on elements that align with our current emotional or physical state, and we do this mostly without awareness.

A user who is hungry attends to food references in a way they wouldn't ordinarily; someone trying to break an addictive habit has their attention drawn toward it even when they're actively trying to avoid it. Internal state shapes perceptual priority.

For product design, this creates a useful tension. On the one hand, we can design to remove triggers from attention: a health app that de-emphasises foods a user is avoiding, or strips out the language of excess when someone is trying to moderate, operates in service of the user's stated goals.

On the other hand, surfacing timely nudges - a gentle reminder to meditate when a user opens a wellness app - uses attentional bias constructively. The ethical question is whether the direction we're nudging attention genuinely serves the user or serves us.

Expectation bias

When we form an expectation about an outcome, that expectation actively shapes what we perceive.

A researcher who expects participants to respond positively will, without trying to, hear more agreement in ambiguous feedback; a designer who expects a feature to land well will weight the test sessions where it does. Expectation bias doesn't just affect perception - it shapes what data we collect and what we decide to do with it.

Surfacing this bias in research settings is genuinely hard. The corrective is structural rather than motivational: separating the designer from the moderator role, using standardised question formats, and building in explicit stages for null hypothesis testing rather than assuming they'll happen naturally.

The goal is not to become bias-free, which is not possible, but to prevent our pre-existing expectations from determining the outcome before the research has even begun.

Trivialization effect

When someone does something kind or loyal - contributes a review, recommends a product, completes a long survey - a small financial reward in return can actually diminish their relationship to the product rather than strengthen it.

This is the trivialization effect: what felt like a meaningful act of goodwill gets reframed as a transaction, and transactional relationships feel thinner than the original.

The research behind this has direct implications for loyalty mechanics. A genuine, personalised acknowledgement - a thank you that feels like it came from people who noticed what the user did - lands differently to a discount code.

Both are gestures; one feels like reciprocity, the other feels like an exchange rate. Understanding which you're designing will shape whether users feel valued or processed.

Completeness effect

We assign more value to things shown in their whole form than to the same things broken into parts - even when the parts add up to exactly the same quantity.

A full product feels more substantial than its visible components; a bundled offering reads as more generous than the same items sold separately, regardless of the maths.

This plays out in both directions depending on what we're trying to achieve. Showing a product whole creates a perception of abundance and completeness; breaking it into visible components creates a sense of modular value that encourages engagement with each piece.

A fitness app that shows an entire week's programme as a single plan will feel different to one that surfaces one day at a time - not better or worse, but meaningfully different in how the commitment it implies gets processed by the user.

Speak-easy effect

Words that are easy to say are easier to trust.

Research on processing fluency consistently shows that names, terms, and phrases with high phonetic accessibility are rated as more credible, more familiar, and more likely to be chosen than functionally equivalent alternatives that are harder to pronounce. GoodFoods feels like a company you'd hand your money to; ParticularlyCuratedNourishment does not, regardless of the quality of either product.

The practical upshot extends well beyond naming. Microcopy, onboarding language, error messages, and the phrasing of calls to action all benefit from the same principle: clear words, short sentences, and an absence of jargon.

Not because users can't read complex language, but because the cognitive ease of reading directly affects how much they trust what they're reading.

What makes this second set of biases distinct from the first is where they sit in the relationship between a product and its user.

These are not just decision-shaping mechanisms - they govern the quality of the relationship: whether effort feels visible, whether the brand feels like a person or a machine, whether the language creates closeness or distance.

Attending to them is less about avoiding manipulation than about understanding what builds a product that users genuinely want to return to.