What is cognitive load?

What are the three types of cognitive load?

The distinction comes from cognitive load theory in education, and it transfers usefully because it identifies which load to cut.

Intrinsic load is the difficulty inherent in the task. Choosing between three genuinely different pricing tiers requires real thought, and that thought cannot be designed away without removing the choice.

Extraneous load is effort caused by presentation rather than by the task. Confusing labels, inconsistent layout, decorative animation, and irrelevant options. This is the load worth attacking, and nearly all of it is avoidable.

Germane load is effort that builds understanding. A good comparison table costs attention and produces a confident decision, so the expenditure is productive.

Key takeaway: optimize by removing extraneous load. Cutting intrinsic or germane load removes the information people need, which feels like simplification and performs like omission.

What increases load on a web page?

Too many simultaneous options without hierarchy. See Hick's law.

Inconsistency. Buttons that look different across pages, navigation that moves, terminology that shifts. Each inconsistency forces re-learning.

Unfamiliar vocabulary. Internal product names and jargon force translation work. See information scent.

Dense text without structure. Long paragraphs, no headings, no whitespace. Correct information at a high extraction cost gets skipped anyway.

Competing visual weight. Three elements all styled as primary means none of them is, and the visitor has to decide what matters, which was the design's job.

Requiring memory across steps. A multi-step form that does not show previously entered answers forces the visitor to hold them in working memory.

Movement during reading. Animation beside live content, layout shift, and a modal arriving mid-sentence all reclaim attention that had been allocated to the task.

Why does simplification sometimes fail?

Because "reduce cognitive load" gets applied as "remove content", and those are different operations.

Removing information buyers need. A B2B page stripped of technical detail is simpler and no longer answers the questions a technical evaluator has. The load moves from the page to a support conversation, or to a competitor's page.

Hiding content behind interaction. Accordions and tabs reduce visible density and add a click plus a discovery cost. Sometimes worth it, frequently not, and a closed pane is not part of what a visitor scans, which also makes it invisible in click maps.

Minimalism that removes affordances. Buttons that no longer look like buttons increase load, because the visitor must work out what is interactive. This produces dead clicks.

Shortening copy on considered purchases. Long copy is not automatically high load if it is well structured. A buyer signing a year-long contract wants the detail. The requirement is structure, not brevity.

The reliable version of the principle: reduce the effort required to find and use information, rather than reducing the information.

Which Webflow habits add extraneous load?

Class-based styling makes a new one-off style faster to create than finding the existing one, so sites accumulate near-identical button and heading styles. That inconsistency is extraneous load manufactured by the build. Work from a small set of base classes, and remember that an interaction attached to a class fires on every element carrying it.

Related terms

Hick's law · Fitts's law · Information scent · Above the fold · Form abandonment

Service: Conversion Rate Optimization.

FAQ

Does reducing cognitive load always improve conversion?

Reducing extraneous load generally helps. Reducing intrinsic or germane load removes information or reasoning the decision requires, which can lower conversion or attract poorly qualified buyers.

Is long copy bad for cognitive load?

Not inherently. Unstructured copy is. Long content with clear headings and scannable sections carries lower load than short dense copy.

How do you measure cognitive load on a website?

Not directly. Proxies include time on task, error rates, rage clicks, back-and-forth navigation in session recordings, and usability testing where participants think aloud, which is the most informative of these.