← All notes

Cognitive load: why your homepage feels noisy (and what I do about it)

Decades of research on working memory and instructional design explain why cluttered sites exhaust visitors. Here is how I apply that evidence when I build products.

About 9 min read

I have lost count of how many founders have asked me to “add one more thing” above the fold. A badge row, a testimonial carousel, three CTAs, and a live chat widget. I get the impulse. You are proud of what you have built, and you want visitors to see all of it immediately. The trouble is that human attention does not work that way.

When I push back, I am not being precious about minimalism for its own sake. I am thinking about cognitive load: the total amount of mental effort your working memory is being asked to spend at once. That idea comes straight out of cognitive load theory, developed from John Sweller’s work on how people learn from structured information (Sweller, 1988). The same constraints show up when someone lands on a marketing site, not only in a classroom.

Three kinds of load (and why only one is helpful)

Cognitive load theory usually splits mental effort into three buckets. Intrinsic load is the difficulty of the task itself—understanding your pricing model is harder than reading a headline. Extraneous load is everything the interface adds that does not help: decorative motion, redundant navigation, competing messages. Germane load is the productive effort of connecting new ideas to what someone already knows.

Good product and marketing design tries to protect intrinsic load (you cannot make “enterprise procurement” trivial), cut extraneous load aggressively, and leave room for germane load by sequencing information instead of dumping it. Richard Mayer’s research on multimedia learning—summarised in Mayer (2009)—gives practical rules that translate surprisingly well to the web: people learn and decide better when you present words and visuals in tight alignment, avoid redundant narration, and signal what matters.

What this looks like on a real project

When I design a hero section, I start from one question: what is the single outcome I want a first-time visitor to hold in working memory after ten seconds? Not three outcomes. One. Everything else—social proof, feature lists, secondary personas—gets a deliberate place further down the page or behind a clear interaction. That is chunking and signalling in practice: the same strategies cognitive scientists recommend to reduce extraneous load in instructional materials.

  • One primary action per viewport height, unless you have a strong reason (comparison tables are a different pattern).
  • Headlines that state a concrete outcome, not an abstract virtue—reduces guesswork and unnecessary search through the page.
  • Visual hierarchy that matches semantic hierarchy: if it looks loud, the brain assumes it is important.
  • Progressive disclosure for detail: accordions, tabs, or dedicated pages instead of foot-long scrolls of everything at once.
If your visitor is still decoding what you sell, they are not ready to weigh fifteen proof points. You are taxing the same limited resource that Sweller’s participants had when poorly designed lessons buried the underlying idea.

The ethical line

Cognitive load is not an excuse to hide material information. Regulations and honest business practice still require clear terms, pricing where it matters, and accessibility. The goal is to remove noise, not to manipulate. When I talk about reducing load, I mean respecting working memory so people can make a real decision—not weaponising confusion.

If you are auditing your own site, try this: open it on a phone you have never used before, set a ten-second timer, and ask a colleague what they believe you want them to do next. If the answer is vague, you probably need less above the fold, not more.

References & further reading

  1. Mayer, R. E. (2009). Multimedia Learning (2nd ed.). Cambridge University Press.
  2. Sweller, J. (1988). Cognitive load during problem solving: Effects on learning. Cognitive Science, 12(2), 257–285.
  3. Sweller, J., van Merriënboer, J. J. G., & Paas, F. (2019). Cognitive architecture and instructional design: 20 years later. Educational Psychology Review, 31, 261–292.