Usability

Definition

Usability is the measurable ease and efficiency with which a user can accomplish a defined task with a product, evaluated along five dimensions: learnability, efficiency, memorability, error rate, and satisfaction.

Usability is a measurable property, not a feeling. Jakob Nielsen's five-component definition (1993) is still the standard: how easy is it to learn the first time, how fast can experienced users go, how well do users remember after time away, how often do they make errors and how easily do they recover, and how pleasant is the experience overall.

A usable product is not the same as a simple product. A flight-booking interface is complex by necessity — multiple legs, fare classes, baggage rules, seat maps. Usability means the user can finish the booking despite the complexity, not that the complexity is hidden. The test is whether the user gets the job done, not whether the screen looks minimal.

Origin

Brought into wide use by Jakob Nielsen's 1993 book Usability Engineering. The ISO 9241-11 standard formalised it as "the extent to which a product can be used by specified users to achieve specified goals with effectiveness, efficiency and satisfaction in a specified context of use."

How it works

Define the user, the task, and the context — usability is always relative to those three.
Set measurable success criteria (e.g. 90% of new users complete signup in under 90 seconds with no errors).
Run usability tests — moderated (you watch and probe) or unmoderated (tools like Maze, UserTesting).
Analyse where users hesitate, click the wrong thing, or give up.
Fix the highest-impact issues first; re-test; repeat until the success criteria are met.

When to use it

Use when

Before launching any flow that drives revenue (signup, checkout, booking, lead capture).
After a redesign — to confirm the new version is at least as usable as the old.
When support tickets cluster around a specific screen or task.
Continuously, on a quarterly cadence, for products with a wide audience.

Skip when

As a substitute for product-market fit. A perfectly usable product no one needs is still a failure.
On flows used once a year by trained internal users — the ROI doesn't justify the spend.

Key metrics

Task success rate (binary or partial).
Time on task (compared to a baseline or competitor).
Error rate per user per task.
System Usability Scale (SUS) — a 10-question survey scored 0–100; ≥68 is above average.
Single Ease Question (SEQ) — "How easy was that task?" on a 7-point scale.

Examples

A usability test with 5 users surfaces 80% of the issues, according to Nielsen.
High usability means the user finishes the task without thinking about the interface.
The SUS score climbed from 54 to 78 after the second iteration.

In practice at Makreate

Every Makreate UX engagement bakes in moderated and unmoderated usability testing. We ship interfaces that have been observed in actual use, not just designed in a Figma file. A recent e-commerce client had 11% checkout completion before the engagement; after two rounds of usability testing exposed three friction points (forced account creation, hidden shipping cost, unclear address autocomplete), we shipped fixes and the rate climbed to 27% — without changing the visual design.

UX Design →

Common mistakes

Confusing usability with simplicity. A usable product can be complex; what matters is whether the user can finish.
Testing only with the team. Engineers and designers share too much context with each other to surface real friction.
Asking 'do you like it?' instead of giving users a task. People will be polite about anything they're not actively trying to use.
Calling 5 users a sample for a quantitative metric. 5 users find qualitative issues; quantitative claims need 20+ users.
Ignoring accessibility. A flow that fails for screen-reader or keyboard-only users has terrible usability for that audience.

Frequently asked

How many users do I need for a usability test?

5 users will reveal 80% of usability issues for a given audience segment, per Nielsen's research. If you have multiple distinct audiences (B2B admin vs. end-user), test 5 in each.

What's a good SUS score?

The benchmark average is 68. Above 80 is excellent, 70–80 is good, 50–70 is okay, below 50 is poor and typically blocks adoption.

Moderated or unmoderated testing?

Moderated for early-stage exploration where you need to ask follow-up questions. Unmoderated for fast iteration on specific tasks once the design is stable. Most teams use both.