Four cheap tests that fail a design before an audience does

You cannot see your own work after two hours

This is not a character flaw. You have spent two hours learning where everything is, so your eye goes straight to each element because it already knows the route. A viewer has none of that, and the gap between your reading and theirs widens with every minute you spend.

The four checks below are all ways of degrading your own vision deliberately so that you see something closer to a first glance. Each takes under a minute and costs nothing.

1. The blur

Apply a Gaussian blur of roughly 1 to 2% of the design’s longest edge: about 10 to 20px on a 1000px canvas. Free everywhere: GIMP (Filters → Blur → Gaussian), Krita, Photopea, or stand back and squint, which is the same test with worse repeatability.

What it tells you: what survives peripheral vision, which is what decides where the eye goes first. If two shapes survive with equal strength, your entry point is unresolved. If nothing survives, the design has no first level at all.

What it cannot tell you: anything about whether the surviving message is the right one. A blur validates rank, never content.

2. Greyscale

One menu item in every image editor, and a filter in browser developer tools.

What it tells you: whether your hierarchy is built on tone or on hue alone. Hierarchies built on hue collapse here, and they also collapse in photocopies, in cheap newsprint, and for viewers with colour vision deficiency.

3. The flip

Mirror the design horizontally. In Inkscape it is the H key; in GIMP, Image → Transform → Flip Horizontally.

What it tells you: balance and awkward alignment, freed from your familiarity. Because you cannot read the text any more, you see the composition as shape. A layout that looked settled often turns out to be leaning, and edges you thought were aligned are visibly not. Painters have used a mirror for this for centuries, for exactly this reason.

Note: do not conclude that the mirrored version is better and ship it. Reading direction matters, and a flipped layout reverses the entry point.

4. Actual size, actual distance

The one people skip, and the one that catches the most expensive mistakes. Print the poster at true size and pin it where it will hang, then walk to where people will stand. Put the reel on a phone and hold it at normal viewing distance. Open the email on the oldest device in the house.

A useful rule of thumb from signage practice: legibility requires roughly 25mm of cap height for every 10 metres of viewing distance, with generous contrast. A shop sign read from across a road needs letters far larger than anyone estimates on a laptop.

What it tells you: whether the smallest necessary element is actually legible in the real viewing condition, which no amount of zooming on a monitor will reveal.

The fifth check, which needs another person

Show it to somebody for three seconds (count them out loud) then take it away and ask two questions: what was it about and what did it want you to do. Do not explain first. Do not defend. Write down what they say verbatim.

This is the only check that tests meaning rather than form, and it is uncomfortable in proportion to how useful it is. Ask two or three people, because one person’s confusion may be idiosyncratic and two people’s confusion almost never is.

What the checks cannot do

All five are tests of transmission. They tell you whether the message arrives, and they say nothing about whether the message is worth sending or whether the viewer will believe it. A piece can pass every check and still fail commercially, because the claim was not credible or the offer was not wanted.

There is also a trap in over-testing. A design optimised solely to survive a blur tends towards one enormous word on a flat ground: maximally legible, frequently empty. The checks are a floor to clear, not a score to maximise, and the right response to passing all of them is not satisfaction but the question the checks cannot answer: does this say the thing the brief asked for?