Most logo presentations show the mark at around 400 pixels, centred, on white. That is the one context in which it will almost never actually appear.

The real test starts after the handover, in places nobody designed for: a browser tab, a stitched polo shirt, a sticker on a curved bottle, someone's slide deck at 9pm. A mark that only holds up in the presentation is not a finished logo, it is a nice drawing. Here is what I check before I call one done.

It has to work at favicon size

Sixteen pixels square. That is the browser tab, and for a lot of brands it is the single most frequently seen version of the mark anywhere.

At that size, detail is not just lost, it turns to mud. Thin strokes disappear, counters fill in, and anything clever collapses into a grey smudge. This is why a logo suite needs a submark: not as a bonus deliverable, but because the primary mark physically cannot do that job. If the identity has no small-size version, someone will make one badly, and it will be the version everyone sees.

It has to work in one colour

Solid black. Solid white knocked out of a dark background. No gradient, no second colour carrying the shape.

One-colour reproduction is where borrowed structure gets exposed. If the mark only reads because a lighter shape sits on a darker one, it stops reading the moment someone faxes it, engraves it, or prints it on a receipt. The one-colour version is not a fallback for cheap jobs, it is the proof that the shape itself is doing the work.

It has to survive being made out of something

Embroidery has a minimum stitch width. Embossing has a minimum relief. Etching, vinyl cutting, and foil all have their own floors, and none of them care how the file looks on screen.

Any mark that is going onto a uniform, a bag, a menu, or a bottle will eventually be handed to a machine with physical limits. Checking that early is much cheaper than redrawing it after the client has already quoted a supplier.

It has to hold still, and it has to move

Animating a mark is the fastest way to find out whether it is actually built well. Motion exposes structure: a logo assembled from unrelated parts falls apart the moment you try to bring those parts in separately, because there is no internal logic to animate along.

I do a lot of motion work alongside identity, and it has changed how I draw static marks. If I cannot describe how the logo would build itself on screen, that usually means the geometry underneath is weaker than it looks.

It has to sit next to someone else's logo

Sponsor walls, partner pages, app store listings, co-branded packaging. The mark will spend a lot of its life in a row of other marks, none of which were designed to flatter it.

That is a weight and proportion problem. A logo that is optically much heavier or lighter than its neighbours looks either aggressive or apologetic, and neither is what the client is paying for. Clear-space rules exist for this exact situation, which is why they belong in the guidelines rather than in the designer's head.

It has to work when someone else is using it

This is the one that decides whether the identity survives the first year.

After handover, the person applying the logo is usually not a designer. It is an office manager building a slide deck, or a supplier laying out a form. If the only correct usage lives in the designer's judgement, the brand starts drifting the week the project ends: wrong colours, squashed proportions, four typefaces nobody chose.

The deliverable that prevents this is not the mark. It is the system around it, which is why every brand identity project I take on ships with a logo suite, a type hierarchy, a palette specified for print and screen, and guidelines that cover misuse as explicitly as correct use.

The mark is the smallest part

A logo is the most visible thing an identity produces and the least of what it actually is. The work that makes it hold up over time is the structure around it, and most of that structure is decided long before anyone approves a shape.

You can see how this plays out across a full system in the Literature & Code and Skinhance projects, where the same mark had to hold across screen, print, and packaging.