Illustration and icon sets drawn for one brand only

What custom artwork is actually for

Two things bring people here. The first is sameness. A company looks at its own site and realises the photographs are the same stock images its competitors are using, the icons are from the same free library everybody installs, and nothing on the page could only belong to them. That is not vanity — it is the whole purpose of spending money on a brand, and stock material quietly cancels it.

The second is explanation. Some products are genuinely hard to describe. A good illustration can carry in one glance what three paragraphs fail to land, which is why the technique survives in software, finance and anything with an abstract mechanism at its centre. Icons do the same job at a smaller scale — they are how an interface names things without words, and a set that is even slightly inconsistent makes an otherwise careful product feel assembled from parts.

It is worth separating the two disciplines, because they are not the same skill. Illustration is drawing, and drawing carries a style that has to be developed and held. Icons are closer to systems work — a grid, a stroke weight, a corner radius, a rule for terminals, applied with discipline across dozens of small shapes. A brand can need one, the other, or both, and the honest first conversation is about which of them will actually change how the product looks.

How a set gets made

1. Inventory and cut the list. Every place artwork is needed — empty states, feature sections, onboarding, the actions in the interface — mapped against what exists. Near duplicates get merged, because two icons for almost the same idea is a usability problem rather than a generosity. Half a day to a day.
2. Agree the style before anything is finished. Two or three directions explored on a small representative sample rather than a full set — one dense icon, one simple one, one illustration if both are in scope. Direction is decided here, on cheap work, and not after twenty pieces have been drawn in a style you turn out not to want. Two to four days.
3. Set the construction rules. For icons that means the grid the artwork is built on, the stroke weight, the corner radius, the join and terminal treatment, and how much optical correction is allowed so a circle and a square look the same visual weight. The stroke weight is matched to your interface typeface — a hairline icon next to a medium-weight label looks broken, and it is one of the most common reasons a free library never quite fits. One day.
4. Draw the set, checking constantly at final size. Icons are drawn on the grid and viewed at the sizes they will actually be used at, not at the size they are drawn at. Detail that dissolves at sixteen pixels is removed rather than admired. Illustrations are built with a shared palette and a shared vocabulary of shapes so the set reads as one family. Four to twelve days, depending on count and complexity.
5. Export, optimise and document. Clean SVG with unnecessary nodes and metadata stripped, set up to take its colour from the page so one file serves light and dark themes, plus print vector and raster exports. Alongside it a short style note — the grid, the weight, the radius, what is allowed — so the set can be extended without drifting. One to two days.

What the cost depends on

Count and complexity. A tight icon set of simple, familiar concepts moves fast. A set that has to invent symbols for ideas with no visual convention is slower per piece, because each one needs sketching, testing and often rejecting. Full illustrations sit an order of magnitude above icons on the same scale.

Whether a style has to be developed or applied. Producing artwork in an established style — yours, or one agreed in an earlier engagement — is production work. Developing a style that is distinctive, holds across an expanding set and still looks like your brand is the expensive, valuable part, and it is what the upper end of this range buys.

Where the artwork has to work. Screen-only artwork is the simplest case. Artwork that must also survive print, embroidery, a favicon, an app icon and a dark background is constrained from several directions at once, and each constraint removes options — fine detail, thin strokes, subtle gradients and tight counters all fail somewhere on that list.

Where this work goes wrong

The artwork comes from three different places. One illustration from a free library, one bought from a marketplace, one generated. Each is defensible alone; together the line weights, perspective, palette and level of detail disagree, and the page looks unsettled. You can rarely name the cause — it registers only as something not feeling right — which is why it goes unfixed for months. One source, one style, all the way through, even when that means using fewer pieces.

Icons are drawn large and never checked small. They look immaculate on the artboard at four times size and turn to mush in the interface — hairlines disappear on a standard display, interior detail closes up, and two icons that were clearly different at large size become indistinguishable in a toolbar. Icons have to be judged at sixteen, twenty and twenty-four pixels, on a real screen, throughout the drawing process and not at the end.

Raster is delivered where vector was needed. Flat PNGs — exported, purchased or generated — look fine until the site is opened on a high-density display, or someone needs the mark at billboard size, or dark mode arrives and every illustration carries a white box behind it. Nothing about that is fixable by editing the file; it is a redraw. Anything that will be scaled, recoloured or printed has to be built as vector from the first stroke.