Start with the verdict. A mood board's value is convergence, not presentation. A board that can actually steer modelling holds 9 to 12 images, each one pinned to a single explicit visual feature, and the whole set has to land on three kinds of numbers by the end: a colour code, a texture grade, and a fillet-radius range. Past 20 images the client's attention slides off the product and onto the backdrop, the model and the photography style.
The short version: write down the three adjectives, break each one into visual features that can be compared (cool or warm, hard or soft, gloss or matte), then find images for the features. Keep a page to 9 to 12 images. Every image gets one line stating what we want from it; if the line cannot be written, the image is deleted. What the client signs at the end is a sheet of parameters, not a stack of pictures.
Three adjectives that mean nothing
Every client opens with the same three: minimal, premium, tech-forward. They are true in every industry, which is exactly why they carry no information. Our first move is to dismantle them into axes where the client has to choose a side — cool against warm, hard against soft, gloss against matte, busy against plain — six to eight axes in total.
"Tech-forward" is the clearest example. Pushed cool, it becomes deep space grey and titanium silver with sharp part lines: industrial equipment. Pushed warm, it becomes champagne gold or cream white with generous radii: consumer electronics. One adjective, two completely different product languages, and the client has one of them in mind without knowing how to say which.
Once the axes exist, the skeleton of the board is finished. Each page sits on one end of one axis, and the client chooses between two pages. Two alternatives is a far easier question than "what do you want?", and the answer arrives faster.
In practice the axes get settled in about three rounds. We bring two pages of the same axis, the client picks one, and we ask why. The "why" is worth more than the pick, because it tells us which axis actually matters to them and which ones we can stop asking about. Usually one or two axes turn out to drive everything, and the rest are decoration.
The 40-image review that taught us the rule
We learned this the expensive way. Early on we opened the stock library and accumulated more than 40 images, every style represented, and called it comprehensive. In the review the client flipped to the end, pointed at one and said "I like this blue." We went back to check. The blue in that image was a background wall; the product in the photo was white. That version of the board was wasted. So was more than a week of work built on a direction nobody had actually agreed to.
The rule that came out of it: a mood board shows products only. No lifestyle scenes, no models, no atmospheric photography. If a use scenario genuinely needs to be communicated, it gets its own page, titled "scenario reference — not part of the form decision." Separating that page once removes a whole class of arguments later.
One image, one job, one line of text
Every image on the board carries a line underneath stating what we want from it: "the part line hidden in the side edge," "button and shell in the same colour but different materials," "the slow transition where the radius meets the flat." If I cannot write that line, the image is deleted on the spot.
The test is whether an image can answer a narrow question: what should this particular area of this product look like? Images that only say "this feels nice overall" fail it. Thirty stock images run through that filter usually leave about ten, and each survivor can be traced to a specific modelling decision later — where the part line goes, how the button relates to the shell, how fast the corner opens out.
There is a useful side effect. The client learns the language. In the first review they say "this one looks good." After reading the captions, by the second review they say "I want the feel of this corner, but not its gloss." Once the conversation moves to a local feature, disagreement has somewhere to land.
Layout is about visual weight, not beauty
Two or three key images are enlarged to half a page; the rest run in a single column of small thumbnails. The eye falls on the large ones by itself — that is a steer, and it costs nothing. If every image is the same size, the client shops by colour instead: the red one first, the dark one after.
We do not spend much time making the layout pretty. The board is a decision instrument, and reading order is the only thing it has to get right.
The board still has to be translated into parameters
A nod is not enough. After the client agrees, the modelling team has to model and the factory has to sample, and "premium matte" is not a workable instruction. So a parameter page goes behind every board:
| Dimension | What we commit to | Why this number |
|---|---|---|
| Main colour | Pantone code, ΔE held within 1.5 | Above 2 the client sees the drift on a prototype at a glance |
| Surface | Fine EDM texture or chemical etch, specified by VDI 3400 grade; consumer-electronics matte usually lands around grade 24 | Anything coarser starts reading as tooling rather than product |
| Main radius | A range, not a single value: R3-R5 for handheld products, R8 and above for desktop equipment | A radius is a family of transitions, and the family is what has to be approved |
| Trim parts | Material and process listed separately, e.g. anodised aluminium with a 0.8 mm colour-separation line | Trim is where suppliers improvise unless the spec is explicit |
When the client signs, every adjective on the board has a number behind it. If the sample comes back off, there is something to compare it against instead of an argument about whether it looks premium.
Three grades of mood board
Not every board does the same job. The useful distinction is what the board can carry downstream:
| Collage | Keyword-anchored | Parameter-anchored | |
|---|---|---|---|
| Typical image count | 30-60 | 12-18 | 9-12 |
| Typical client response | picks the images they like | agrees with the direction, still vague on details | can point at a local feature and name what they want |
| Can it guide modelling directly | no | partly | yes |
| Rework risk later | high | medium | low |
Parameter-anchored boards ask more of the client: they have to sit down and confirm line by line, and not every client will. If speed is the priority, keyword-anchored is fine — add the parameter page at concept stage and the project still moves.
When to skip the board
If the client arrives with a specific reference product and wants differentiation by small adjustment, a mood board only slows the schedule down. The opposite case is where the board is almost the only tool that works: the client says "it has to be distinctive" and can supply no reference at all. Then the board is not a presentation aid, it is the only way to get the adjectives out of their head and onto a page where they can be tested.
One rough internal number worth keeping: 80% of aesthetic disagreement happens in the first two rounds of conversation. The board's job is to detonate that 80% while modelling cost is still zero. Fewer images, more captions, and the modelling and sampling that follow run smoother. The client also walks away with a decision they can point at, rather than a folder they have to re-read.
FAQ
Q: How many images should a product mood board hold?
A: Usually 9 to 12, in three groups matching the three keywords. Above 20, the client's attention drifts to backdrops, models and photography style, and the product's own form features get diluted. We also caption every image with the one thing we want from it and delete anything we cannot caption, which typically takes 30 images down to about 10. Each of the survivors then maps to a concrete modelling decision — part-line position, the material relationship between button and shell, the way a corner opens out.
Q: The client cannot describe the style they want. Where does the board start?
A: Break the adjectives into comparison axes — cool or warm, hard or soft, gloss or matte, busy or plain — and find images for both ends of each axis. Choosing between two options is far easier than describing a style from nothing. "Tech-forward" pushed cool means deep space grey with sharp part lines; pushed warm it means cream white, pale gold and generous radii, which is a different product language entirely. Once the axes exist, preferences surface quickly; usually three rounds is enough to converge on one or two directions.
Q: Should competitor products appear on the board?
A: Yes, but on their own page, labelled "difference reference," never mixed with form references. Mixed in, the client unconsciously starts asking for the competitor, and the output becomes a slightly adjusted copy. We put that page last and title it "what we are avoiding," marking the visual features that cannot be used. Used that way, competitor images work as exclusion rather than imitation, and the direction gets clearer for it.
Q: How does the mood board relate to the CMF proposal?
A: The board sets direction; CMF sets parameters. There is one layer of translation between them. The board says "premium matte"; CMF has to turn that into a colour code, a texture grade and a material stack, and if the two do not line up the sample stage loops. That is why a parameter page follows the board: Pantone code with ΔE inside 1.5, surface specified by VDI 3400 grade, radii given as a range, trim listed separately. What the client signs is that page, and it is what the samples get checked against.
Originally published at https://www.hezidesign.com/jingyan/1232.html
