Davide
Melotti

The problem

A production house sells imagination and has to prove it on everything it prints. The mark had to hold up small on a card and at two metres on a poster, and stay recognisable even reduced to a single letter.

What I designed

• A mark built inside a slanted quadrilateral, like a frame seen at an angle: the shape says “screen” before the name is even read.
• The M is drawn to live on its own: detached from the logotype it becomes a monogram and goes onto bags, shirts and folders without losing identity.
• Three high-contrast colour pairs — mint, magenta, white, always against black — and seven declared lock-ups, so the mark works on light, dark and coloured grounds with no improvised decisions.
• A constant photographic direction: low angle, sky, high saturation. Always the same point of view, that of someone looking up.
• The Think big payoff carried from the poster to the site, where the home opens on those two words and a single call to action.
• The system applied to cards, folders, outdoor posters, t-shirts, tote bags, social and website.

The problem

A small local grower ends up on the shelf next to industrial brands with photographic budgets. It couldn’t win on realism: it had to win on recognition, with something you remember after seeing it once.

What I designed

• The mark comes from botany: the five-pointed star a blueberry carries on its crown becomes the graphic sign running through every application.
• The tittles on the i replaced by blueberries with leaves. The name carries the product inside it and stays legible down to a few millimetres.
• Two lock-ups for two uses: stacked for square formats and icons, a horizontal plaque for signs and labels.
• A character — a walking blueberry — for social, packaging and anything that needs a voice rather than a logo.
• A palette of night blue, cream and leaf green: the colour of the fruit becomes the ground rather than the subject, so the real product stands out on top of it.
• Applications down to the detail of the business card, where the brand’s tone ends up deciding the job title printed under the name.

The problem

A coffee shop has a few seconds and a few centimetres to be recognised: the sign glimpsed in passing, the bag on the counter, the post going by. It needed a sign that stayed itself from a two-centimetre roundel on a sachet to the face of a half-kilo pack.

What I designed

• A single graphic device carries the whole system: the horizon line cutting red from green. It is the sun behind the elephant in the emblem, and it is the same cut that divides bag, tote, sachet and post.
• A two-tier mark — round emblem and plaque with logotype — because the two extremes of scale ask for two solutions, not an enlargement of the same one.
• The elephant is drawn as a solid silhouette with no internal detail: it survives small print, embroidery and a phone screen.
• Heavy condensed type used as a speaking voice: Tastes goooood with the o’s stretched until they fill the line. The typographic joke is also how the text is made to fall exactly into the column.
• The back of the pack given over to type alone, so the two faces divide the work instead of repeating each other.
• The system applied to coffee packaging, menus, folders, teapot, sachets, three versions of tote bag and a social editorial plan.

The problem

A CRM has to show a lot of data without becoming unreadable. The problem ran two ways: making the state of dozens of parallel deals legible at a glance, and cutting data entry down to seconds for someone moving between clients.

What I designed

• A stage view in columns — opportunity created, contact made, presentation, offer sent, won, lost — with per-state counters, total value at the foot of each column and bulk actions: move stage, freeze, archive.
• An alternative list view over the same data, sortable by title, company, owner, deadline, stage and value: two ways of reading the same work, for two ways of working.
• A side detail panel with actions grouped by intent — report, schedule, communicate — so the task closes without leaving the pipeline.
• Trovaclienti: AI segments prospects by spend bracket and offers them for purchase, with summary metrics, comparison against known companies and a history of packages already bought.
• A mobile app built on a newsfeed interaction model: voice memos, one-tap meeting outcomes, contacts on a map. Handed to the team for React Native development.
• A documented design system: eight semantic accents and eight grey steps, complete form field states, a hierarchy of primary, secondary and destructive buttons — with the CSS specifications already written for developers.

The problem

Nobody fills in a CRM from the driver’s seat. The job was to reduce every interaction to a single gesture and give something immediately useful back: where the contacts are, how the month is going. It also needed somewhere to put the people who don’t answer first time, who in the field are the majority.

What I designed

• The twenty-eight day cycle as the load-bearing structure: a new list announces how many contacts it holds and how much time is left, and from there every screen carries the days remaining.
• Contact list with colour-coded status — new, contacted, engaged, closed — distance from your current position and a contextual action on the card, switchable to map from the same control at the top.
• Contact record cut down to what you need at the door: phone, address, supply point and price list, with call and navigation as the only primary actions.
• A feedback loop in a single screen: one question, a face that changes with the answer, four levels, and always a declared way out — call didn’t connect, nobody home.
• Recovery for the missed contact: if it was the wrong moment you reschedule with date and time in two taps, otherwise you declare it lost. Recoveries land in a list of their own.
• Personal performance dashboard: status distribution as a donut chart, redemption rate and won/lost ratio on a continuous bar, with an acknowledgement at the end of the month.
• Product identity: logo, palette and a type scale built on Quicksand, documented in a style guide.

The problem

Tyres are a purchase people postpone until it becomes urgent. The idea was to give the driver a reason to open the app before that moment, and the dealer a reason to get them through the door.

What I designed

• Only three sections in the top bar — treasure hunt, garage, tire test — because an app opened twice a year cannot ask anyone to learn a navigation.
• A Garage section: the user’s car and fitted tyres, with wear as a percentage and a traffic-light colour that flags replacement before it becomes a problem.
• Tire test: photographic tread scanning, with the wheel framed inside a guide circle, to estimate wear and compare the tyres currently fitted.
• A treasure-hunt mechanic on promotions — find the complete series to unlock the discount — turning app usage into footfall at the point of sale.
• High-contrast yellow and black direction, condensed italic type, chequered flag and dashed rules: motorsport language, not e-commerce.

The problem

Someone buying a blade isn’t looking for a product: they’re looking for the right blade for the material they have to cut and the machine sitting in their workshop. A catalogue organised by category doesn’t answer that question. It had to be turned around and start from the job to be done.

What I designed

• Filters built on the real problem rather than on the warehouse taxonomy: you start from fourteen materials to be cut — soft and hard wood, plywood, chipboard, veneered, laminate, MDF, plexiglass, aluminium, ferrous and non-ferrous metal — then narrow with sliders on blade diameter, bore diameter and thickness.
• Technical specs brought onto the list card instead of hidden in the product page: diameter, bore, tooth count and thickness, each with its own icon and its minimum and maximum values. They are read while comparing, which is when they matter.
• Search with a scope selector beside the field: anyone arriving with a code in mind looks it up straight away, anyone without one stays on the category path.
• The wishlist works as a quotation request: codes accumulate and are sent in one go. Technical B2B doesn’t end in a payment, it ends in an offer, and the site walks the buyer to that point.
• A three-level taxonomy across six families — blades, knives, drill bits, router cutters, accessories, sharpening — navigable from the menu, the category grid and the breadcrumb, because a technical catalogue is entered from every side.
• Editorial pages on the subjects that justify the price: Frozen treatment, hard metal, blade characteristics. In a technical market the explanation is part of the product.
• Dark ground with a blue accent: the steel of these products is grey, and on white it disappears. Italian and English across the whole catalogue, product codes included.

The problem

Three different audiences on one homepage: the tourist booking a long route, the commuter renewing a season pass, someone after an urban ticket worth a couple of euros. Each is in a hurry in a different way, and none of the three should trip over the other two.

What I designed

• A trip search engine on the first screen, laid over the image rather than pushed below it: origin, destination, outbound, return and passengers. The swap button between the two places saves retyping everything when looking for the way back.
• The return is a dashed, optional field and the outbound date arrives pre-filled with today: the shortest path is pressed first, the others stay one tap away.
• Urban tickets have a colour of their own, green, and sit outside the route search box. Two different products living on the same screen without being confused for each other.
• Season passes on a tab beside tickets: they need the travel card and therefore registration, and the constraint is stated at the exact point where it blocks, not two screens later.
• On mobile a Buy bar stays anchored at the foot of the whole page: you can read the company history, the routes, the FAQ, without ever losing sight of why you came.
• Service promises spelled out before the purchase — digital ticket, contactless on board, booking changeable up to departure, Wi-Fi and sockets, 1+1 luggage — with the FAQ picking up whatever objections remain.
• The network navigable by city as well as by route, with cards that speak about the place before the timetable, and public service obligations — service charter, conditions of carriage, complaints, accessibility — built in without weighing down the purchase path.

The problem

A typeface is only visible when somebody uses it, and that opens two opposite problems. In promotion, nobody looks at a glyph chart, and the signs that would say the most — currencies, mathematical marks, brackets — are exactly the ones that never turn up in a specimen text. On commission the problem inverts: the letter has to carry content that isn’t its own — a title, a date, a cover — without turning into decoration.

What I designed

• One device in the promo posts, repeated: a circle over the glyph. Translucent, it blurs the stroke; solid, it covers it. Either way the shape has to be rebuilt in the head, and that is what holds the eye on a single letter instead of letting it skim a whole alphabet.
• As the subject of those posts, the part of the set usually left out: pound, yen, generic currency, times, division, braces. These are the glyphs that tell you whether a typeface is finished.
• In the commissioned work the letter is structure, not ornament. In the m the dog’s head grafts onto the last stem and the arches become its legs; in The OG the portrait’s neck runs down into the O.
• Where a system was needed rather than a single drawing, a modular alphabet built on a grid: Heavy Chest is one alphabet recomposed across poster, ticket and flyer in four colour pairings.
• Titles over images kept legible with an outline rather than a band behind them, so the photograph stays whole under the text.
• On records the typeface goes into the credits like every other trade: on the back of observer by |rg| the type design is credited next to the mastering.

The problem

A typeface for an exhibition about a fifteenth-century painter, commissioned by a polytechnic’s architecture programme, has to be in two places at once. Look only to the Renaissance and it becomes historical pastiche that no longer belongs to the institution showing it; look only to the present and the painter’s name on the cover carries no authority. And it has to hold both the title on the wall and the caption in the catalogue.

What I designed

• Two systems in one alphabet: serif capitals, sans lowercase. Not two cuts to alternate — the jump happens inside the word, the only place where the two periods actually touch.
• Wedge serifs, cut on the diagonal rather than bracketed: they read as carved, not written. That is where the architectural half of the reference comes from.
• A lowercase built on circles and straight stems, close to monoline. Its job is to stop the two systems competing: the capitals carry the historical register, the lowercase keeps the text in the present.
• Six weights, with serifs that don’t thin in proportion to the stem: at the light weights they would vanish, and the reference with them.
• An extended Latin set drawn rather than inherited — Æ, ß, Ð — because a public exhibition prints names and loans from other languages.
• Punctuation and symbols assigned to the right system: dots and braces follow the geometry of the lowercase, currencies carry the serifs of the capitals.

The problem

A variable display face has to hold two extremes that are usually drawn as separate typefaces. At minimum weight the stems go to hairlines and the curves tend to disappear before the verticals do; at maximum weight the counters close and the letters collapse into blots. Between them sits an infinite number of weights nobody draws by hand: they have to come out of the interpolation already correct.

What I designed

• Contrast kept low and constant along the whole axis: curves and stems are near enough the same thickness at every weight. That is the condition for the hairline to stay a continuous line instead of breaking where the curve thins out.
• One skeleton for the whole family. The double-storey g shows it plainly: a circle, a rounded loop and a connecting stroke, the same at every weight — only how much they thicken changes.
• Counters reduced to simple geometric shapes so they stay open at black: in A and 4 the void is a minimal triangle, enough to tell the letter apart once everything else has closed.
• The extended set carried across the whole axis, not just the middle weights: accents, currencies, mathematical signs and braces all reach black. A headline containing a price or a formula shouldn’t have to change typeface halfway.
• At the extremes the non-alphabetic signs take on shapes of their own: at black the brace becomes a solid S cut flat at both ends, the tilde on Ñ flattens into a compact stroke to stay under the line height.
• Built as a variable font rather than a series of static weights: the user picks the exact point on the axis, and the named weights are instances of it, not separate files.

The problem

A sans built on the circle is normally a display face. In long text the circle is a liability: round letters take more room than straight ones, the rhythm loosens, and a single-storey a strips the word of one of its strongest recognition cues. The same shapes that make a typeface likeable at 60pt make it tiring at 10. Oxilia had to do both: stay readable across full pages and still work when it is large and the words are few.

What I designed

• The lowercase on one circle, repeated: o, b, d, p and q are the same shape rotated and mirrored. That is where the plain look comes from, and near-zero contrast keeps it plain as the text gets denser.
• Single-storey a and g: these are the shapes taught in handwriting, and they are the exact reason the face reads as ingenuous. A stated choice, not a leftover of the geometric construction.
• The capitals, by contrast, are not geometric: they keep classical widths — narrow B E F P R S, round C G O Q, wide M W. An alphabet of uniform-width capitals comes apart in a headline. The face changes system between the two cases, and that is what lets it hold two registers.
• Small breaks from the geometry where long text needs them: the t is cut on a slant at the top of its stem, the y descends on a straight diagonal rather than a tail. They are footholds — in a line of circles the eye needs something that is not a circle.
• Generous ascenders above the x-height: in a paragraph they keep one line clear of the next as much as the leading does.
• The extended Latin set drawn on the same skeleton: the tilde on Ñ is a full wave, and the bar of the Ø runs past the circle at both ends so it cannot be mistaken for an O at small sizes.

The problem

A sine wave can do almost nothing an alphabet needs. It has no thicks: it cannot give weight to a stem. It has no corners or terminals: it cannot close an A. It never breaks: it cannot make the dot on an i. And above all it encloses no space, while a, b, d, g, o, p and q live on closed bowls. The constraint was to keep the starting shape legible in every glyph, rather than using it as a pretext and then drawing normally the moment it became inconvenient.

What I designed

• One amplitude across the whole alphabet: the letters are different cuts of the same wave, not independent drawings. It is why a line of text reads as a single undulating line even when there are fifty letters in it.
• Constant weight, with no modulation at all. A sine wave has no thick and thin points, and the moment contrast is added the letter stops looking like a wave and starts looking like a script.
• Strokes do not stop at the letter's edge: they run past it and overlap the next one. The space between two letters is not a gap but the continuation of the curve, which turns spacing from a technical problem into part of the drawing.
• Bowls close by crossing, not by joining: the curve passes through itself. It is the only way to get an enclosed space out of a line that by nature never returns to where it started.
• Ascenders and descenders are crests and troughs of greater amplitude, not extended stems. The line therefore has an uneven high-and-low profile, which is exactly what a sound trace looks like.
• One element alone does not derive from the curve: the dot on i and j, a small diamond. A wave cannot produce a dot, and replacing it with a fragment of curve would have made it unreadable. It is the declared exception that makes the rule visible.
• Legibility deliberately held at the edge: some capitals are hard to recognise in isolation and only resolve inside a word. This is a face for headlines and marks, not for text — that is the choice.

The problem

A single-weight typeface has no way out. There is no lighter cut to fall back on when it sits too heavy, and no darker one when more force is needed: whatever it does, it does in one version. High contrast makes the sum harder, because solid stems and hairline joins tighten the counters, and the same letter that is crisp on a sign closes up when it drops to a headline on paper. And there is a question of tone: a strongly contrasted serif looks immediately like nineteenth-century advertising, and then the face isn't contemporary, it is a quotation.

What I designed

• Contrast pushed hard, but with a stated minimum for the counters: in e, a and o the inner space is a narrow slit that never closes. That slit is the limit below which the letter would fill in at the smallest size the face is used.
• Thin, flat serifs, barely bracketed: they cut across the stem rather than growing out of it. It is the detail that keeps the face clear of nineteenth-century pastiche — the historical reference sits in the contrast, not in the joins.
• Teardrop terminals on a, c, f, r and y: the only rounded elements in an alphabet otherwise made of straight cuts. They are the elegant part, and they are few by design — one per letter, never two.
• One weight, not a family. The face is built for a range of sizes rather than a range of weights: everything a text face would solve by switching cut has to be solved inside a single drawing here.
• Figures, currencies and accented letters drawn with the same care as the letters, not bolted on at the end. A headline or a mark often carries a year, a name with an accent, a symbol: if those are weaker it shows at once, because they are as large as everything else.
• In the symbols the horizontal bars hold the hairline weight and the curves hold the stem weight. The sign then carries the same colour as the letters beside it, instead of becoming a lighter or darker patch in the line.
• Accents sit low and wide, close to the letter: at display sizes a thin, tall accent detaches from the body and reads as a smudge.