“Language will not be merely a set of unrelated sounds, clauses, guidelines, and meanings; it’s a completely coherent system certain to context and habits.” — Kenneth L. Pike
Article Continues Under
The net has accents. So ought to our design techniques.
Design Programs as Residing Languages#section2
Design techniques aren’t element libraries—they’re dwelling languages. Tokens are phonemes, parts are phrases, patterns are phrases, layouts are sentences. The conversations we construct with customers change into the tales our merchandise inform.
However right here’s what we’ve forgotten: the extra fluently a language is spoken, the extra accents it will possibly assist with out dropping which means. English in Scotland differs from English in Sydney, but each are unmistakably English. The language adapts to context whereas preserving core which means. This couldn’t be extra apparent to me, a Brazilian Portuguese speaker, who realized English with an American accent, and lives in Sydney.
Our design techniques should work the identical manner. Inflexible adherence to visible guidelines creates brittle techniques that break beneath contextual strain. Fluent techniques bend with out breaking.
Consistency turns into a jail
The promise of design techniques was easy: constant parts would speed up growth and unify experiences. However as techniques matured and merchandise grew extra advanced, that promise has change into a jail. Groups file “exception” requests by the a whole bunch. Merchandise launch with workarounds as a substitute of system parts. Designers spend extra time defending consistency than fixing person issues.
Our design techniques should study to talk dialects.
A design dialect is a scientific adaptation of a design system that maintains core rules whereas growing new patterns for particular contexts. Not like one-off customizations or model themes, dialects protect the system’s important grammar whereas increasing its vocabulary to serve totally different customers, environments, or constraints.
When Excellent Consistency Fails#section3
At Reserving.com, I realized this lesson the exhausting manner. We A/B-tested the whole lot—shade, copy, button shapes, even emblem colours. As knowledgeable with a graphic design training and expertise constructing model model guides, I discovered this surprising. Whereas everybody fell in love with Airbnb’s pristine design system, Reserving grew into an enormous with out ever contemplating visible consistency.
The chaos taught me one thing profound: consistency isn’t ROI; solved issues are.
At Shopify. Polaris (https://polaris-react.shopify.com/) was our crown jewel—a mature design language excellent for retailers on laptops. As a product crew, we have been anticipated to undertake Polaris as-is. Then my achievement crew hit an “Oh, Ship!” second, as we confronted the problem of constructing an app for warehouse pickers utilizing our interface on shared, battered Android scanners in dim aisles, carrying thick gloves, scanning dozens of things per minute, many with restricted ranges of English understanding.
Process completion with normal Polaris: 0%.
Each element that labored fantastically for retailers failed utterly for pickers. White backgrounds created glare. 44px faucet targets have been invisible to gloved fingers. Sentence-case labels took too lengthy to parse. Multi-step flows confused non-native audio system.
We confronted a alternative: abandon Polaris totally, or educate it to talk warehouse.
The Start of a Dialect#section4
We selected evolution over revolution. Working inside Polaris’s core rules—readability, effectivity, consistency—we developed what we now name a design dialect:
| Constraint | Fluent Transfer | Rationale |
| Glare & low gentle | Darkish surfaces + gentle textual content | Scale back glare on low-DPI screens |
| Gloves & haste | 90px faucet targets (~2cm) | Accommodate thick gloves |
| Multilingual | Single-task screens, plain language | Scale back cognitive load |
Outcome: Process completion jumped from 0% to 100%. Onboarding time dropped from three weeks to 1 shift.
This wasn’t customization or theming—this was a dialect: a scientific adaptation that maintained Polaris’s core grammar whereas growing new vocabulary for a particular context. Polaris hadn’t failed; it had realized to talk warehouse.
The Flexibility Framework#section5
At Atlassian, engaged on the Jira platform—itself a system inside the bigger Atlassian system—I pushed for formalizing this perception. With dozens of merchandise sharing a design language throughout totally different codebases, we would have liked systematic flexibility so we constructed immediately into our methods of working. The outdated mannequin—exception requests and particular approvals—was failing at scale.
We developed the Flexibility Framework to assist designers outline how versatile they wished their parts to be:
| Tier | Motion | Possession |
| Constant | Undertake unchanged | Platform locks design + code |
| Opinionated | Adapt inside bounds | Platform offers sensible defaults, merchandise customise |
| Versatile | Lengthen freely | Platform defines habits, merchandise personal presentation |
Throughout a navigation redesign, we tiered each aspect. Emblem and international search stayed Constant. Breadcrumbs and contextual actions grew to become Versatile. Product groups may instantly see the place innovation was welcome and the place consistency mattered.
The Resolution Ladder#section6
Flexibility wants boundaries. We created a easy ladder for evaluating when guidelines ought to bend:
Good: Ship with present system parts. Quick, constant, confirmed.
Higher: Stretch a element barely. Doc the change. Contribute enhancements again to the system for all to make use of.
Finest: Prototype the perfect expertise first. If person testing validates the profit, replace the system to assist it.
The important thing query: “Which choice lets customers succeed quickest?”
Guidelines are instruments, not relics.
Unity Beats Uniformity#section7
Gmail, Drive, and Maps are unmistakably Google—but every speaks with its personal accent. They obtain unity by means of shared rules, not cloned parts. One additional week of debate over button shade prices roughly $30K in engineer time.
Unity is a model consequence; fluency is a person consequence. When the 2 conflict, facet with the person.
Governance With out Gates#section8
How do you keep coherence whereas enabling dialects? Deal with your system like a dwelling vocabulary:
Doc each deviation – e.g., dialects/warehouse.md with earlier than/after screenshots and rationale.
Promote shared patterns – when three groups undertake a dialect independently, overview it for core inclusion.
Deprecate with context – retire outdated idioms through flags and migration notes, by no means a big-bang purge.
A dwelling dictionary scales higher than a frozen rulebook.
Begin Small: Your First Dialect#section9
Able to introduce dialects? Begin with one damaged expertise:
This week: Discover one person movement the place excellent consistency blocks job completion. Might be cellular customers scuffling with desktop-sized parts, or accessibility wants your normal patterns don’t handle.
Doc the context: What makes normal patterns fail right here? Environmental constraints? Person capabilities? Process urgency?
Design one systematic change: Concentrate on habits over aesthetics. If gloves are the issue, greater targets aren’t “”breaking the system””—they’re serving the person. Earn the variations and make them intentional.
Take a look at and measure: Does the change enhance job completion? Time to productiveness? Person satisfaction?
Present the financial savings: If that dialect frees even half a dash, fluency has paid for itself.
Past the Part Library#section10
We’re not managing design techniques anymore—we’re cultivating design languages. Languages that develop with their audio system. Languages that develop accents with out dropping which means. Languages that serve human wants over aesthetic beliefs.
The warehouse employees who went from 0% to 100% job completion didn’t care that our buttons broke the model information. They cared that the buttons lastly labored.
Your customers really feel the identical manner. Give your system permission to talk their language.
