Research findings

Desktop and mobile UI/UX research • Reviewed September 15–16, 2026

This review treats each website as a designed product: what it is trying to communicate, how its visual system supports that purpose, what interaction adds, and where the experience breaks down. The strongest shared lesson is that distinctive websites build a repeatable visual language around a specific idea. Oversized type, dark backgrounds and animation alone do not provide that idea.

Reading this research

Observed means I inspected the live page or exercised the control. Source-confirmed means public HTML/CSS/JavaScript contains the specific configuration or handler. Within that source evidence, configured means an actual component invocation or page markup is present, and implemented means a matching handler/component was inspected. Hook only means the markup suggests an interaction but its custom implementation was unavailable. Interpretation is my creative-direction assessment, not a measured business outcome.

Desktop inspection used chiefly 1280×720 and 1440px-wide views, with some larger-window observations. Mobile inspection used a 390×844 CSS-pixel viewport. These were responsive browser tests with pointer/keyboard input, not physical iPhone/Android device tests. Screenshot frames show a moment in moving compositions, not the complete animation.

All five homepages were explored through their major sections. Navigation, galleries, menus, accordions and representative secondary templates were inspected. The route inventory covers 65 additional route requests across Ace, MONOLOG and Lama Lama. It does not mean every case image, browser state, external client site or hardware gesture was tested.

This article brings together the cross-site comparison, measured type hierarchy, public-source evidence, and route coverage. The five website reviews contain the detailed browser observations, direct screenshots, implementation findings, and complete website-specific request tables.

Website reviews

Each website has its own page with its browser review, screenshots, measured typography, public-source evidence, and route findings where available.

• Ace — technical editorial precision.
• Arkon Digital — a portfolio as a real-time scene.
• Neue Montréal — a living type specimen.
• MONOLOG — warm editorial confidence.
• Lama Lama — a complete brand behavior system.

Style map

Themes across the collection

Typography is the main identity system

The sites use remarkably few conventional interface decorations. Their hierarchy comes from font character, scale, spacing, weight and placement. Ace contrasts compressed Anton headings with restrained GT America and mono labels. Arkon uses a deliberate three-way contrast: elegant script, narrow industrial capitals and technical body text. MONOLOG alternates strong sans-serif statements, unusual display lettering and small mono annotations. Lama keeps its main sans family consistent and uses mono labels to make actions feel like instrument controls. Neue Montréal makes one family demonstrate an entire range of personalities.

The useful principle is assign each typographic voice a job. An expressive title, a readable explanation and a technical annotation can coexist if their roles stay consistent. Copying all the declared font files would miss the point: shared bundles often contain fonts that are not visibly important.

Layout alternates restraint and release

Ace's fixed left rail provides stability while the right column changes. Arkon arranges content around a central object like a stage. Neue Montréal alternates dense editorial passages with enormous letterforms and immersive full-screen chapters. MONOLOG's small explanatory columns sit beside oversized statements and generous media. Lama mixes narrow control panels with broad image fields and compact project indexes.

Negative space does structural work. It gives a dominant element room, separates chapters and creates anticipation. It becomes a problem when users cannot tell whether the space is deliberate, content is still loading or an interaction is required. Empty published routes are a separate issue from intentional whitespace.

Motion should have a reason

Three particularly strong reasons appear here: demonstration (Neue Montréal's variable weights), proof of skill (Arkon's real-time rendering), and brand consistency (Lama's pixel/logo treatment across media and controls). Ace's spinning seal and image degradation add a technical signature with less visual dominance. MONOLOG uses movement to support an editorial and persuasive sequence.

Persistent ambient motion has a different job from a click response. Ambient motion creates atmosphere; click motion should clarify what changed and where the user went. These sites sometimes mix those responsibilities. A morphing scene is memorable, but a clear selected filter, readable label and obvious close control still matter.

Discovery is enjoyable until it hides the task

The rewarding discoveries include Lama's pitch deck and contact playground, MONOLOG's footer experiment, Neue's city stories and Ace's reactive image treatments. They are optional depth. Navigation, buying, reading and contacting should remain understandable without finding them. In particular, icon-only controls, unexplained horizontal strips and hover-dependent information deserve explicit touch and keyboard alternatives.

Conversion is woven into the composition

Ace keeps booking beside the work on desktop. Arkon directs attention to the portfolio rather than a long lead form. Neue keeps a purchase link visible while visitors explore the specimen. MONOLOG develops trust through work, outcomes, process and FAQs before its closing invitation. Lama repeats contact access and turns its compact menu into a form/calendar/presentation hub.

These are different strategies for different offerings. A type specimen benefits from letting visitors play with the product; a studio must explain fit and credibility; a developer portfolio can use the interface as a live sample of capability. No conversion rates were measured here.

What was measured and tested

Type was sampled at 1440×1000 desktop and 390×844 mobile viewports. Values below are computed CSS pixels, rounded where appropriate, not estimates from screenshots. Font size and line height are separate measurements; a graphic wordmark or transformed specimen cannot reliably be described by the font size of its surrounding element.

I moved the pointer across titles, buttons, navigation, project imagery, postcards, service names, footer canvas regions and team rows. Browser pointer placement was paired with small wheel increments where needed. I checked the hovered element and resulting styles, alongside screenshots. This establishes hover responses; it does not establish physical-device touch behavior or a sustained mouse-button hold. Continuously animated backgrounds were not treated as proof of pointer response merely because successive frames differed.

The suggested reading orders are a creative-direction assessment based on size, placement, contrast and motion. They are not eye-tracking results. The raw computed-style samples retain the measurements; they also include offscreen elements and animation wrappers, so the curated tables below are the appropriate reference for visible typography.

Measured type hierarchy

Numbers shown as size / line height. Tracking is letter spacing, expressed in em where useful. These are observed roles at the tested widths, not an exhaustive design-token specification.

Two distinctions matter. First, computed weight is not proof of a separate font file: a browser can synthesize a requested weight. Second, “headline” can describe visual prominence without meaning an HTML h1. Ace's 40px service terms are h3 elements, while its explanatory h1 is 16px. MONOLOG's semantic h1 is its small positioning statement, while the huge wordmark carries the strongest visual presence. That separation is not inherently wrong; it means visual hierarchy and document structure must be reviewed independently.

Information hierarchy compared

Three levels should be kept distinct when borrowing ideas from these sites. Page hierarchy determines which chapters matter. Component hierarchy determines how a project, quote or service is understood. Interaction hierarchy changes what is emphasized or revealed as the visitor acts. MONOLOG's service list and Lama's team list are clear examples of the third level.

The strongest takeaway is that hierarchy is not simply a descending list of font sizes. Alignment can make equally sized text understandable; contrast can turn a list into a selection interface; a hover can introduce a temporary layer of detail. The successful examples preserve a stable reading structure beneath those changes.

Pointer and interaction findings

The individual website pages document the live pointer evidence: Ace's project images zoom to 1.05; Arkon's CTA grows and navigation arrows appear; Neue's postcards tilt and purchase letters roll; MONOLOG's service list changes emphasis with its preview, and its cursor-following instruction was confirmed; and Lama's team rows reveal portraits and individual personality captions.

An additional responsive discovery is Lama's awards: desktop uses a press-and-hold reveal in its implementation, while mobile supplies a persistent View our awards disclosure that was opened and verified. The Lama Lama page distinguishes that source-confirmed desktop behavior from the live mobile test.

Desktop → mobile comparison

Responsive quality here is best judged by what gets rethought, not only by whether columns fit. Neue's accordion and Lama's services are strong examples. Ace's introduction demonstrates the opposite: changing the outer layout without releasing an inner fixed width breaks the reading experience.

What I would send back to the teams

These priorities are editorial recommendations based on observed impact, not formal severity scores from a full QA program.

Principles worth taking into a new design

1. Choose a concept before choosing effects. Technical precision, a live coding stage, a travel guide, a founder's editorial voice and a pixel-based agency identity are distinct starting points.
2. Give motion an explanatory job. A changing weight shows a font capability; a persistent scene links routes; an opening row exposes portfolio depth. These are easier to justify than an unrelated decorative effect.
3. Keep a quiet place to read. Large media can surround the copy, but the copy still needs stable contrast and a sensible line length.
4. Separate expression from control clarity. Brand the button's behavior, but preserve recognizable selection, focus, labels and exits.
5. Design mobile interactions deliberately. Replace a desktop arrangement when its geometry no longer works. A vertical accordion or horizontal service strip can preserve the idea better than shrinking it.
6. Make hidden interactions optional rewards. A visitor should be able to assess the work and contact the team without finding the experiment.
7. Audit the edges of the experience. Sitemaps, old project URLs, menus, footer links and validation states are part of the delivered design. The most polished homepage does not compensate for an empty destination.

My strongest references by purpose are Neue Montréal for product education, Lama Lama for a coherent interaction language, MONOLOG for narrative and trust, Ace for restrained technical art direction, and Arkon for demonstrable creative-development skill. These are qualitative judgments about this research set, not claims about which site converts best.

Public-source evidence

The public-source inspection, reviewed September 15–16, 2026, explains the fonts, visual tokens, breakpoints, motion settings, and interaction implementations behind the browser observations. All findings come from public HTML, CSS, JavaScript, and sitemaps; no source repository access was available. A shipped definition alone is not proof that a feature is active, and a library import is not proof of a visible effect.

Exact website-specific findings sit under the public-source sections for Ace, Arkon Digital, Neue Montréal, MONOLOG, and Lama Lama. The public origins and inspected asset URLs are retained here:

• Ace HTML: https://acedesign.io/; page bundle https://framerusercontent.com/sites/3ZCVgDgJmNFmhxsu3U6NDN/ZtjpnfryboeNkIbiqgmIC58OsEq7Uq5TjX8rLjG-wag.BV0EGYq6.mjs.
• Arkon HTML: https://arkon.digital/; application https://arkon.digital/index.9bd9d15c.js; CSS https://arkon.digital/index.eaed5af2.css.
• Neue HTML: https://neuemontreal.com/; application https://framerusercontent.com/sites/45hQNNcMArzuoHlzx9jZaS/7OrBmzrr11EPkzHu6vKirIbGo0YLgybo7shxc4A6qus.mA0RSvhl.mjs.
• MONOLOG HTML: https://bymonolog.com/; CSS https://cdn.prod.website-files.com/68b652bbd6c64a44c8fe3e5e/css/bym0n0l0g.webflow.shared.1ee2b3b31.min.css.
• Lama HTML: https://lamalama.com/; CSS https://lamalama.com/wp-content/themes/lamalama2025/dist/assets/main-xY9fs9fO.css?ver=7.1; app https://lamalama.com/wp-content/themes/lamalama2025/dist/assets/app-B43p2XC9.js.

Evidence and boundaries

Temporary source snapshots, route snapshots, and extraction manifests from the research session are not included in this repository. The public origin URLs and module references are cited above and on the website pages; versioned assets may change or become unavailable. Source bundles are minified, so quoted selectors/function names are more useful retrieval anchors than line numbers. Every inventoried route is itself a primary source, although live content may have changed since inspection.

Screenshots are direct captures from the live research session. No image-generation or remote-media downloads were used, and no font binary downloads were necessary. Some animations, particularly pointer effects and mobile touch handling, were explained from inspected implementation instead of claiming they were all exercised. A font declaration does not establish actual rendered usage.

No forms were submitted, bookings made, external AI prompts sent, purchases performed or camera access granted. No Lighthouse or other performance benchmark was run. This research does not certify color-contrast compliance, screen-reader usability, keyboard coverage, reduced-motion compliance, or that every interaction is usable in every browser or viewport. Real-device Safari/Chrome, sustained long-press, full keyboard traversal, slow-network loading and measured frame rate remain outside this review.

Route inventory

65 public URL requests completed on 2026-09-16: 8 Ace, 8 MONOLOG, and 49 Lama Lama routes, including every English Lama case URL in its sitemap (41 URLs, 40 unique destinations). Every request ended at HTTP 200; that does not mean every page is complete. One redirect and several empty/incomplete content templates were found. This inventory describes delivered markup and selectively inspected public component code, not live interaction tests.

Most consequential findings

• Ace /works is an empty page component. The public page bundle creates only a black 1200×1080 div and overlay, with no children. This is stronger than a missing server-rendered text observation. Page, actual page module.
• Ace /laboratorium is not an experiment gallery. It is a personal-profile layout: headline, avatar, biography, social links, employment timeline and email CTA. Its content/style differ materially from the main studio site. It may be residual/template content; the source alone cannot establish intent or verify the claimed career history. Page.
• MONOLOG's seven case URLs deliver the same generic old-looking shell. Each has title “MONOLOG,” no case heading/body, a navigation/footer with 2025 copyright, different positioning/contact copy, and the same text. These are incomplete-looking responses despite HTTP 200, not confirmed 404s. Client hydration can alter source responses; the MONOLOG review records a live confirmation for one case, not all seven.
• Lama /contact has a camera playground. It obtains a front-facing webcam stream on activation, textures it into the branded WebGL grid, offers grid-size and dark/light controls, and stops camera tracks when reverting. This was source inspected only. Contact.
• Lama /ai is sparse. Main page content is an AI heading followed by the standard contact/footer, with no substantive AI service content in the delivered markup. Page.
• Lama Cubby case contains lorem ipsum. Its case introduction is placeholder Latin. Case.
• Lama has duplicate sitemap destinations. /cases/10-10/ redirects to /cases/10-out-of-10/; both appear in the source sitemap.

Route findings by website

The complete request tables are on the corresponding website pages.

Arkon's three main routes and Neue Montréal's one-page structure were inspected separately from these 65 requests. Their observations are on the Arkon Digital and Neue Montréal pages.

Coverage limits and reading guidance

This is a public response/template and selectively inspected interaction-module inventory, not a report that all listed media or controls were exercised in a browser. HTTP 200 only proves a response was returned, not that content is complete or interactions work. The individual website reviews supply runtime evidence on main routes.

Third-party client sites, external portfolios/demos and transactional destinations were not exhaustively explored. Form submission, booking, purchases, camera permission, and client project functionality are outside this inventory. English routes were inventoried; alternate Dutch content was not exhaustively duplicated.

HTML snapshots sometimes contain repeated text for responsive variants, animated hover layers or accessibility/display copies. Those source repetitions were not classified as visual duplication defects. Similarity of case templates is intentional system reuse; the substantive concerns are empty/sparse content, placeholder prose, or navigation to a generic shell. Where case copy describes a client tool or creative system, the description is not evidence that the tool can be operated on Lama's case page.

Use the source findings as an investigation map and explanatory evidence alongside the live observations. Where a feature was only identified in markup, retain that distinction when reading or reusing the research.