Web-Development
Kinetic Rebalance
Eleven city microsites for a BD firm and a branding agency, designed in the branding agency's own language. Four days, brief to live.
Web-Development
Kinetic Rebalance
Eleven city microsites for a BD firm and a branding agency, designed in the branding agency's own language. Four days, brief to live.
BD-AEC and BLDG Brands launched a joint offer for architecture and engineering firm owners in eleven cities. We designed the sites in BLDG's own editorial language (chamfered color-block posters, cream and ink, a continuous energy line running the whole scroll), ran a demand study before writing a single title, caught our own first targeting pass aiming at the wrong searcher, and shipped eleven sites from one template in four days. Lighthouse: 96 accessibility, 100 best practices, 100 SEO, and 99 mobile performance under real throttling.
- Web Design
- Next.js Development
- SEO Strategy
- AEO & AI Visibility
- Motion Design
- Performance Engineering
- Multi-Site Architecture
The site
11
City microsites, one template
Measured
4
Days, brief to eleven live builds
Measured
100/100
Lighthouse Accessibility, Best Practices, SEO
Measured
99/100
Mobile performance, real throttling
Measured
1,278ms
Observed LCP, equal to first paint
Measured
50ms
Main-thread blocking time
Measured
50+
Phrasings demand-tested before targeting
Measured
638
AE establishments mapped, Cincinnati alone
Measured
Challenges
What stood in the way
3 problems, and what each one took to clear.
Challenge 01
Problem
Out-design a design agency
The client partner, BLDG Brands, is a creative agency with a strong print-era identity: chamfered tiles, editorial cream, saturated color blocks. Their own site does not express it well. The brief we set ourselves: what could the BLDG homepage look like.
Solution
We built their language as a system: a hand-coded poster wall hero in their palette, chamfered corners at their scale via CSS clip-path, an editorial serif with a weight that settles as headings land, and a dashed energy line that conducts the reader through all twelve sections of a continuous scroll.
Result
A design the branding agency's own principal is credited inside, coded rather than pictured: the hero posters are CSS, not images.
Challenge 02
Problem
The obvious keyword was a trap
The one phrase with search presence in every metro was 'engineering firms {city}'. We had already targeted it when we caught the problem: that searcher is hiring an engineering firm, not running one. Volume with inverted intent is worse than none.
Solution
We retargeted every title to 'Grow Your {City} Architecture or Engineering Firm', a phrase only an owner searches, and moved the volume play into a ten-question FAQ written in the exact full-form phrasings principals type: what a firm is worth, what owners actually make, when succession planning starts.
Result
Matched intent on every page before launch, documented in the code so the reasoning survives the project.
Challenge 03
Problem
A continuous-scroll design against a mobile budget
Twelve sections of choreography (an energy line, weight-settling type, a flywheel cursor) initially spent 660ms of main-thread blocking on phones, and the score dropped under our 90+ bar.
Solution
We pulled the Chrome trace instead of guessing: the static hero painted at 1278ms, but 34 one-shot entrance animations were forcing the animation library to evaluate during hydration. They became IntersectionObserver and CSS transitions with identical trigger points and easing; the fonts were instanced to the axis ranges actually used, 144KB to 69KB.
Result
99 mobile performance under real throttling, blocking time 660ms to 50ms, zero layout shift, and the choreography unchanged.
Strategy
How the work ran
5 phases, in the order they happened.
Phase 01
Demand research before design
A Google-suggest demand study across the metros: 50+ candidate phrasings tested nationally and city by city. The offer's natural vocabulary has zero query presence, so the sites would have to rank for what people actually type.
- Suggest-oracle sweep
- Per-metro keyword verification
- The intent catch and retarget
- FAQ long-tail map
Phase 02
The design system
BLDG's editorial language built as a working system: poster wall hero, chamfer scale, cream-to-ink atmosphere, energy line with per-section entry and exit points, flywheel cursor, weight-settling headings. Above the fold stays static by rule.
- Poster wall (hand-coded)
- Energy line with 4 path shapes
- Type system: Fraunces + Barlow
- Motion rules: below-fold only, reduced-motion and no-JS floors
Phase 03
One template, eleven cities
Cincinnati built as the template; a fork manifest changes only city config, photography, and market data. Each city gets its own photography (never the sister site's), its own Census numbers, and its own Sonor form.
- kinetic-cincinnati.com template
- Fork tooling with build verification
- 11 per-city Sonor forms
- Per-city Census CBP 2022 market sections
Phase 04
The AEO layer
Schema for the real entities only: two Organizations, Person for both principals, Service, FAQPage, speakable. No LocalBusiness, because the microsites have no local premises. llms.txt per site, a coded OG poster card, and every same-owner cross-link nofollowed by a single policy.
- JSON-LD graph per site
- llms.txt and llms-full.txt
- OG poster card
- relFor() link policy: one dofollow, to the independent partner
Phase 05
Diagnose, harden, ship
Trace-level performance diagnosis, a full mobile audit (tap targets, input sizes, caption floors), and fleet verification on every fork before deploy. Screenshots and scores in this study are from the live production site.
- Chrome-trace LCP attribution
- Mobile audit fixes
- 11 verified Netlify deploys
- Post-deploy live verification
Tech stack
What it's built with
9 technologies.
Next.js
React
TypeScript
GSAP
@sonordev/site-kit
Sonor
Netlify
U.S. Census CBP
Lighthouse