Make the service experience complete.
Cellar design, wine racking, cooling, doors and the process are explained on the site, with an inquiry path that keeps the visitor in the same service experience.
Wine Cellars San Francisco / Case studyRoom for the collection.
Room for the collection.
A clearer way to begin.
Cellar design, racking, cooling and installation planning in one self-contained service experience for San Francisco and the Bay Area.
- Work
- Cellar design and wine storage
- Place
- San Francisco and the Bay Area
The approved site brings the collection and the room into the same conversation. Visitors can explore racking and cooling, look through cellar imagery, understand the design process and send a project inquiry directly on the site. Retained cellar articles remain part of the browseable information rather than disappearing in the redesign.
Live on Seppies. Production captures and the final three-run mobile and desktop measurements were recorded after the November 4, 2026 cutover.
01 / The visible work
Move the line.
See the change.
Compare the actual websites. Choose a recorded phase, switch to the phone view, or open either capture at its original size.
Drag anywhere across the image or use the slider. With the slider focused, use arrow keys, Home or End. Phone captures retain vertical page scrolling.
Before cutover · Previous production website
The maintained WordPress website at https://winecellarssanfrancisco.com, captured November 3, 2026. It shows the site as maintained at that date, rather than a reconstruction of its original launch.
Open earlier capture ↗November 4, 2026 · Current live website
The live Seppies production website, captured after the November 4, 2026 cutover. The viewport basis matches the earlier captures. Its separately dated lab results include the final launch changes and normal analytics.
Open later capture ↗The record / Observed website eras
Keep the
source trail.
The record distinguishes the previous website, the selected design and the production rollout.
Current production files and databases, together with the supplied source, are retained privately. File modification times and database dates do not independently establish a public launch date or authorship.
- Before cutover
The actual maintained website
Matched desktop and phone captures, together with dated raw lab observations, document the public website as maintained at the capture date.
- Selected design preview
Design and inquiry handling
The selected design was carried into a separate production build with retained inquiries, customer-facing copy and a full hostname and indexing check.
- November 4, 2026
The production cutover
The production domain now serves the new Seppies build. Dated captures and normal production lab observations followed the hostname, certificate, indexing and legacy URL checks. Historical inquiry reconciliation is a separate retained record.
02 / What we built
The system
behind the screen.
Give the ideas a tangible form.
Supplied cellar imagery, a cellar tour, bottle study and airflow illustration help visitors think about storage, display and the room. Still compositions remain available on unsupported devices.
Keep the useful information.
Fifty-six published cellar articles and the public media they reference are retained at their original article paths, with a journal index and clear routes into a cellar inquiry.
Use a direct, branded contact path.
The production inquiry saves the project details before acknowledging receipt. A domain email gives the service a consistent public contact point; the business's internal inbox remains private.
03 / Retained speed evidence
Measure the change. Keep the context.
The final production homepage measured a median performance score of 94 on mobile and 99 on desktop. Mobile LCP changed from 6.56 s to 2.87 s.
November 3-4, 2026 · Homepage Lighthouse performance scores / 100
| Tested version | Mobile | Desktop | Mobile main-content paint (LCP) |
|---|---|---|---|
| Previous production website | 62 | 81 | 6.56 s |
| Selected design preview | 56 | 73 | 3.61 s |
| Current live website | 94 | 99 | 2.87 s |
| Device / version | LCP | FCP | TBT | CLS | Transfer |
|---|---|---|---|---|---|
| Mobile · Previous production websitePerformance 61-62; LCP 6.53 s-7.02 s; n = 3 | 6.56 s | 4.54 s | 26 ms | 0.0338 | 2.37 MB |
| Desktop · Previous production websitePerformance 80-81; LCP 1.83 s-2.08 s; n = 3 | 1.91 s | 1.35 s | 0 ms | 0.0007 | 3.53 MB |
| Mobile · Selected design previewPerformance 56-57; LCP 3.61 s-3.62 s; n = 3 | 3.61 s | 1.01 s | 2744 ms | 0.0000 | 1.12 MB |
| Desktop · Selected design previewPerformance 73-74; LCP 0.70 s-0.71 s; n = 3 | 0.70 s | 0.27 s | 697 ms | 0.0000 | 1.14 MB |
| Mobile · Current live websitePerformance 94-95; LCP 2.83 s-2.88 s; n = 3 | 2.87 s | 1.38 s | 37 ms | 0.0000 | 0.98 MB |
| Desktop · Current live websitePerformance 99-99; LCP 0.90 s-0.96 s; n = 3 | 0.94 s | 0.44 s | 0 ms | 0.0000 | 1.33 MB |
Read the test method and its limits
Lighthouse 13.0.1 and Chrome 151 from the same pinned Sitespeed 42.6.0-plus1 image on Eggbox. Fresh Chrome processes, standard simulated mobile and desktop presets, 2 CPUs and 4 GiB memory. Three samples per version and device. Before and design-preview observations were interleaved; final production-after observations followed the public cutovers and are separately dated. Medians and ranges are retained.
- These are lab results for the exact pages and dates tested, not field Core Web Vitals, sales or search-ranking outcomes.
- Preview indexing protection is intentional. Preview SEO scores do not represent production indexability.
- The selected preview showed a software-rendering blocking-time regression. Its production build uses a capability fallback. The preview record is retained separately from the final production observations.
- Before, preview and final production are different recorded phases. Final production includes its normal analytics. Launch-day analytics also includes operator and laboratory visits and does not establish business growth.
05 / The source trail
Real sites. Retained evidence.
The visual comparison and selected lab record document the work at the stated dates. The complete original report and supporting files remain available through detailed review access.
Your next chapter
Make the work
speak for itself.
A cellar website should help someone plan a useful room for the collection they love, then make it easy to talk through the details.