How We Redesigned a Toronto Restaurant's Website in 2 Weeks

In March 2026 a Toronto restaurant owner called us on a Thursday. Their website was built in 2019, hadn't been updated since COVID, and their reservation system was broken. They had a soft opening for a new menu in 16 days and needed a site that worked. Here's exactly what we did, what went wrong, and what the numbers looked like 90 days later.

Day 1-2: The audit that changed the scope

We ran a PageSpeed Insights audit and the site scored 18 on mobile. Eighteen. The homepage loaded a 14MB background video that autoplayed on every device, including phones on cellular data. The menu was a PDF — not indexed by Google, not readable on mobile, and 3 months out of date. The reservation widget was an iframe from a service they'd stopped paying for in 2022. It showed a perpetual loading spinner.

Day 3-7: Build sprint — four pages, zero fat

We replaced the video hero with a single high-quality AVIF image at 180KB — the same visual impact at 1.3% of the file size. The menu became a structured HTML page with schema markup (Menu and MenuItem types) so Google could surface dishes directly in search results. The reservation page got a direct OpenTable embed with a fallback phone number. The about page got a 200-word story with two real kitchen photos.

What we cut and why

• Blog section — zero posts in 4 years. Dead sections hurt more than missing sections.
• Instagram feed widget — added 3 external JavaScript files and 2.1 seconds to load time for a feed nobody scrolled.
• Events calendar — they had 2 events per year. A simple 'Upcoming Events' text block replaced a 400KB calendar plugin.
• Gallery page — 47 uncompressed photos from 2019. We picked the best 8, compressed them, and put them on the homepage.

Day 8-10: The problems we didn't expect

The restaurant's email was tied to their old hosting provider. Migrating the site without breaking email required setting up MX records separately before touching the hosting. We also discovered their Google Business Profile listed a different phone number than the website — a discrepancy that was silently hurting their local SEO for years. We fixed the GBP to match the site and submitted a recrawl request.

Day 11-14: Testing, training, launch

We spent two full days on testing — mobile responsiveness on 6 device sizes, reservation flow end-to-end, menu readability at every viewport, and load time verification from three Canadian cities using WebPageTest. Then we trained the owner's manager to update the menu page using a simplified WordPress editor with exactly one template. No other pages were editable. Constraints prevent future bloat.

The 90-day results

• PageSpeed mobile score: 18 → 91.
• Average page load time: 8.2s → 1.4s.
• Online reservations per week: 12 → 34 (183% increase).
• Google Business Profile views: up 40% — the NAP consistency fix compounded with the schema markup.
• Bounce rate: 72% → 41% — mostly from eliminating the mobile video autoplay that was driving people away.

What we'd do differently

We should have checked the email-hosting dependency on day 1, not day 8. That cost us two days of unplanned DNS work. We also should have set up Google Search Console on day 1 to get a baseline crawl before making changes — without it, we couldn't prove the before/after indexing improvement with hard data. Lesson learned: audit infrastructure before auditing content.

The best redesign isn't the one with the most features. It's the one that removes everything the business doesn't actually need.