Restaurant websites look simple on the surface — a menu, some photos, a reservation button — but they come with a surprisingly tricky set of constraints: heavy image galleries, mobile-first ordering flows, and owners who need to update content themselves without touching code.

I recently built Dinecraft, a free WordPress theme aimed at restaurants, and wanted to share a few technical decisions that made the biggest difference — plus a couple of mistakes I'd avoid next time.

1. Menus are content, not just markup

Most restaurant theme demos hardcode the menu into the template. That looks great in a screenshot but breaks the moment a real owner needs to swap out a seasonal dish.

Instead, I structured menu items as a custom post type with taxonomies for category (starters, mains, desserts) and dietary tags (vegan, gluten-free). This meant:

• Owners can add/edit dishes from the standard WP admin, no page builder needed
• Menu items are filterable on the front end with a small vanilla JS filter (no jQuery dependency)
• The same content structure can be queried for a "today's specials" widget later

php

register_post_type('menu_item', [
    'public' => true,
    'has_archive' => true,
    'supports' => ['title', 'editor', 'thumbnail'],
    'taxonomies' => ['menu_category', 'dietary_tag'],
]);

2. Image weight was the biggest performance lever

Restaurant sites live and die by food photography, which means large images are non-negotiable — but they're also the easiest way to tank your Core Web Vitals.

What actually moved the needle:

• Serving WebP with a JPEG fallback via <picture>
• Lazy-loading everything below the fold with native loading="lazy"
• Capping hero images at 1920px and letting srcset handle the rest

This alone took the homepage from a ~4.2s LCP down to under 1.8s on a throttled 4G test.

3. Reservation forms shouldn't require a plugin bloat

A lot of restaurant themes bundle a heavy booking plugin by default. I wanted Dinecraft to work with popular reservation plugins (OpenTable embeds, Calendly, etc.) without requiring one, so the base theme ships with a lightweight contact-form fallback and clearly marked template hooks for anyone who wants to drop in their own booking widget.

php

do_action('dinecraft_before_reservation_form');

This kept the core theme lighter and gave more flexibility than baking in a single vendor's plugin.

4. Mobile-first, not mobile-adapted

About 70% of restaurant site traffic is mobile — usually someone standing outside deciding whether to walk in. I built the layout mobile-first in CSS rather than adapting a desktop layout down, which avoided the usual pitfalls of oversized tap targets and hidden nav items on small screens.

What I'd do differently

• I underestimated how many owners would want a dark mode toggle for evening ambiance branding — adding that after the fact meant retrofitting CSS variables I should have set up from day one.
• I'd build the menu filter with IntersectionObserver from the start instead of a scroll-based check I later had to replace.

Try it / feedback welcome

The theme is free and open for anyone to use or fork: Dinecraft — Free Restaurant WordPress Theme.

If you've built something similar, I'd love to hear how you handled menu content structuring or image optimization — always curious how other devs solved the same problems differently.

Hey everyone,

I built a Magento 2 extension for product feeds and wanted to share it / get feedback from merchants and developers who deal with Google Merchant Center, Meta catalogs, Bing, TikTok, etc.

**Problem I was trying to solve**

Most stores end up maintaining messy CSV/XML exports or multiple one-off feed tools. I wanted one Magento-native manager that:

- generates feeds from live catalog data

- supports multiple channels/formats

- has a readable admin UI (filters + attribute mappings without raw JSON)

- can schedule regeneration via Magento cron

- supports public or token-protected feed URLs

**What it supports**

- Channels: Google Merchant / Shopping Ads / PMax, Meta Facebook, Instagram, Bing, Amazon-style, TikTok, Pinterest, Snapchat, ChatGPT Shopping, Custom

- Formats: XML, CSV, JSON, TSV

- Filters: stock, categories, product types, visibility, price range, brand, include/exclude SKUs

- Attribute mapping grid + “Load Recommended Mappings”

- Feed URL format: `/feed/{filename}` (optional `?token=`)

- Dashboard, logs, analytics

- Optional OpenAI title/description optimization (encrypted API key)

**Stack / compatibility**

- Magento Open Source / Adobe Commerce 2.4.6+

- PHP 8.2–8.5

- Service contracts, declarative schema, ACL, Magento UI components

**Links**

- GitHub: https://github.com/hiteshagrawal84/Magento2-product-feed

- Adobe Marketplace: Yet to be live

- Download Link : https://www.createswowtech.com/plugin/universal-product-feed-manager-for-magento-2

Happy to answer technical questions (mapping, cron, attribution, ChatGPT feed shape, token security, etc.).

If you use feeds in production:

1. What channel is hardest for you right now?
2. What do current feed extensions get wrong?
3. Anything you’d want before trusting a feed tool on a live catalog?

Thanks — feedback welcome (including “this is overkill / missing X”).