boyceheating.com

• FAIL: 1 junk URL in the XML sitemap: coming-soon stub (/blog/new-website-and-blog-coming-soon). Fetched and confirmed indexable (no noindex): /blog/new-website-and-blog-coming-soon.
• 8 render-blocking stylesheets in <head> - nothing paints until they load.
• 27 blocking scripts in <head> with no async or defer.
• 32 of 32 images have no width/height, so the page shifts as they load.
• A Canadian business declared as US English (lang="en-US", og:locale -); en-CA fits.

How this was measured - The homepage response, robots.txt, the XML sitemap (index plus up to five child sitemaps), and a sample of up to four junk URLs fetched to read their robots meta and X-Robots-Tag. Junk is judged by URL pattern (default posts, -old copies, author and tag archives, uncategorized archives, cart/checkout/account, thank-you pages, coming-soon and email-signature stubs, lorem ipsum and testimonial post types). A sitemap the scan could not fetch is inconclusive, not a pass. Page weight: One HTML response (plus a HEAD probe when the GET loses Content-Encoding). Transfer size is the compressed weight when Content-Encoding is visible (or a gzip estimate when that size is unknown). The uncompressed finding only fires when Content-Encoding is visible and is not gzip/br/zstd — platforms that strip encoding after decompress are treated as inconclusive, not as a fail. Uncompressed markup is reported alongside and only flagged when extreme. Scripts, images and CSS linked from the page are not downloaded.

• Title tag is 13 characters ("Boyce Heating"). 15-65 reads best in results.
• No meta description, so Google writes its own snippet.
• More than one H1 on homepage (4), /furnace-service (2), /maintenance-plans (2), /new-furnace-installation (2) and 3 more.

How this was measured - The homepage plus up to six key pages linked from its navigation (contact, about, services, locations, pricing or membership, then city or service pages). Each is checked for exactly one H1 that is more than the brand name, a title that still names a city or a service once the brand is removed, and word count. Image alt text is read on the homepage. Pages that could not be fetched are left out rather than counted as passing.

• FAIL: LocalBusiness is missing telephone (the phone is visible on the page but not in schema).
• Schema is missing geo and sameAs.
• No Google Maps link or embed, a common local trust and relevance signal.

Capped at 35: no LocalBusiness schema with complete NAP.

How this was measured - Cities are read from visible text on the fetched pages (addresses, '<City> location', 'locations in ...', service-area statements) and from schema addressLocality and areaServed. A dedicated city page means a URL in the sitemap or navigation whose path names the city (product, blog and archive URLs do not count). Schema is every JSON-LD block on the homepage; NAP means name, telephone, and either a matching PostalAddress street (when a street is visible on the pages) or areaServed / locality on a LocalBusiness-type node when no street is shown (service-area-only). Storefronts with a visible street still need that street in schema. This scan cannot see a Google Business Profile.

• 32 images and no srcset - phones download desktop-sized files.

• No online booking or scheduling link, so every lead waits on a callback.
• Enquiry path found via link (/contact-us), not a form on this page.

How this was measured - One HTML response. A <form> on that page scores highest for the enquiry path; same-site contact, booking, quote or referral links, and recognised booking providers, count as a pass but the linked page itself is not fetched. Tel and mailto links are read from hrefs only.

• a generic cookie banner is present, but the tracking scripts are not marked as blocked in the markup - a banner that appears after the trackers have already loaded does not record consent, it announces it.
• No link to a privacy policy anywhere on the page.
• Trackers or cookies are in use but there is no cookie policy linked, so a visitor cannot find out what is set or how to refuse it.
• No terms of use or terms of service linked.

How this was measured - Checked by fetching the page once, server-side, without running JavaScript: which tracking scripts are in the markup, whether a consent tool is present and actually gating them, which cookies the server set on that first response, and which policy pages are linked. Cookies set later by JavaScript are not visible to this scan. These are observations about a single response, not legal advice.