Service Page Generator

The Zipper Method is straightforward — services × locations = a page per combination, each targeting a "money keyword" with $3-$20 CPC in paid search. The trap is scale. Service pages are inherently similar (unlike blog posts), and at high counts the near-duplicate matrix trips Google's scaled-content-abuse and doorway-page signals. The same matrix that looks efficient is the pattern those systems demote. The skill caps recommendations at 25-100 pages and pushes back on anything bigger with a prioritised first batch — highest-commercial-value services crossed with the locations that actually have demand.

Every page runs the stand-alone test before it's emitted. Mentally delete the location name from everywhere it appears. Read what's left. If the content collapses into generic boilerplate, the page failed and gets rewritten with real service depth before it ships. Cosmetic variation — a swapped opening sentence, one tacked-on landmark — doesn't pass. Neither does inventing local detail to survive the test, because that breaks the integrity rule and makes the failure worse.

Integrity is non-negotiable. Star ratings, review counts, years in business, jobs completed, named testimonials — every trust signal traces to something the user supplied. If a number is missing, the skill uses a clearly-marked placeholder like [REVIEW COUNT — supply from Google Business Profile] instead of inventing a plausible-sounding figure. A made-up "4.9 stars, 800 reviews" isn't embellishment; on a money page it's a false advertising claim, and in legal, medical, financial, or immigration verticals it's a licensing risk for the client. The skill delivers hard-converting copy through legitimate levers (clarity, strong CTA, real credentials, urgency framing) instead of prohibited promises.

It also routes the SEO payload into the user's actual framework. Title, meta, schema and OG tags only count if they render — frontmatter sitting inert in a component file is a silent failure. Next.js gets the metadata export plus JSON-LD via <script type="application/ld+json">. Markdown gets the site's existing SEO component. Page one is always written fully and held for approval before the bulk run — even when the user says skip approval, because one round-trip beats replicating a flawed template across dozens of pages.