Jerry Coker Insurance
Local agency site with a lead workspace behind it.
Jerry Coker runs a GEICO Exclusive Agency in Spring, Texas. The public site is three pages and has one job, which is to get a local driver to a quote. The interesting half is what sits behind it: the leads those three pages produce have to be searchable, sortable, and impossible to lose.
- Sitemap, breadcrumbs, schema, and the AI index all read one route list
- Leads are searchable and filterable with a per-lead note and event history
- Image delivery tuned against per-transformation hosting costs, not left on defaults
One codebase, two views.
Desktop and mobile are the same components with different breakpoints, not two component trees kept in sync by hand. Every section owns its own spacing rules, which is the part that usually rots first when a site gets edited a year later by someone who was not there for the build.
The business record Google actually reads.
The site publishes a complete local business record: opening hours written in the convention that makes a closed day render as closed instead of disappearing, coordinates at the precision Google asks for, an industry classification code, a price range, and links to the canonical off-site profiles. What is not known is left out rather than guessed, because a wrong founding date is worse than no founding date.
One list of routes drives four systems.
Every indexable page is declared once, with its breadcrumb name, its last significant change date, and a one-line summary. The sitemap, the breadcrumbs, the structured data, and the AI index all read that one list, so a page cannot exist in the navigation and be missing from the sitemap. The change dates carry the UTC offset Texas was actually observing on that date, not a hardcoded one that happens to be right until winter.
A small CRM inside the website.
The admin holds a lead inbox with statuses, search, filters, notes, and an event history per lead, plus reviews, staff, and FAQs the client reorders by dragging. Filtering and search are pure functions with their own tests, so they keep returning the right leads as the volume grows. A database trigger keeps the twenty most recent snapshots of the editable settings, so a mistaken change is recoverable without a developer.
