Saraswati Dental needed the enquiry captured before the visitor left the page — and routed somewhere a busy practice actually reads. The build pairs a validated intake form with a WhatsApp hand-off that arrives already knowing what the patient asked for.
CRM dashboard, captured from saraswati-dental-clinic.vercel.app/dashboard.html.
Every lead, name and number on this dashboard is seeded demonstration data — “Priya Sharma”, sequential placeholder phone numbers, IDs running SDC-001 onward. That is deliberate and it is the right call: the page is publicly reachable with no authentication, so real patient records here would be a data-protection exposure rather than a better demo. Read it as a working surface, not as a real pipeline.
Enquiries arrived by phone during treatment hours, when nobody could answer.
A dental audience searches by treatment — implants, root canal, orthodontics — not by clinic name.
The practice serves a bilingual catchment, so the page had to set Devanagari as competently as Latin.
A structured intake sits in the page rather than behind a mailto link, so the enquiry is typed while intent is highest.
The form validates an Indian mobile number and requires an explicit consent tick before it will submit, so what leaves the page is reachable and permissioned.
Click-to-chat links carry an intent-specific pre-filled message, so the thread opens with the reason for contact already written.
A separate dashboard turns captured enquiries into a worked pipeline — four summary tiles, two charts, and a kanban running the six statuses from new enquiry to closed.
A Dentist entity graph publishes the practice, its hours, its coordinates and its treatment questions in a form search engines and assistants read directly.
Every row was read off the running deployment, not from a project brief. Open the site and view source to check any of it.
A booking with an unreachable number is not a booking. The field constrains input to a valid Indian mobile — ten digits opening 6–9, with an optional country code — and explains the rule in place instead of failing silently after submission.
Verify: View source on the booking form: the phone input carries a pattern attribute, a numeric inputmode, and its own error message element.
A public clinic form attracts bot submissions, and a CAPTCHA taxes every real patient to stop them. An off-screen field that humans never see and scripts routinely fill costs the patient nothing and drops the automated traffic.
Verify: The form contains a visually hidden, aria-hidden field with tabindex -1 and autocomplete off, labelled to be ignored.
The practice contacts patients on WhatsApp afterwards. Making the tick a submission requirement means every record it holds was volunteered on the record, rather than assumed from a pre-checked box.
Verify: The consent checkbox is required; the form will not submit without it.
The dashboard is reachable at a public URL with no login. Seeding it with real enquiry records would have made a demonstration surface into a data-protection exposure — names, phone numbers and treatments for identifiable people, served to anyone who guessed the path. Fictional records make the surface reviewable without putting a single real patient on the open web. The constraint the page is actually under is privacy, not fidelity.
Verify: Open the dashboard and read the records: “Priya Sharma”, sequential placeholder numbers, IDs running SDC-001 onward.
It is an internal surface for the practice, not a page for patients, so it earns no navigation entry and no sitemap presence. Reachable by URL for whoever is given it; absent from the path a visitor takes.
Verify: The public site contains no reference to dashboard.html — search the source and there are zero matches.
Serving Hindi through a fallback renders it visibly worse than the English beside it. Noto Sans Devanagari ships alongside the Latin faces so both scripts are set deliberately.
Verify: The Google Fonts request includes Noto Sans Devanagari with the Latin families.
Not publicly disclosed.
No business outcome has been measured against a stated method and confirmed with this client, so no figure is published. We would rather show you the system and let you judge it than print a number nobody verified.
This build uses the following SheetFlow services:
