LockPoint Start a systems audit

Case study — 01

Frenchy
Smiles

A custom website and the patient acquisition system behind it, for a porcelain veneer practice in Atlanta.

Client
Frenchy Smiles
Dr. David Motoban — Atlanta, GA
Scope
Website · lead capture pipeline · staff CRM · cloud infrastructure
Status
Website live.
CRM built and staged, launch held until the vendor BAAs are signed.

01 — The problem

A five-figure
service with
no front door

Frenchy Smiles does porcelain veneer work in the tens of thousands of dollars per case, with a large Instagram following driving the interest.

What it did not have was anywhere for that interest to land. There was no website of the practice’s own. The link in the bio went to a bare Google Form.

01

No site to sell the work

Someone considering a $30,000 decision had nothing to look at except a form, and no way to see the work, understand the process, or find out what a visit involved.

Positioning
02

Enquiries went into a spreadsheet

The Google Form dropped submissions into a sheet. Nothing ordered them, nothing chased them, and nothing recorded whether anyone had called.

Lead handling
03

Patient information in the wrong places

People type real medical detail into a consultation form. A general-purpose form tool is not built to hold that, and a practice carries the consequences.

Compliance
04

No picture of the pipeline

There was no way to answer how many people were waiting for a call, who had been waiting longest, or which enquiries had already been handled.

Visibility

02 — What we built

Part one:
the website

Eleven public pages, built from scratch. A scroll-driven 3D sequence carries the home page, and the rest of the site does the unglamorous work of answering what a patient actually wants to know before they call.

The Frenchy Smiles home page: dark background, a 3D porcelain tooth with the Eiffel Tower visible inside it, and the headline ‘Porcelain, made to your face’.
Fig. 01Home — desktop
A page from the Frenchy Smiles site explaining how a veneer case is made, laid out as a scroll-driven sequence with large typography.
Fig. 02“The making” — scroll sequence
The Frenchy Smiles services page listing treatments with descriptions in an editorial layout.
Fig. 03Services
The Frenchy Smiles financing page, explaining payment options in a clear editorial layout.
Fig. 04Financing — one of the pages patients reach for first
The Frenchy Smiles home page on a phone: the 3D tooth above the headline, with the whole layout rebuilt for a narrow screen.
Fig. 05Home — iPhoneIts own layout, not a squeezed desktop

The 3D sequence is a decoded image sequence driven by scroll position, with a separate portrait frame set for phones so a narrow screen isn’t downloading and discarding landscape frames. That is the sort of detail nobody notices when it is right.

03 — What we built

Part two:
the system

A website that collects enquiries and does nothing with them is a nicer version of the Google Form. The second half of this project is the software the front desk works in.

The Frenchy Smiles CRM: a left sidebar with call list and digest links, a middle column of waiting enquiries ordered by how long each has waited, and a record panel showing treatment interest, source, contact details hidden behind Reveal links, a dated history of call attempts, quick flags, and outcome buttons for Booked, Call back, No answer and Voicemail.
  1. 1The queue. Every source, one list, longest wait first.
  2. 2Sealed contact details. Behind a Reveal control; each opening is recorded against the staff member.
  3. 3Every attempt, dated. A new staff member picks the conversation up cold.
  4. 4Outcome in one keystroke. Number keys one to five, then the next one is up.
  5. 5Source on every record. Instagram, web form, phone or referral, logged at capture.

Fig. 06Call queue and patient record — staff viewDemonstration data, not patients

What it does

01

One queue, longest wait first

Every enquiry from every source lands in the same list. The order is decided by who has been waiting, not by who happens to be scrolling.

02

Contact details stay sealed until opened

Phone and email sit behind a Reveal control. Opening one is an action recorded against the staff member who did it.

03

Outcomes logged in one keystroke

Booked, call back, no answer, voicemail, not interested — number keys one to five. If logging is slow, it stops happening.

04

Accounts are issued, never self-served

The practice administrator creates staff logins. There is no sign-up page, and no role a person can grant themselves.

The same CRM on a phone: the call queue as a single column of cards with a fixed bottom tab bar for List, Today, Business and Digest.
Fig. 07CRM on a phone

04 — System view

Where an enquiry goes

One path, from the form to the phone call, with every step recorded.

01
Website

Consultation request submitted from any page on thefrenchysmiles.com.

02
Intake API

Posted straight from the browser to the practice’s own endpoint on AWS. The site has no backend of its own to compromise.

03
Sealed store

Identifying details are written into a separately encrypted vault. The analytics side of the database never touches it.

04
Queue

The enquiry appears in the call list, ordered by wait, flagged if it carries a promise or a deadline.

05
Front desk

Staff call, log the outcome, and the notification of the original enquiry has already reached the office by email.

The practice’s clinical records system and its phone system were both left exactly where they were. Neither is integrated in version one, and that was a decision rather than an omission — see below.

05 — Decisions

The calls
that shaped it

A build is mostly decisions. These are the four that changed the shape of this one, each written down at the time so nobody has to re-derive it later.

01

Patient data lives in one place only

Everything touching patient information runs inside a single vendor boundary covered by a business associate agreement. It ruled out several cheaper, more convenient services, which is the point of writing the rule down before you are tempted.

Architecture
02

Don’t integrate what you can’t integrate honestly

The clinical system’s API is partner-gated and the phone system has no public send API. Rather than build something brittle against neither, version one leaves both alone and the practice keeps using them as-is.

Scope
03

Availability is our own source of truth

Appointment availability is held in the system’s own database with a constraint that makes a double-booking impossible at the storage layer, rather than trusting a calendar API to be right.

Correctness
04

Public intake fails closed

Until the system is formally promoted to production it refuses anonymous submissions outright and tells the patient to phone the practice. A half-open front door is worse than a closed one.

Safety

06 — The build

What is actually in it

Counts from the repository, not adjectives.

11

Public pages

On the live site, each with its own metadata and its own mobile layout.

21

Database migrations

Every schema change recorded in order and replayable from empty.

21

Test suites

Covering capture, authentication, roles, booking, notification and erasure.

04

Decision records

The architectural calls above, written down with their reasoning.

07 — Where it stands

Site live.
System staged.

The website is live and taking consultation requests, and the site’s form now posts directly to the practice’s own infrastructure. The legacy Google Form has been retired.

The CRM is finished and deployed, and deliberately not yet carrying real patient data. Opening it is waiting on things that are not engineering: a signed business associate agreement, an attorney review of the contract terms, production sending access from the email provider, and a handful of decisions only the practice can make.

We could have opened it months ago and dealt with the paperwork afterwards. On a system holding patient information, that is the wrong order, and we would make the same call again.

08 — Outcome

What changed

A practice that had no website now has one that matches the price of the work, and a system built around how its front desk actually spends a day.

We are not going to put a conversion percentage on this page. The system has not yet run a month of real patient traffic, so any number here would be invented. When there is something real to report, it will appear here and it will be measured.