A custom CRM and franchise brand database built for The Franchise Consulting Company. Cedar Web Services
designed it, built it, and operated it in production for nine years, where it was
used every day by more than 500 franchise consultants, hundreds of brand owners, and
the brokerage’s admin team.
years in production
9
consultants using it
500+
fields on every brand record
98
third-party integrations
13
Who uses it
One system, three audiences.
A franchise brokerage sits between people who want to buy a business and the brands
that sell franchises. FCC BrandFinder serves both sides and the team in the middle.
Consultants
Search roughly 300 franchise brands against a candidate’s budget, industry, and ownership preferences. Register leads, run territory checks with the brands, browse and list businesses for resale, and coordinate with each other on an internal board.
Brand owners
Maintain their own listing through a self-service portal with no account and no password. Each brand gets a tokenised link. They edit the fields the brokerage has chosen to expose, upload documents, and manage their resale listings.
Administrators
Decide field by field what each audience sees, add new fields without a deploy, watch the sales pipeline on a CRM dashboard, and audit every email sent and every value changed. Six scheduled jobs kept the data in sync with email, spreadsheets, and the CRM.
The core tool
Every brand, every number, one grid.
Each brand is held as a 98-column record: investment ranges, franchise fees by
unit count, royalty and ad-fund percentages, liquid capital and net worth
requirements, unit counts, average unit volume, ramp-up time, veteran and
first-responder discounts, visa suitability, international availability, and
the contacts for territory checks and lead registration.
The grid searches server-side, filters per column, remembers each consultant’s
column selection, and exports to CSV, Excel, or PDF. A brand profile page lays the
whole financial picture out in tabs, with documents and contacts alongside.
A brand profile. Screenshots on this page are from a demo build with generated data.
Territory checks
Sending email as 500 people, safely.
Before a consultant introduces a candidate to a brand, they have to confirm the
territory is available. Brands want that request to come from the consultant, not
from a generic platform address. That one requirement shaped the most important
piece of automation in the system.
1
A consultant submits a territory check from a form in the app. The message is composed from HTML templates with merge fields: candidate name, territories requested, consultant contact details, private notes.
2
The consultant has already authorised the app against their own Google account, so the app holds an OAuth token for them. That token travels with the queued email row. No shared mailbox, no stored passwords.
3
The email is written to a queue table as QUEUED. Nothing is sent inline, so a slow mail API never blocks the person clicking the button.
4
A background worker runs on a schedule, claims a batch of queued rows, flips each to PROCESSING, and sends it through the Gmail API as the consultant. It lands in the brand’s inbox from a real person and threads in the consultant’s own Sent folder.
5
Success marks the row SENT with a timestamp. Failure returns it to QUEUED with the error recorded, so it retries on the next pass. Every send is logged and searchable by brand, consultant, recipient, and date.
The territory-check log. Every send is searchable and exportable.
The same queue carries resale checks, lead registrations, and reminder emails.
Consent is captured once per consultant, the token rides on the row, and a
background worker claims, sends, and retries. No shared mailbox and no
"on behalf of" in the From line.
The portal
One record, three views, zero deploys.
An admin sees all 98 fields on a brand. A consultant sees the subset relevant to
placing a candidate. The brand owner, in their portal, sees only what the brokerage
has chosen to let them edit. That split is not written into the code. It is two
switches per field in a settings table, edited from an admin screen, and every
page renders against it at runtime.
Admins can also define entirely new fields with their own type, label, and options.
They appear in the grid, the profile, and the portal immediately. This is the
feature that kept the brokerage from needing a developer for routine changes,
which is the real business value of the design.
Field settings: a consultant switch and a portal switch per field, plus custom fields.The brand owner portal, reached by a tokenised link with no login.
Home
Dashboard with profile completeness, announcements, and upcoming events.
My Profile
Edits the full brand record, limited to the fields admins have exposed. Save per field or all at once.
Presentation and Elevator Pitch
The answers that generate the brand’s PDF one-pager and the 30-second pitch consultants use.
My Resales, Consultants, Map
Add or bulk-upload resale listings, browse the consultant roster, and see where consultants are located.
CRM and automation
The jobs that run while nobody is watching.
FCC BrandFinder is wired two ways into GoHighLevel, the brokerage’s CRM, through a
dedicated Python client library covering auth, contacts, opportunities, pipelines,
locations, and lead sources. Around it sits a set of background jobs that keep email,
spreadsheets, forms, and the CRM in agreement without anyone touching them.
Email queue worker
Sends queued territory checks, resale checks, lead registrations, and reminders through each consultant’s own Gmail authorisation, with retries on failure.
Spreadsheet sync
Keeps brand data in step with the Google Sheets the brokerage works from.
Inbox processor
Reads incoming applicant email over IMAP, extracts PDF résumés, and turns them into structured leads with Google Gemini.
Form-to-stage movers
Moves JotForm submissions and completed questionnaires to the matching pipeline stage in GoHighLevel.
Lead-source downloader
Pulls lead-source attribution from GoHighLevel into the local CRM mirror for reporting.
Maintenance scripts
Lead groomer, stale-opportunity marker and emailer, lead-source standardiser, orphan file cleanup, logo downloader. Run on demand.
Event-driven automations run alongside: address geocoding on save, logo discovery,
do-not-call checks on lead capture, CDN upload on every document, and change logging
on every field edit.
Lead intake
Turning an inbox into structured leads.
Applicants email résumés to a shared address. On a schedule, a job connects over
IMAP, walks the new messages, pulls the PDF attachments, and sends them to Google
Gemini to extract name, email, phone, current position and location, and the role
applied for. The result is a structured row that the lead log charts and the CRM
consumes.
Two details make the difference between a demo and something that ran unattended for years:
A global rate limiter across worker threads. A minimum
wall-clock interval between fetches, enforced across every thread, keeps the job
inside Gmail’s bandwidth quota.
Resumable, idempotent imports. Before fetching anything,
the job classifies each message as complete, partial, or new. Complete ones are
skipped, partial ones are retried for just the missing fields. A crash mid-run costs
nothing and a re-run never double-imports.
The lead log, charted daily, weekly, and monthly.
The rest of the platform
Everything a brokerage needs in one place.
Brand marketplace, directory, and map
A card-based marketplace for browsing brands by industry, a logo-first directory, and a map of every active brand, clustered and filterable, with a results panel that updates as you pan and zoom.
Resale marketplace
Businesses for sale with asking price, revenue, EBITDA, cash flow, FF&E value, employee count, and reason for sale. Table and map views, per-brand management, bulk CSV upload, and automatic geocoding of every address.
Consultant directory and map
A roster of consultants with bios, territories, photos, and contact details, plus a map for territory coverage. Supports aliases for consultants operating under another consultant’s brand.
Internal message board
Groups, topics, threaded replies, pinned posts, attachments, and search. Includes a migration that imported years of history from the company’s WordPress intranet.
CRM pipeline dashboard
Lead counts by stage from first contact to contract out, with 7, 14, 30, and 90-day windows, filters by location and lead source, stage-conversion bars, and a consultant leaderboard. Backed by a dedicated Python client library for GoHighLevel.
Brand marketplace
The consultant-facing brand browser: investment range, liquid cash, net worth, ownership model, and area-development availability at a glance, with faceted filters down the side.
Also in the box
Do-not-call checker against two DNC providers, single number or in bulk across CRM contacts
A suite of 16 public forms: NDAs, confidential questionnaires, invoice requests, deal announcements, lead registration, with UK and Ireland variants
Brand presentation generator that turns portal answers into a seven-page PDF one-pager per brand
AI-assisted logo finder that searches, scores candidates, and stores the pick on the CDN
In-app assistant that answers questions about the brand database
Change log recording every field edit with old value, new value, source, and timestamp
Login audit, portal announcements and events, email and territory-check logs with export
Feature flags so sections of the app switch on and off from a settings table
Google sign-in with an email allowlist, role-based navigation, and tokenised no-login access for brand owners
Stack and scale
Plain PHP, built to be operated.
No framework and no build step. Two databases: the main application schema and a
CRM mirror. Server-side DataTables processing for the large grids, Chart.js
dashboards, file storage on a CDN with local fallback, and mPDF for document
generation. Feature flags, field-level permissions, tokenised public access, and
audit logging throughout.
It grew feature-first against live users over nine years, and it shows in places.
That is the honest shape of software that a business actually runs on. Cedar Web
Services has since handed it off and no longer maintains it. The
screenshots here come from a demo build of the same code that runs against a
generated dataset, so no client data appears anywhere on this page.
PHP 8.3 MySQL Python Bootstrap DataTables Chart.js Google OAuth Gmail API Google Sheets API Google Maps Google Gemini GoHighLevel Mailgun Bunny.net CDN mPDF JotForm
PHP files
311
API endpoints
91
database tables across two schemas
40
background jobs running unattended
6
Running your business on spreadsheets and a generic CRM?
We build custom systems that fit the way your team already works, then keep them
running for years. Directly for businesses, and as a white-label partner for agencies.
