AI-native playbook · Franchising

How to make a franchise brand AI-native: a playbook

A staged, sourced path for the company that sells franchises under an FDD, writes the manual and collects royalties. Agents take development follow-up, franchisee questions, field follow-up and local marketing, held to the FDD, each agreement and the manual. Statements about money, fees, defaults and franchisees' employees stay with named people, on the record.

Last reviewed
October 2026
Written for
Franchisors with 20 to 3,000 units and a corporate team of 15 to 400
Reading time
About 35 minutes
On this page
01

The problem: a small team, a big system, and every sales message is regulated

  • $17M

    To go back to Xponential Fitness franchisees under a March 2026 FTC settlement over sales claims and late FDDs, the largest amount the FTC has sent back in a franchise case[1]

  • ~400,000

    FDDs sent each year by about 4,000 franchisors, by the FTC's estimate[2]

  • ~2%

    Corporate overhead as a share of system sales at McDonald's (2.2%) and Wingstop (2.4%), our arithmetic from their 2025 10-Ks[42,43]

  • 10.1%

    Accommodation and food services businesses that used AI in a recent two-week period, against 23.8% of all US businesses[21]

The FTC's $17 million Xponential settlement was about sales talk and paperwork: sellers told prospects that studios typically open within six months when they typically took more than a year, FDDs left out franchisees who had left the system, and some FDDs arrived less than 14 days before signing[1]. The same company separately paid $22.75 million to settle with 509 current and former franchisees[48].

The FTC estimates that about 4,000 franchisors send roughly 400,000 FDDs a year, most of them electronically[2]. Now put an AI agent on the franchising site, answering every prospect at midnight. If it works out a payback period for a curious prospect, it has made an earnings claim, and unless that claim is in Item 19 it breaks the Franchise Rule[7].

A franchisor's corporate team is small next to the system it runs: Wingstop runs 3,056 restaurants with 345 corporate and regional staff[43]. Agents could take on much of that team's work, but the contract and the disclosure rules decide where they may go, and the brand's tools reach into businesses it does not own and people it does not employ.

This playbook is written for the franchisor, not the multi-unit franchisee. Franchisee operations appear where the franchisor's choices reach into them, because that is where most of the law and most of the failures sit.

02

The short version: 60 seconds

03

AI-enabled vs AI-native

AI-enabled

Development uses a CRM with AI lead scoring, marketing uses an AI review tool, field consultants use an audit app with photo checks. Each tool has its own logins, data and rules. The FDD, the agreements and the manual are still PDFs that only people read.

AI-native

The FDD, Item 19, every agreement by version, the manual and the approved claims are something agents read and are held to. Agents do the first pass on development follow-up, disclosure logistics, franchisee questions, audits, local marketing and royalty checks. Named people keep money statements, who gets a franchise, fees, defaults and anything that controls a franchisee's employees, and every agent action leaves a record a regulator could read.

A test: a former franchisee's lawyer writes that your chatbot promised her client a three-year payback. An AI-native franchisor can show, in minutes, which Item 19 text was in force that day, what the agent drafted, what the rule blocked, what was actually sent and which rep called back.

AI in a franchise system sits in two layers, and they need different rules.

Corporate AI
Agents working for the franchisor's own teams: development, legal, field, marketing, finance. The franchisor is the employer and owns the data. Mistakes stay inside the company.
System AI
Tools the franchisor offers or requires at franchised units: ordering, hiring, scheduling, reviews, audits. Here the FDD, fees, data access, security and joint-employer limits apply. Most of the failure cases in this playbook sit in this layer.
04

The missing layer: where OrchKernel fits

No franchisor will replace its franchise management suite, development CRM, audit app, reputation tool or royalty system to become AI-native. Several of those vendors now sell agents of their own. What none of them holds is the whole rulebook: the Item 19 text by state, the waiting periods, each unit's agreement terms, the line on franchisees' employees, and a record of every agent action across all of those tools.

OrchKernel is built to be that layer. It connects to the systems you run and does not replace them; they remain the record for units, agreements, sales and royalties. Details are in the OrchKernel blueprint.

Agents working for the franchisor
  • Development follow-up and scheduling
  • FDD delivery and receipt tracking
  • Franchisee support answers from the manual
  • Field visit prep and corrective actions
  • Local marketing and review replies
  • Royalty reconciliation and reminders
OrchKernel checks each action
  • Rules: waiting periods, Item 19 text, claims
  • Approvals by named role
  • Acting on one person's authority
  • Data access by role and field
  • Human queue
  • Tamper-evident log
Systems the brand already runs
  • Franchise management suite and development CRM
  • E-signature and FDD receipt tools
  • Audit, checklist and learning systems
  • Reputation and listings tools
  • POS and back-office feeds, where Item 11 allows
  • Royalty billing and accounting
The franchise management suite, the POS and the accounting system stay the systems of record. The layer records what each agent asked to do, which rule applied, who approved it and what happened.
05

Who controls what in a franchise system

Most large systems are almost entirely franchised: about 95% of McDonald's restaurants, 98% of Wingstop's and 97% of Yum's 63,285 units[42,43,44]. Mature systems are run by a few hundred multi-unit operators, not thousands of single-unit owners: Wingstop has 186 domestic franchisees averaging about 14 restaurants and 16 years in the system[43]. In our reading, owners at that scale have the counsel and the bargaining power to push back on a tool the brand tries to impose, and to ask what it costs them and what data it reads.

The franchisor sets standards and checks outcomes. The franchisee runs the business and employs the people. Three documents sit under both, and an agent at the franchisor touches all three.

Franchisor

Standards and outcomes. Its agents work here.

  • Brand standards and the operations manual
  • Approved suppliers and required systems (FDD Items 8 and 11)
  • Outcomes it can audit: food safety, cleanliness, guest scores
  • The national ad fund and brand marketing
  • Who gets a franchise, where, and on what terms
The line

Agents send recommendations across it. The franchisee's own manager accepts, changes or ignores them.

Writing schedules, ranking applicants or disciplining staff would cross it[18].

Franchisee

Employees and day-to-day decisions.

  • Hiring, firing and discipline of unit staff
  • Schedules, hours and pay
  • Day-to-day supervision and direction
  • Local decisions the agreement leaves to the owner
  • The unit's profit and loss
Underneath both: the rulebook
  1. The FDD

    23 disclosure items, delivered at least 14 calendar days before a prospect signs or pays, revised within 120 days of year end[3,5]

  2. Each franchise agreement

    Signed per unit; McDonald's agreements generally run 20 years[42]. Terms differ by vintage.

  3. The operations manual

    Part of the agreement, changed by the franchisor over time, and not a route for new fees[8,27]

Where exactly the line falls depends on the agreement, the state and the statute; joint-employer tests differ under wage, discrimination and state law. Treat this as the design rule, and ask employment counsel where your line is.

What this means for AI. Agents belong on the franchisor's side of the line. They can recommend across it. They cannot write a franchisee's schedule, reject a franchisee's applicant or discipline a franchisee's staff, because that is the control that makes a joint employer[18]. Restaurant Brands International already lists franchisees failing to meet "artificial intelligence" requirements among its risks[45], so at least one large franchisor already holds its franchisees to AI requirements.

06

Where the franchisor's hours and money go

Public franchisors' 2025 filings show how the business is paid and how lean the corporate team is.

Wingstop
System
3,056 restaurants, about 98% franchised; $5.343B system sales
How it is paid
Royalty 6.0% and ad fund 5.5% of gross sales. Royalties, franchise fees and other $321.8M; ad fees $247.6M; total revenue $696.9M[43]
Corporate cost, staff and structure
SG&A $128.4M, 2.4% of system sales (our arithmetic); 345 corporate and regional staff
McDonald's
System
45,356 restaurants, about 95% franchised; $139.4B system sales
How it is paid
Rent, royalties, company stores, and fees that recover part of its technology and digital platform costs[42]
Corporate cost, staff and structure
SG&A $3.039B, 2.2% of system sales (our arithmetic)
Yum! Brands
System
63,285 units, 97% franchised; $68.3B system sales; about 1,500 franchisees
How it is paid
Royalties and fees; offers franchisees its Byte by Yum platform[44]
Corporate cost, staff and structure
About 40% of franchised units run under master franchise programs
Xponential Fitness
System
3,097 studios
How it is paid
Royalty typically 7% to 8% of gross sales; marketing fund 2%; technology fees[47]
Corporate cost, staff and structure
About 340 HQ staff, 116 of them part-time
Planet Fitness
System
Franchise and corporate clubs
How it is paid
Royalty 7% at the current rate, 6.7% on average because only about 63% of clubs pay the current rate[46]
Corporate cost, staff and structure
Four-wall adjusted EBITDA margin about 42.7%, or about 35.3% after a 7% royalty

The franchisor's own cost base is small. Corporate overhead runs about 2% of system sales at McDonald's and Wingstop, and Wingstop and Xponential each run about nine units per corporate employee (our arithmetic)[42,43,47]. AI saves corporate hours, but most of the value, and most of the legal exposure, sits in what the franchisor can do for hundreds of franchisees, such as getting signed units open sooner.

The franchisor is paid on sales; franchisees keep the profit. Royalties and ad fund contributions are a percent of gross sales. An AI tool that lifts sales pays the franchisor first. A tool that cuts labor helps only the franchisee's margin.

Technology is now a revenue and cost line. McDonald's recovers part of its technology and digital platform costs through fees[42], Xponential reports technology fee revenue[47], and Yum offers its Byte platform to franchisees "with advantaged economics made possible by the scale of YUM"[44]. FTC staff have named "ever increasing payment processing and technology fees" as a concern when they were not disclosed[8].

07

The franchise lifecycle end to end, and where AI works today

Eight stretches of work run from a lead to a renewal or an exit. Two federal gates sit before signing. At each step some work is already sold and running, some is ready for drafts a person approves, and some should stay with a person.

AI works todayAI drafts, person approvesPerson only
  1. 01Lead
    • AI works today

      Instant reply, qualification, scheduling

    • AI drafts, person approves

      Follow-up emails and texts from the approved claims list

    • Person only

      Any answer about sales, income, profit or payback outside Item 19

    Tools: Development CRM, franchise portals, brokers

  2. 02FDD and discovery
    • AI works today

      FDD version by state, delivery, receipt log, validation-call lists

    • AI drafts, person approves

      Discovery day packs, committee summaries

    • Person only

      Approving the candidate and awarding territory

    Gate: 14 calendar days from FDD receipt before any signature or payment; 7 days after any term the franchisor changed

    Tools: E-signature and receipt tools, CRM

  3. 03Signing
    • AI works today

      Waiting-period and registration checks

    • AI drafts, person approves

      Agreement packages from the current templates

    • Person only

      Changed terms, side letters, fee concessions

    Tools: Contract and e-signature tools

  4. 04Opening
    • AI works today

      Project tracking and slippage alerts

    • AI drafts, person approves

      Site approval packs, opening checklists

    • Person only

      Site approval and opening date commitments

    Tools: Franchise management suite, learning system

  5. 05Field and support
    • AI works today

      Answers from the manual; visit prep; photo checks

    • AI drafts, person approves

      Visit reports and corrective action plans

    • Person only

      Anything that directs a franchisee's employee

    Tools: Audit and checklist apps, support desk

  6. 06Marketing
    • AI works today

      Listings, review reply drafts, local ads from brand templates

    • AI drafts, person approves

      Ad fund reports, off-template ads, menu copy

    • Person only

      Nutrition and allergen content

    Tools: Reputation, listings and social tools

  7. 07Royalties and fees
    • AI works today

      Reconciling sales against POS feeds

    • AI drafts, person approves

      Late-fee notices, sales audit requests

    • Person only

      Any new or changed fee, and waivers

    Tools: Royalty billing, POS feeds, accounting

  8. 08Renewal, transfer or exit
    • AI works today

      Term-date calendar, transfer checklists

    • AI drafts, person approves

      Renewal packages, notices for counsel

    • Person only

      Defaults, terminations, nonrenewals, transfer approvals

    Tools: Contract records, legal

The gates come from the FTC Franchise Rule[3]. "AI works today" means vendors sell it and franchisors run it, not that it is safe without the controls later in this playbook.

Who sells it

We list categories and a few examples, not recommendations. Every performance figure below is the vendor's own claim.

  • Franchise management suites with agents. FranConnect sells development CRM, audits, royalty management and learning, plus agents for sales coaching, field work and franchisee support. It says its customers sell about 29,847 franchises and open about 15,286 units a year[56].
  • Development follow-up. ClientTether sells two-way text, email and call follow-up with agents that answer inbound messages, and cites replies "within 30 seconds"[57].
  • Field audits. Crunchtime uses AI to flag audit photos that fail brand standards[58].
  • Local marketing, listings and reviews. SOCi sells agents for local search, reputation and social to franchise brands such as Jersey Mike's and Sport Clips[59]; Yext manages listings and reviews across what it says are 9M+ locations[60].
  • Guest-facing voice and hiring at units. SoundHound sells phone, kiosk and drive-thru ordering; its revenue rose 99% to $168.9 million in 2025, mostly from acquisitions[51]. Paradox runs high-volume hiring chat for franchise systems and says one franchisee group automates 90% of its hiring process[61].
08

Adoption and where the money went

The Census Bureau asks businesses every two weeks whether they used AI in any function. It does not separate franchisors from franchised units, but it shows the sectors where franchised units cluster[21]:

Accommodation and food services
September 2026
10.1%
Late June 2026
9.7%
Mid-November 2025
7.8%
Other services, including personal services
September 2026
15.0%
Late June 2026
11.5%
Mid-November 2025
9.7%
Retail trade
September 2026
16.5%
Late June 2026
17.4%
Mid-November 2025
13.1%
Administrative and support services
September 2026
20.8%
Late June 2026
20.2%
Mid-November 2025
14.2%
Arts, entertainment and recreation, including fitness
September 2026
22.6%
Late June 2026
18.3%
Mid-November 2025
17.7%
Health care
September 2026
25.5%
Late June 2026
24.7%
Mid-November 2025
21.4%
Educational services, including tutoring
September 2026
30.0%
Late June 2026
30.9%
Mid-November 2025
23.6%
All US businesses
September 2026
23.8%
Late June 2026
Not shown
Mid-November 2025
Not shown

The point is narrow: restaurant and hotel units, the largest franchise sector, adopt AI at well under half the national rate. "Any AI in any function" also counts a manager using a chatbot. We found no independent survey of franchisor AI adoption; the figures that exist are vendor customer claims.

The money went to vendors and brand platforms, not new franchisors. FranConnect, SOCi and Paradox built agents for multi-location brands, and SoundHound grew by buying ordering companies[51]. The largest systems built their own: Yum's Byte by Yum covers ordering, POS, kitchen and delivery, menu management and "inventory and labor management and team member tools", designed to speed up AI adoption[44]. Wingstop runs its Smart Kitchen in all domestic restaurants and requires "continual and unlimited independent access" to POS data[43].

Some failed loudly. Presto, once in hundreds of franchised drive-thrus, disclosed SEC and DOJ inquiries into its AI statements[49] and was delisted in 2024[50].

We found no AI-native franchisor at meaningful scale. For a mid-size brand the realistic path is to make its own operating system (manual, standards, development, support) work with agents, under rules its franchisees can see.

09

The staged path

Six stages that follow the franchise contract. First make the rulebook readable. Then use agents where the franchisor is the only party, then in sales, where the volume and the enforcement are, then in opening and the field, and only then in tools that run inside franchisees' businesses. Durations are rough for a system of 50 to 1,000 units.

  1. 0

    Stage 0: Make the rulebook machine-readable

    Know which version of which rule applies to which unit, prospect and state.

    About 1 to 3 months, longer for a system with many agreement vintages

    What to do

    • Build the contract map: every unit and development deal with its agreement version, royalty and ad fund rates, term dates, state and side letters.
    • Put the FDD, state addenda, Item 19 text and an approved claims list in one controlled place, with versions and effective dates. Version the manual by chapter.
    • List the AI features already on in your tools and mark each as corporate AI or system AI.
    • Read Item 11 and the agreements for what franchisee data the brand may access. Agents get no more than that.

    Why now

    Every later agent depends on this. Only about 63% of Planet Fitness clubs pay its current 7% royalty, so the system average is 6.7%[46]. An agent that reads the current template will be wrong about a third of that system.

    In place first

    • A franchise compliance owner and a manual owner, named.
    • Franchise counsel's sign-off on the approved claims list.

    What to measure

    • Share of units with a complete, checked contract record (target: all)
    • FDD, addenda and Item 19 versions with an owner and effective date
    • Manual chapters with a named owner

    Common mistakes

    • Pointing an agent at the current FDD and assuming it describes what existing franchisees signed.
    • Skipping Item 11, then building a dashboard on data the agreement never gave the brand.
  2. 1

    Stage 1: The corporate back office

    Agents work where the franchisor is the only party and mistakes stay inside.

    Starts once the contract map and manual versions exist

    What to do

    • Franchisee support answers from the approved manual, with the chapter cited and a handoff to a person.
    • Royalty and ad fund reconciliation against POS sales; reminders on each unit's own terms.
    • Drafts for the annual FDD update: Item 20 tables from the unit ledger, the former-franchisee list, the registration calendar.

    Why now

    Errors stay internal and can be reversed. The FDD update is due within 120 days of year end[5], and a missing former-franchisee list was part of the Xponential case[1].

    In place first

    • Stage 0's contract map.
    • Agent data access set to what Item 11 allows.

    What to measure

    • Support questions answered without a person, and the share a person later corrected
    • Days to close the monthly royalty reconciliation
    • FDD update filed before day 120; Item 20 discrepancies found before filing

    Common mistakes

    • Letting the support agent improvise when the manual is silent.
    • Treating a franchisee's question about a fee as routine. Fee questions go to a person.
  3. 2

    Stage 2: Franchise development

    First stage that speaks to prospects

    Lead follow-up, disclosure logistics and waiting periods, under the Item 19 rule.

    After the claims list and Item 19 rules are tested on the back-office agent

    What to do

    • Lead response, qualification and scheduling.
    • FDD version chosen by the prospect's state and the unit's location; delivery and receipt logged; waiting periods checked before any agreement or deposit link.
    • Validation-call lists from Item 20, and discovery day logistics.

    Why now

    Revenue starts here and speed of reply matters. So does enforcement: Xponential's $17 million case turned on sales claims and late FDDs[1].

    In place first

    • The Item 19 text per state, and a rule that blocks any other sales or profit figure.
    • Recorded consent before calls and texts, including AI voice[17].
    • The list of franchise sellers named in Item 23, and rules for brokers[7].

    What to measure

    • Lead response time; days from FDD receipt to signing
    • Waiting-period exceptions (target: zero)
    • Earnings questions routed to a person

    Common mistakes

    • A chat agent that helps with payback math.
    • Letting a prospect who wants to move fast sign or pay a deposit before day 14.
    • One FDD for every state.
    • Brokers running their own AI scripts outside the brand's claims list.
  4. 3

    Stage 3: Opening and field operations

    Track openings, prepare visits and close corrective actions, without managing franchisees' staff.

    Can overlap with Stage 2 once the field team has a written line on what it may direct

    What to do

    • Opening project tracking from site to grand opening, with slippage alerts.
    • Visit prep and report drafts; audit findings turned into corrective actions with owners and due dates.
    • Training assignments and tracking.

    Why now

    The gap between signing and opening is large and costly. Xponential had 832 studios committed and not yet open against 2,606 open in North America at the end of 2025[47]. Field consultants are the brand's main point of contact with each owner, and today much of their week goes to visit reports and chasing corrective actions.

    In place first

    • Franchisee consent and Item 11 coverage for any data pulled from unit systems.
    • A written line: consultants advise the owner and do not direct the owner's employees.

    What to measure

    • Median days from signing to opening; openings on the promised date
    • Corrective actions closed on time; repeat findings
    • Consultant hours on coaching against paperwork (your own time study)

    Common mistakes

    • Using opening forecasts in sales conversations. A timeline is a claim; it was one in the Xponential case.
    • Field agents assigning tasks to a franchisee's employees by name.
  5. 4

    Stage 4: System tools at franchised units

    Where fees, data security and joint-employer limits apply

    Offer tools inside franchisees' businesses by pilot and choice, not by mandate.

    After the advisory council has reviewed a pilot design, fees and data flows

    What to do

    • Local marketing from brand templates, with approvals for anything off-template.
    • Listings and review replies across every unit.
    • Hiring help, ordering and labor recommendations, menu content, offered to franchisees.

    Why now

    McHire, Presto and the drive-thru voice tests all happened here, and restaurant units adopt AI at under half the national rate[21]. Franchisees, not the brand, will set the pace.

    In place first

    • Advisory council review; a pilot with opt-in units, a written exit and matched comparison units.
    • Counsel's review of disclosure and any fee[8].
    • A security review of the vendor; recommendations only for anything touching franchisees' employees.

    What to measure

    • Adoption by choice; opt-outs
    • Pilot units against matched non-pilot units on sales, labor and guest complaints

    Common mistakes

    • Mandating before measuring.
    • Calling a tool automated when people do much of the work behind it.
    • Adding a technology fee through the manual.
  6. 5

    Stage 5: The AI-native operating model

    Roles, councils and the FDD cycle built around agents.

    After a full FDD cycle with Stages 1 to 3 running

    What to do

    • Staff the compliance owner, system data steward and manual owner roles, and give the advisory council a standing AI agenda.
    • In each FDD cycle, review which agents speak to prospects, which tools franchisees use, and what claims, fees and data access changed.

    Why now

    Corporate overhead at large franchisors is small next to the system: about 2.2% of system sales at McDonald's and 2.4% at Wingstop (our arithmetic)[42,43]. Cutting that team saves little. The return shows up in units opened on the promised date, clean FDD cycles and franchisees choosing the brand's tools.

    In place first

    • Two quarters of stable results and clean logs from earlier stages.

    What to measure

    • Units per corporate employee; corporate overhead as a share of system sales
    • Franchisee satisfaction from an independent survey; regulator inquiries and disputes

    Common mistakes

    • Cutting field consultants before franchisees agree the agents help them.
    • Leaving the advisory council out until a mandate is ready.
10

Your first 90 days

Stage 0 for the whole system, then one corporate agent and the start of development. Nothing in the first 90 days touches a franchisee's business or employees.

  1. Days 1 to 30

    Build the contract map for every unit and development deal. Collect the current FDD, state addenda and Item 19 in one controlled place. Write the approved claims list with counsel: opening timelines, support promises, awards, anything about AI. Inventory the AI already switched on in your tools and sort it into corporate and system layers. Name the compliance owner and the manual owner.

  2. Days 31 to 60

    Run one corporate agent: franchisee support answers from the manual, with the chapter cited and a handoff to a person, read-only on every system. Turn on the FDD receipt log and the 14-day and 7-day rules in the development CRM. Start a time study of consultants and development reps so Stage 3 has a baseline.

  3. Days 61 to 90

    Add development follow-up under the claims rules: no earnings statements outside Item 19, consent before any text or call. Read a sample of agent conversations each week. Take the results to the franchisee advisory council and agree how any system pilot would be chosen, paid for and stopped. Set Stage 3 targets from the time study.

11

How roles change

This is our reading of the evidence, not a survey finding. Most of these changes fall on a corporate team that is already small. What does not change: franchisees own their units, their staff and their local decisions.

  1. 1

    Franchise development

    Agents start and qualify conversations, book calls, send the right FDD and track receipts. Reps spend their time on discovery days, validation and judging fit, with a hard line on what can be said about money. Their pipeline stops depending on how fast they answer texts at night.

  2. 2

    Franchise compliance owner

    A new role or a bigger one. Owns the FDD calendar, state registrations, the approved claims list, Item 19, the seller list and the receipt log, and decides what the agents may say. The Xponential case shows what it costs when this slips[1].

  3. 3

    Field consultants

    Fewer hours writing visit reports and chasing corrective actions; more time coaching owners. A consultant might cover more units, but no public baseline exists for units per consultant, so measure your own before and after.

  4. 4

    Brand standards and manual owner

    The manual becomes a maintained, versioned source that agents read and answer from. Any change that adds a cost goes to counsel first[8,27].

  5. 5

    System data steward

    Owns what franchisee data the brand may access under Item 11 and the agreements, how it is shared, and which agents see which fields. Answers the franchisee who asks what the brand's tools can see.

  6. 6

    Franchisee advisory council

    Becomes the forum for system AI: which pilots run, who opts in, what it costs, what data flows and how a pilot ends. Taco Bell described its voice AI decisions as made "in partnership with our franchisees"[62].

12

What not to fully automate

An agent can prepare each of these. A named person with authority decides, and the decision is recorded.

Any statement to a prospect about sales, income, profit or payback
Why it stays with a person
It must match Item 19 and be substantiated[4,7]. The Xponential case shows how far sales talk can travel[1].
Approving a candidate as a franchisee; awarding territory
Why it stays with a person
A judgment of fit, capital and territory that binds both sides for the whole agreement term. Whether discrimination law reaches franchise sales, and how far, is to be confirmed state by state.
Releasing an FDD, a state addendum or an amendment
Why it stays with a person
Version and timing errors are violations in themselves[3,5].
New or changed fees, required purchases or required technology
Why it stays with a person
Fees added through the manual without disclosure are the pattern regulators named[8,27].
Defaults, cure notices, terminations and nonrenewals
Why it stays with a person
Good cause and cure periods differ by state: at least 60 days in California, no more than 30 required in Washington[26,28].
Transfers and renewals
Why it stays with a person
Terms differ by agreement vintage; the franchisor's approval rights and fees apply.
Hiring, scheduling, pay and discipline of franchisees' staff
Why it stays with a person
The franchisee is the employer. Control over these terms is what makes a joint employer[18], and AI hiring laws apply.
Mandating a new tool at franchised units
Why it stays with a person
Item 11 disclosure, cost, security and franchisee consent all apply[4,13].
Pricing recommendations franchisees feel pressed to follow
Why it stays with a person
California's pricing algorithm law reaches coercion[30].
Menu, nutrition and allergen content
Why it stays with a person
Nutrient declarations need a reasonable basis[20], and a guest's safety depends on them.
Replies to franchisee complaints, association letters and regulator inquiries
Why it stays with a person
Nothing may discourage franchisees from talking to regulators[9], and these letters often become evidence.
Public statements about the brand's own AI
Why it stays with a person
Presto's regulators looked at its AI claims[49], and the FTC says there is no AI exemption[11].
13

The rules that bite

Ten groups of rules recur for franchisors. The first three govern what the brand says and charges; the rest govern the tools it puts into franchisees' businesses. Each lands on something an agent does.

The Franchise Rule and earnings claims

Timing
The FDD must reach the prospect "at least 14 calendar-days before the prospective franchisee signs a binding agreement with, or makes any payment to, the franchisor", and terms the franchisor changes unilaterally need at least seven calendar days[3]. The FDD is revised within 120 days of year end and amended quarterly for material changes[5].
Earnings claims
A franchisor that makes a financial performance representation must have a reasonable basis and written substantiation and must state it in Item 19[4]. The FTC's definition covers oral, written and visual statements, and "includes a chart, table, or mathematical calculation that shows possible results based on a combination of variables". A broker cannot make one unless it already appears in Item 19[7]. No seller may say anything that contradicts the FDD[6].
Items 8 and 11
Item 8 covers required purchases, including "computer hardware and software", and the franchisor's revenue from them. Item 11 says whether the franchisor requires POS or computer systems, what they cost, whether the franchisor has independent access to the data in them and any limits on that access, and how the ad fund is spent[4].
Enforcement
The FTC announced the Xponential settlement on 18 March 2026: $17 million to go back to franchisees, which the FTC called the largest amount ever to go back to consumers in a franchise case[1].

For agents: An agent that tells a prospect what a unit might earn, or works out a payback period, is making a financial performance representation. Agents quote Item 19 as filed, and waiting periods are counted in calendar days.

Fees, technology fees and the operations manual

FTC staff guidance
"If a franchisor imposes or collects a new fee, through its operating manual or otherwise, that was not disclosed in the FDD and included in the franchise agreement, the franchisor may be engaging in an unfair act or practice." The guidance names "ever increasing payment processing and technology fees". It is staff guidance and does not bind the Commission[8].
FTC policy statementPart to be confirmed
Contract provisions may not restrict franchisees from talking to the FTC or other regulators[9]. It was adopted 3-2 over the dissents of Commissioners Holyoak and Ferguson[10]. Both documents were still listed on ftc.gov in October 2026; whether the current Commission relies on them is to be confirmed.
WashingtonPart to be confirmed
"A franchisor cannot impose a fee through the operations manual or otherwise, without pre-sale disclosure in the Franchise Disclosure Document."[27] California's regulator has published guidance on the same question; we could not open it, so its terms are to be confirmed.

For agents: A required AI tool with a new monthly fee, added through the manual, is the pattern the guidance describes. Any draft that adds a cost goes to counsel.

State registration, relationship laws and brokers

RegistrationPart to be confirmed
California, Minnesota, New York and Washington make it unlawful to offer or sell a franchise until the offer is registered or exempt[22,23,24,25]. Hawaii, Illinois, Indiana, Maryland, Michigan, North Dakota, Rhode Island, South Dakota, Virginia and Wisconsin are commonly listed too; those, and NASAA's multi-state forms, are to be confirmed.
Termination and curePart to be confirmed
California: no termination before the end of the term except for good cause, with a cure period of at least 60 days and generally no more than 75[28]. Washington: good cause is also required, but the opportunity to cure "in no event need be more than thirty days", and nonrenewal notice rules tie to goodwill compensation[26]. Other states' good-cause, notice and nonrenewal rules are to be confirmed.
Brokers (California)Part to be confirmed
SB 919 requires franchise brokers to register. It becomes operative on the later of 1 July 2026 or one year after the Legislature funds it, and implementation depends on that funding[29]. Whether the money has been appropriated is to be confirmed.
State enforcement
Maryland entered a consent order against Xponential on 14 April 2026, with an administrative penalty and required disclosures[48].
Outside the US
Australia's new Franchising Code of Conduct started on 1 April 2025, with some rules applying from 1 November 2025[41]. Other countries' franchise codes are out of scope here.

For agents: Which FDD a prospect gets depends on where they live and where the unit will be. A 30-day cure may be enough in Washington and too short in California. Agents choose versions and periods by rule; people sign.

Joint employer: the line system AI can cross

NLRB
A business is a joint employer if it shares or codetermines essential terms such as "wages, benefits, hours of work, hiring, discharge, discipline, supervision, and direction" through "substantial direct and immediate control". Setting "the objectives, basic ground rules, or expectations" under a contract is not that control[18]. The Board restored this text effective 27 February 2026, after a court vacated its broader 2023 rule[19].
What franchisors say
Restaurant Brands International says it could be liable for franchisees' labor violations[45]. Wingstop names "vicarious liability, joint employer, or other theories"[43].
Other lawsPart to be confirmed
Wage and hour, discrimination and state laws use their own tests. How each applies to a brand-provided AI tool is to be confirmed with employment counsel.

For agents: A brand agent that writes franchisee schedules or rejects applicants is exercising control. The same agent sending a recommendation the franchisee's manager accepts or changes is closer to setting ground rules. Design for the second.

Systems the franchisor requires: security and data

FTC v. Wyndham
Wyndham required its franchised hotels to buy and configure property systems to its specifications and ran the network connecting them. Three intrusions in 2008 and 2009 led to more than $10.6 million in fraudulent charges, and the court upheld the FTC's authority to treat poor security as unfair[13]. Wyndham settled in December 2015[14].
Consumer privacyPart to be confirmed
Loyalty and app data often sit with the franchisor while orders happen at franchised units. Who is the business and who is the service provider for each data flow under the CCPA is to be confirmed with privacy counsel. California's rules on automated decision-making technology took effect on 1 January 2026, with some duties phasing in later[34].

For agents: A brand that mandates a tool and connects it to franchisees' systems answers for its security. Agents see only the fields their role needs, and every read is logged.

Pricing tools shared across franchisees

California AB 325Part to be confirmed
Unlawful to use or distribute a "common pricing algorithm" as part of a conspiracy to restrain trade, or to coerce another person to adopt a price or term it recommends. A common pricing algorithm is one used by two or more persons that uses competitor data to recommend or influence a price[30]. The bill was chaptered on 6 October 2025 with no urgency clause or later date, so by California's default rule it took effect on 1 January 2026 (our reading). The text does not define competitor or mention franchisees; whether franchisees of one brand count as competitors is to be confirmed with antitrust counsel.

For agents: A brand pricing tool fed with other units' prices and pushed hard enough to count as coercion sits close to this text. Recommendations carry no penalty for ignoring them.

Menus and nutrition

FDA menu labeling
Chains with 20 or more locations under the same name selling substantially the same menu must post calories, must have a reasonable basis for nutrient declarations, and must give FDA recipes and signed certifications on request[20].

For agents: An agent writing a limited-time offer or menu board copy can introduce a calorie or allergen error at every unit at once. Menu and nutrition content needs the R&D or QA owner's approval, with the substantiation attached.

What AI says to guests and prospects

Reviews
The FTC's reviews rule bans fake and AI-written reviews, review incentives conditioned on sentiment and suppressing negative reviews[15]. The FTC's order against Rytr, an AI review-writing tool, was set aside in December 2025; the rule still stands[16].
Calls and texts
AI-generated voices count as "artificial" under the TCPA, so its consent rules for artificial-voice calls apply[17]. Xponential settled a putative nationwide TCPA class action in 2025[48].
AI income claims
In Operation AI Comply the FTC acted against schemes promising AI-powered income: "There is no AI exemption from the laws on the books."[11]
Overstating the AIPart to be confirmed
Presto disclosed an SEC investigation and a DOJ request about its statements on its AI technology, while most of its drive-thru orders involved a person[49]. A reported SEC settlement with Presto in January 2025 is to be confirmed against the SEC order.

For agents: One central tool answers reviews for every unit, and one agent texts every lead. Agents never post or gate reviews, never text without consent, and describe the brand's AI only from the claims list.

Hiring tools the brand provides

AI hiring laws
New York City requires bias audits and notices for automated employment decision tools[35]. Illinois amended its Human Rights Act on AI in employment from 1 January 2026[36]. Colorado's SB26-189 applies from 2027[37], and California's civil rights rules on automated decision systems took effect in October 2025[33]. California's broader SB 7 was vetoed[32].
Who carries themPart to be confirmed
The franchisee is usually the employer using the tool. A franchisor that selects, configures and mandates it may be treated as an agent, the theory tested against a vendor in Mobley v. Workday[38]. Who carries the bias audit in a franchise system is to be confirmed.

For agents: Brand hiring tools recommend; the franchisee's manager decides. The brand's agents do not see applicant fields they do not need.

Labor rules that shape franchisee tools

California fast food wage
$20 an hour from 1 April 2024 at fast food restaurants in national chains, with a Fast Food Council that can raise it[31]. This is one reason franchisees ask for labor-saving tools.
Predictable schedulingPart to be confirmed
Seattle requires schedules 14 days ahead for hourly staff at retail and food service employers with 500 or more employees worldwide[39], and Oregon requires 14 calendar days[40]. New York City, San Francisco, Los Angeles, Chicago and Philadelphia have their own rules, and how each counts franchisees' employees is to be confirmed.

For agents: A scheduling recommendation pushed by the brand must respect each unit's local rules, and the franchisee's manager publishes the schedule.

14

When it goes wrong

Real cases first, with what each teaches a franchisor and the control point that answers it.

  1. McHire: a hiring tool the brand chose, 2025

    A hiring chatbot used by 90% of McDonald's franchisees exposed 64 million applicant records through a default "123456" admin login[55]. The vendor, Paradox, is now part of Workday[61].

    The lesson: When the brand picks the tool its franchisees hire with, the breach carries the brand's name, whoever built the software.

    Control points: Tool onboarding (13), data access by role and field (10)

  2. Presto voice AI at franchised drive-thrus, 2023 to 2024

    In early 2024 Presto disclosed that its voice AI sites used a person to enter orders "in all instances" pending a rollout; at the 44 sites on its most advanced version, about 30% of orders needed no person. The SEC and DOJ looked into its AI disclosures[49], and it was delisted in September 2024[50].

    The lesson: Measure what the AI really does at the unit before calling it automation in an FDD, a sales deck or a press release.

    Control points: Approved claims register (3)

  3. Taco Bell and McDonald's voice ordering tests

    In August 2025 Taco Bell said it was rethinking voice AI after errors and pranks (secondhand account)[62]. McDonald's ended its IBM order-taking test in 2024; details are to be confirmed[63].

    The lesson: Pilots at franchised units need an exit and franchisees' agreement before any system mandate.

    Control points: System rollout gate (12)

  4. Xponential Fitness, 2023 to 2026 (not an AI case)

    Timeline claims, a hidden litigation history, missing former franchisees and late FDDs: $17 million in FTC redress for franchisees[1], $22.75 million to 509 franchisees, and a Maryland consent order[48].

    The lesson: Every sales message is regulated speech. A development agent can repeat a bad claim to thousands of prospects faster than any rep.

    Control points: Earnings statements (2), claims register (3)

  5. FTC v. Wyndham, 2008 to 2015 (not an AI case)

    Wyndham specified and managed its franchised hotels' property systems; three breaches led to more than $10.6 million in fraud[13].

    The lesson: Mandating and connecting franchisees' systems makes the franchisor answerable for their security.

    Control points: Data access (10), tool onboarding (13)

  6. Rytr and AI-written reviews, 2024 to 2025

    The FTC alleged an AI tool generated fake reviews; the order was set aside, but the rule against fake reviews stands[15,16].

    The lesson: A central review agent working for hundreds of units must never write or gate reviews.

    Control points: Reviews and local marketing (8)

We did not find a published case of a franchisor's AI agent making an earnings claim to a prospect, or of a brand AI tool being the basis of a joint-employer finding. We do not claim these have happened. The scenarios below are what an agent with too much freedom could do.

Agent failures to design against

Each is a scenario, not a reported case, and each maps to one of the control points.

The helpful payback math. Asked how long it takes to earn back the investment, the chat agent combines Item 7 costs with a sales figure from a press release and answers "about three years".
What stops it
Only the Item 19 text for that state may be quoted; the question goes to the development queue (control 2).
The prospect who wanted to move fast. A prospect asks to sign before a lease option expires, and the agent books the signing call and sends the deposit link 12 days after FDD receipt.
What stops it
Agreement and deposit links stay blocked until 14 calendar days after the logged receipt (control 1).
The manual update that added a fee. Drafting the technology chapter, an agent adds "a monthly AI subscription of $149, paid to the franchisor".
What stops it
Any manual change that adds a cost or a required purchase goes to counsel first (control 5).
The schedule the brand wrote. A brand labor agent pushes next week's schedule straight into franchisees' scheduling tool.
What stops it
Brand agents cannot write to franchisee schedules; they send recommendations the manager accepts or changes (control 9).
The default notice with a 30-day cure. An agent drafts a California notice from a Washington template.
What stops it
Counsel approves every default notice, and a rule checks the cure period against the state (control 7).
Review requests only to happy guests. A reputation agent sends review links only after five-star ratings.
What stops it
Requests go to all eligible guests or none; agents never post reviews (control 8).
Unit margins in a benchmark email. An analytics agent shares one owner's margins, name visible, with the whole system.
What stops it
Unit financials are visible to that unit's consultant and finance; benchmarks are aggregated (control 10).
The price nudge. A pricing assistant warns that units "not following the recommended price will be flagged".
What stops it
Recommendations carry no penalty or flag, and counsel approves the data (control 11).
The limited-time offer's calories. A menu agent drops an allergen and reuses a smaller portion's calorie count.
What stops it
The R&D or QA owner approves menu content, with substantiation attached (control 6).

One scenario, step by step: an earnings question at 11:40 p.m.

  1. 01Approved text· People

    The compliance owner loads this year's Item 19, the state addenda and the claims list, with effective dates.

  2. 02Question· Development agent

    At 11:40 p.m. a Minnesota prospect asks what a unit makes and how long until she earns back $400,000. The agent drafts a payback period.

  3. 03Claims rule· OrchKernel

    The figure is not in the Item 19 text for Minnesota. The rule blocks it.

  4. 04What the prospect gets· Development agent

    Item 19 as filed, and an offer of a call with a named rep.

  5. 05Human queue· OrchKernel

    The question lands with that rep, marked as an earnings question.

  6. 06Call· People

    Next morning the rep calls, stays inside Item 19 and points her to the franchisees listed in Item 20.

  7. 07Record· OrchKernel

    The blocked draft, the reply, the rule and the call note go into the tamper-evident log.

A scenario, not a reported case. The FTC counts "a chart, table, or mathematical calculation" as a financial performance representation[7], so a payback estimate is one.
15

The OrchKernel blueprint for a franchise brand

OrchKernel is the layer between AI agents and the systems a franchisor runs, as drawn in the missing layer. Agents send their actions through it; it checks the rules for that prospect, unit and state, holds what needs a person, and records what happened.

What it is not. OrchKernel is not a franchise management suite, a CRM, a POS or a royalty system, and it does not write your FDD. It governs only the actions sent through it: a vendor's own agent that texts leads directly is outside it unless routed through it.

The mechanisms

Approvals
The action waits for a named person, who sees exactly what will go out: the manual chapter with the new cost marked, the default notice with its cure period. It runs once, as approved.
Rules
Checked at the moment of action, the same way for every agent: no agreement link before the waiting periods end, no sales or profit figure outside the Item 19 text for that state, no text without recorded consent. A rule allows, holds or denies, with a reason.
Acting on a named person's authority
Development follow-up runs as the named rep, inside that rep's territory. Field follow-up runs as the assigned consultant, for that consultant's units only. The agent has no more access than the person.
Data access by role and field
A unit's P&L is visible to its consultant and finance; everyone else sees aggregated benchmarks. Applicant and employee fields stay hidden from corporate agents.
Tamper-evident audit log
Every request, rule result, approval and outcome, chained so an edited or deleted entry shows: which FDD version went to whom and when, what an agent was about to say and what was blocked. It is the record you need when someone says "you told me".
Human queue
Earnings questions, franchisee complaints, regulator contact and anything that looks like a decision about a franchisee's employee land with a named owner and a response time.
Connections to your systems
The brand connects its franchise management suite or development CRM (FranConnect, ClientTether, Salesforce or others), e-signature and FDD receipt tools, audit apps (Crunchtime, RizePoint or others), reputation tools (SOCi, Yext or others), POS feeds where Item 11 allows, and royalty billing and accounting. OrchKernel holds the credentials so agents never do.

Sixteen control points

Where a franchisor needs a control whatever tools it uses, who owns it, and what enforces it.

Selling franchises

01
FDD delivery and waiting periods: right version and addendum, receipt logged, 14 calendar days, 7 after changed terms
Owner or approver
Franchise compliance owner
What enforces it
Rules pick the version by state and block agreement and deposit links until both periods pass; the log keeps version, send and receipt.
02
Earnings statements to prospects: only the Item 19 text for that state
Owner or approver
Compliance owner; development lead
What enforces it
Rule blocks any other sales or profit figure and sends the question to the human queue.
03
Approved claims register: opening timelines, support, awards, the brand's AI
Owner or approver
Compliance owner; marketing
What enforces it
Rule: register text only. Approval to add a claim.
04
Who may sell: the Item 23 list, brokers, broker registration
Owner or approver
Development lead
What enforces it
Agents act on a named seller's authority. Brokers' own tools sit outside OrchKernel.
14
Consent for calls and texts; AI voice disclosure
Owner or approver
Marketing; development
What enforces it
Rule: no call or text without consent on record.

Running the system

05
Fees, required purchases and required technology, including through the manual
Owner or approver
General counsel; CFO
What enforces it
Approval: any draft that adds a cost waits for counsel.
06
Menu, nutrition and allergen content
Owner or approver
R&D or QA owner
What enforces it
Approval with substantiation attached.
07
Defaults, cure notices, terminations, renewals, transfers
Owner or approver
General counsel; COO
What enforces it
Rule checks the cure period by state; counsel approves.
08
Reviews and local marketing
Owner or approver
Marketing lead
What enforces it
Rules: requests to all guests or none, no agent-posted reviews; approval off-template.
11
Pricing recommendations to franchisees
Owner or approver
General counsel
What enforces it
Approval of data sources; rule: no penalty tied to adoption.
12
System rollout gate: pilot, opt-in, exit, council review
Owner or approver
COO; advisory council
What enforces it
Approval before agent actions reach units outside the pilot.

Franchisees' people and data

09
Franchisees' employees: hiring, schedules, pay, discipline
Owner or approver
COO; employment counsel
What enforces it
Rule: no write access, recommendations only; apparent decisions go to the human queue.
10
Franchisee financials, guest, applicant and employee data, only as Item 11 allows
Owner or approver
System data steward; IT security
What enforces it
Data access by role and field; every read logged.
13
Tool onboarding: security, data location, hiring-tool roles, FDD disclosure
Owner or approver
IT security; counsel
What enforces it
Approval before an agent uses a new tool. A vendor's own security is outside OrchKernel.

Records and regulators

15
Complaints, association letters, regulator inquiries
Owner or approver
General counsel
What enforces it
Human queue only.
16
Decision records: what each prospect received, who approved fees, notices and claims
Owner or approver
Compliance owner
What enforces it
The tamper-evident log; people set retention.

What belongs elsewhere

Systems of record
The franchise management suite, the POS and the accounting system keep units, agreements, sales and royalties. OrchKernel does not replace any of them.
Drafting the FDD and judging earnings claims
Franchise counsel. OrchKernel can hold the approved Item 19 text and block anything else; it cannot decide what counts as a claim.
Joint-employer exposure
Employment counsel decides where your line is. OrchKernel can keep brand agents to recommendations.
Vendor security
A flaw inside a vendor's product, like the McHire default password, is for the vendor and your security review. OrchKernel limits what its own agents can reach.
Bias audits of hiring tools
HR, the franchisee as employer, and independent auditors.
Antitrust review, food safety testing, franchisee consent
Counsel, R&D and QA, and the advisory council under the agreement and the FDD.

OrchKernel is source-available under the Business Source License and runs on your own servers, so your counsel and your franchisees' advisers can read the code that enforces these controls.

16

Scorecard by stage

Record your baseline before Stage 1, then track the same numbers at each stage. Industry benchmarks for lead conversion, signing to opening, units per consultant and support volume are paid (FRANdata[53]), vendor claims, or do not exist. We have not quoted any we could not source.

Complete contract recordsStage 0
How to count it
Units and development deals with agreement version, rates, term dates and state checked
Public benchmark
No public benchmark. Internal target: all of them
Support answered from the manualStage 1
How to count it
Questions answered without a person, and the share a person later corrected
Public benchmark
No public benchmark found
FDD update on timeStage 1
How to count it
Days before the 120-day deadline the update was finished
Public benchmark
The legal floor is day 120[5]; no benchmark
Lead response and conversionStage 2
How to count it
Minutes to first reply; lead-to-signing rate
Public benchmark
No independent benchmark. ClientTether cites a customer going from 9% to 24% lead conversion (vendor claim)[57]
Waiting-period exceptionsStage 2
How to count it
Agreements or payments before 14 calendar days, or 7 after changed terms
Public benchmark
The rule sets the target at zero[3]
Earnings questions routedStage 2
How to count it
Questions about sales or profit sent to a person, and drafts the claims rule blocked
Public benchmark
No public benchmark
Signing to openingStage 3
How to count it
Median days; share of units open on the promised date
Public benchmark
None across the industry. One filing signal: Xponential had 832 committed studios against 2,606 open[47]
Units per field consultantStage 3
How to count it
Units covered per consultant, and coaching hours against paperwork
Public benchmark
No public benchmark found
Pilot resultsStage 4
How to count it
Pilot units against matched units on sales, labor and complaints; opt-outs
Public benchmark
No independent benchmark; vendor case studies only
Review response rateStage 4
How to count it
Share of reviews answered, and time to answer
Public benchmark
Vendor only: a SOCi customer went from 18% to 99% (vendor claim)[59]
Corporate overheadStage 5
How to count it
SG&A as a share of system-wide sales
Public benchmark
Filing examples: 2.2% at McDonald's and 2.4% at Wingstop, 2025 (our arithmetic)[42,43]
Units per corporate employeeStage 5
How to count it
Units in the system over corporate and regional staff
Public benchmark
Filing examples: about 9 at Wingstop and Xponential (our arithmetic)[43,47]
Franchisee satisfactionStage 5
How to count it
An independent survey, the same one each year
Public benchmark
Paid surveys exist; brand results are not public[54]
17

What we could not find

Open questions from our research. If you have good data on any of them, write to support@prefero.ai.

  • An independent survey of AI use by franchisors. The IFA's outlook counts about 832,500 franchise establishments in 2025[52] but says nothing public about AI; vendor figures are customer claims.
  • Industry benchmarks for lead conversion, days from signing to opening, units per field consultant and support tickets per unit.
  • How many franchisors already write AI rules into their manuals or FDDs. A scan of FDDs in state filing databases would answer it.
  • How franchisees use their right to complain. The GAO counted about 5,900 franchise complaints to the FTC in 2018 to 2022, and most of its discussion groups did not know they could file[12].
  • Whether the current FTC relies on its 2024 fee guidance and policy statement, and what California's regulator says about undisclosed fees.
  • The full list of registration states, other states' termination rules, and whether California has funded broker registration.
  • Who carries privacy duties for loyalty and app data, and bias audits for a hiring tool, when the brand picks the tool and the franchisee uses it.
  • The SEC order settling with Presto, the McDonald's and IBM test details, and how city scheduling laws count franchisees' employees.
18

Sources

Sources were read in October 2026; dates are publication or data dates.

Primary sources

Federal and state regulators, legislatures, courts and SEC filings. Ratios marked as our arithmetic are calculated from the filed statements.

  1. 1
  2. 2
  3. 3
    16 CFR 436.2: obligation to furnish documents. Cornell Legal Information Institute.
  4. 4
    16 CFR 436.5: disclosure items. Cornell Legal Information Institute.
  5. 5
    16 CFR 436.7: updating requirements. Cornell Legal Information Institute.
  6. 6
    16 CFR 436.9: additional prohibitions. Cornell Legal Information Institute.
  7. 7
    Franchise Rule Compliance Guide. Federal Trade Commission.
  8. 8
    Staff guidance on undisclosed fees in franchising. Federal Trade Commission staff, 12 July 2024.
    Staff guidance, not binding on the Commission
  9. 9
    Policy statement on franchisors' use of contract provisions, including non-disparagement, goodwill and confidentiality clauses. Federal Trade Commission, 12 July 2024.
    Still listed in the FTC's legal library in October 2026
  10. 10
  11. 11
  12. 12
    Franchises: additional actions could strengthen FTC oversight (GAO-23-105338). US Government Accountability Office, 5 April 2023.
  13. 13
    FTC v. Wyndham Worldwide Corp., No. 14-3514. US Court of Appeals for the Third Circuit, 24 August 2015.
  14. 14
    Wyndham Worldwide Corporation (case page and settlement). Federal Trade Commission, settled 11 December 2015.
  15. 15
    16 CFR Part 465: use of consumer reviews and testimonials. eCFR, effective 21 October 2024.
    Reused from our trade contractor research, not reopened
  16. 16
    Rytr LLC, In the Matter of (case page). Federal Trade Commission, complaint 25 September 2024; order set aside 22 December 2025.
    Reused from our trade contractor research, not reopened
  17. 17
    FCC makes AI-generated voices in robocalls illegal. Federal Communications Commission, 8 February 2024.
    Reused from our staffing research, not reopened
  18. 18
    29 CFR 103.40: joint employers. Cornell Legal Information Institute.
  19. 19
  20. 20
  21. 21
    Business Trends and Outlook Survey, sector and national data, question 7. US Census Bureau, reference period 24 August to 6 September 2026.
  22. 22
  23. 23
    Minnesota Statutes 80C.02: registration requirement. Minnesota Office of the Revisor of Statutes.
  24. 24
  25. 25
    RCW 19.100.020: registration required. Washington State Legislature.
  26. 26
    RCW 19.100.180: franchisee bill of rights. Washington State Legislature.
  27. 27
    Franchise Act interpretive statement FIS-09: fees not disclosed in the FDD. Washington Department of Financial Institutions, 1 November 2023.
  28. 28
  29. 29
    SB 919 (Chapter 518, Statutes of 2024): franchise brokers. California Legislative Information, approved 24 September 2024.
  30. 30
    AB 325 (Chapter 338, Statutes of 2025): common pricing algorithms. California Legislative Information, chaptered 6 October 2025.
  31. 31
  32. 32
    SB 7 (2025): automated decision systems in employment, bill status. California Legislative Information, vetoed 13 October 2025.
  33. 33
    Civil Rights Council secures approval for regulations on employment discrimination related to artificial intelligence. California Civil Rights Department, 30 June 2025.
    Reused from our staffing research, not reopened
  34. 34
    CCPA updates, risk assessments and automated decisionmaking technology regulations. California Privacy Protection Agency, effective 1 January 2026.
    Reused from our staffing research, not reopened
  35. 35
    Automated employment decision tools (Local Law 144). NYC Department of Consumer and Worker Protection, enforced since 5 July 2023.
    Reused from our staffing research, not reopened
  36. 36
    HB 3773 (Public Act 103-0804), amending the Illinois Human Rights Act. Illinois General Assembly.
    Reused from our staffing research; ilga.gov could not be opened in October 2026
  37. 37
    SB26-189: automated decision-making technology. Colorado General Assembly, signed 14 May 2026.
    Reused from our staffing research, not reopened
  38. 38
    Mobley v. Workday, Inc., No. 3:23-cv-00770 (N.D. Cal.): docket. CourtListener (RECAP archive).
    Reused from our staffing research, not reopened
  39. 39
    Secure Scheduling Ordinance. Seattle Office of Labor Standards.
  40. 40
    ORS 653.436: advance notice of work schedule. Oregon Revised Statutes (via oregon.public.law).
  41. 41
    Franchising Code of Conduct. Australian Competition and Consumer Commission.
  42. 42
  43. 43
  44. 44
    Yum! Brands, annual report on Form 10-K for 2025. SEC EDGAR, filed 20 February 2026.
  45. 45
  46. 46
    Planet Fitness, annual report on Form 10-K for 2025. SEC EDGAR, filed 25 February 2026.
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51

Industry bodies and independent research

The International Franchise Association's outlook (prepared by FRANdata), franchisee survey and benchmark firms whose full data is paid, and independent security researchers.

  1. 52
    2026 Franchising Economic Outlook. International Franchise Association, prepared by FRANdata, 2026.
    Headline figures only; the full report is gated
  2. 53
    Franchise data and benchmarking. FRANdata.
    Benchmarks are sold, not public
  3. 54
    Franchisee satisfaction research. Franchise Business Review.
    Survey scale only; brand results are paid
  4. 55
    McHire: 64 million applicant records exposed. Ian Carroll and Sam Curry, independent security researchers, 30 June 2025.

Vendor sources

Published by companies that sell software to franchisors and multi-location brands. Directional, not an industry benchmark.

  1. 56
    Franchise management platform and Frannie agents. FranConnect, read October 2026.
    Vendor source
  2. 57
    Franchise development CRM and lead follow-up. ClientTether, read October 2026.
    Vendor source
  3. 58
    Operations execution and audits. Crunchtime, read October 2026.
    Vendor source
  4. 59
    Local marketing, reviews and Genius Agents. SOCi, read October 2026.
    Vendor source
  5. 60
    Listings, reviews and AI search visibility. Yext, read October 2026.
    Vendor source
  6. 61
    Conversational hiring for high-volume employers. Paradox (part of Workday), read October 2026.
    Vendor source

Company and press

News coverage and an encyclopedia summary of reporting we could not open directly.

  1. 62
    Taco Bell (section on voice AI at the drive-thru, citing The Wall Street Journal, August 2025). Wikipedia.
    Secondary source; the underlying article was not opened
  2. 63
    McDonald's to end IBM AI drive-thru test. CNBC, 17 June 2024.
    Could not be opened in October 2026; details to be confirmed

Become a design partner

Run Stage 0 and a first corporate agent with us, on your own systems. You get early access, help with setup, and a say in what we build next.

Or email support@prefero.ai