Invariants and local choices: the only two lists a franchise standard needs

01

The third list is the one nobody admits to writing

Open any operating manual and read the verbs. A small number of clauses say must. A smaller number say the site may choose. The rest say should, or recommended, or wherever practical, or expected, or best practice.

That third group is not a standard and it is not a freedom. It is an unresolved argument, printed. Somebody wanted the rule, somebody else did not want to defend it, the wording split the difference, and the manual went to print. In the manuals I have worked with, the middle group was the largest of the three by a distance.

What follows is predictable. Every site reads the middle privately, according to temperament and pressure. A conscientious manager treats should as must and slows down. A stretched manager treats should as optional and moves on. Both believe they are compliant, and neither is wrong, because the document never decided.

Then a complaint arrives and head office discovers it had an opinion after all. It enforces a should. The site that ignored that clause for two years is now in trouble for behaviour the manual permitted. That single event costs more than almost anything else a franchisor does, because afterwards the network reads the document as a mood rather than a rule.

02

What functioning networks already do

The two list model is not my invention. It is a description of what working networks do, and the research says so plainly.

The Service Industries Journal published a study of forty franchisors in the United Kingdom on exactly this tension, local conditions pushing against a uniform format. Adaptation happened at the edges. Peripheral parts of the format were allowed to move to fit local circumstances, and the core parts were held unchanged everywhere. Those franchisors were not choosing between standardisation and adaptation. They had already split the format in two and were applying a different rule to each half.

The same study found something less comfortable. Most of them described franchisees as a source of innovation. Very few actually took a franchisee's idea and rolled it across the system.

That is the failure mode of the local list. You permit variation, the variation produces something better, and the discovery dies in the site that made it. If you are going to free the method, build a route home for what the method finds. One quarterly page where each site records what it changed and what happened is enough to start. Reading it is the hard part.

03

How do you decide what is an invariant?

Four questions, and a rule has to pass all four. On Monday, take the ten clauses your network argues about most and run them through it.

Does a difference change the promise or only the route? If customers at two sites would describe what they were promised differently, that is an invariant. If they would describe the same promise arriving by different means, it belongs to the site.

Would you enforce it against your strongest site, in its worst week? This is the honest question. Every network has a site that carries the numbers and knows it. If you would quietly let that site off, the rule is not an invariant anywhere.

Does following it leave evidence somebody can check without a visit? An invariant that produces no artefact cannot be held, only asserted. If the only proof is a person saying yes, you have written an aspiration.

Can every site do it today, with the staff and equipment it actually has? If two sites physically cannot comply, you have three options and only three. Fund them so they can. Move the rule to the local list. Or accept that you are running a network with known non-compliance and say that out loud. What you must not do is leave the rule written and unmet, because that teaches every site that written rules are negotiable.

Anything failing a question moves to the local list or gets rewritten until it passes. Nothing stays in the middle.

A rule you would not enforce against your strongest site is not a standard. It is an opinion with a document number.

04

Local choices should be published, not merely tolerated

The second list usually gets treated as a residue, whatever the manual did not mention. That is a mistake, because silence is read differently by different people. A cautious franchisee reads an unmentioned thing as forbidden and asks permission for a decision that was always theirs. A confident one reads it as licence and changes something you would have refused.

So write the local list on purpose. Name the decisions each site owns outright. Name the ones it owns provided it tells you. Name the ones that need approval first. Three columns, and every operational decision your network makes should be findable in one of them.

That list is also what makes the invariants credible. A short list of things that never vary reads as serious precisely because it sits beside a long list of things the site is trusted to decide. A network that fixes everything is not signalling high standards. It is signalling that it cannot tell which of its rules matter.

05

The brand does not carry the quality by itself

Health Policy and Planning published a systematic review asking whether franchising in health care is valuable. Fifteen empirical studies met the bar for inclusion. Franchising came out mostly positively associated with client volumes, physical accessibility and some types of quality, and the authors were direct about how weak the underlying evidence was. Hold on to that. The model is associated with reach more clearly than with quality.

Which puts the weight back on the design. If the format does not deliver quality on its own, the two lists are what deliver it, and how you run them matters as much as where the line sits.

A mixed methods study in PLoS One covering Dutch health care franchises looked at the relationship between franchisor and units. What went with better outcomes was relational rather than contractual: partnership instead of hierarchical control, mutual involvement in developing new practice, open communication, and trust that respected professional autonomy. Those associations were stronger for unit satisfaction, survival and quality of care than for financial performance.

Read it as an operating instruction. The line between the two lists is a negotiation you hold every quarter, not a wall you build once.

06

What is yours to decide, and what is not

You do not control how a franchisee reads a document you are not in the room for. You do not control the local labour market, the regulator who changed a rule in one country and not the others, the landlord, or whether a good site manager stays. You do not control the pressure that makes a busy team skip a step.

You control which promises are invariant and how few of them there are. You control whether each one leaves evidence without anybody being asked. You control whether the local list is written down or left to inference. You control what you do the first time your strongest site breaks an invariant, because the network learns more from that decision than from the manual. And you control whether the third list exists at all.