Call five of your own stores on a Tuesday afternoon and write down what you hear. One answers with the brand name and the location. One answers "yeah, hold on." One picks up on the eleventh ring with fryer noise behind it. One rolls to voicemail. One does it exactly the way the operations manual says.
That spread exists in nearly every franchise system, and it exists for a structural reason rather than a discipline one. Every other brand-standard surface is visible. A dirty dining room shows up on a field visit. A wrong uniform shows up in a photo. Bad phone answering happens hundreds of times a week at every location and nobody at the franchisor ever hears a single one of those calls.
What a phone standard is actually made of
Most brand manuals have a phone section, and most of them are one page. Written out fully, the standard has more parts than operators expect.
There is the greeting itself, which is the wording and the order of information. There is answer time, which is really a staffing rule dressed up as a phone rule. There is the hold policy, including whether hold is allowed at all during peak. There is how the person on the phone describes menu items, which matters for allergens and for upsell language and for not promising something the kitchen stopped carrying in March. There is complaint handling, which is the part where a bad thirty seconds turns into a public review. And there is the closing, including whether the caller is told a pickup time and whether that time is real.
A franchisee can comply with the greeting and fail every other piece. Most audits only check the greeting, because the greeting is the only part a mystery shopper can score reliably.
Why the standard drifts, and why it isn't laziness
A phone standard is the only brand standard that competes directly with guest service in the moment.
When a shift lead is running a counter with six people in line and the phone rings for the fourth time, the manual's greeting loses to the people standing in front of them. That is a rational trade under pressure. Repeated for a year, it becomes the store's actual phone culture, and no amount of retraining survives the next Friday rush.
Turnover finishes the job. The person you trained in February left in May. The person who replaced them learned the phone by listening to whoever was nearest, which means the store's drift compounds instead of resetting. A brand with two hundred locations is running two hundred independent copies of that decay process, at different speeds, invisibly.
What a voice agent locks, and what it doesn't
The honest framing is that a voice agent converts a behavioral standard into a configuration setting, and configuration does not drift.
Locked by construction: the greeting wording, the tone, the answer time, the way menu items and allergens get described, the upsell prompt and whether it fires, the complaint routing rule, and whether the caller is quoted a real pickup time or a guessed one. These stop being training topics because there is no human in the loop to deviate from them. The custom greetings and brand voice piece covers how far the wording control goes in practice.
Not locked: everything that happens after the caller arrives. The agent cannot make the store bag the order correctly, hit the quoted time, or hand it over pleasantly. It also cannot fix a store whose real problem is that they are two people short on Fridays. If your phone standard is failing because staffing is failing, the phone gets better and the rest of the guest experience does not.
Say that plainly to franchisees during rollout. Overselling it as a fix for store-level operations is how a brand program loses credibility in month three.
Splitting the standard between brand and operator
The design question that decides whether this works is which settings the franchisor controls and which the store controls. Get it wrong in either direction and you either get revolt or you get drift back.
The workable split, in most systems, gives the store everything that is local and changes often:
- Hours, including the holiday exceptions the franchisor never hears about until after the holiday
- Delivery zone, delivery minimum, and whether delivery is running at all tonight
- Location-specific facts: parking, the entrance that is actually easier to find, whether the patio is open
- Local promotions and LTOs the store is running with permission
- The escalation phone number and who is on it during which dayparts
Everything else sits with the brand. Greeting wording, brand-voice tone, complaint routing rules, allergen language, the national promotion script, and the reporting definitions all belong centrally, because they are the things that only mean something if they are identical everywhere. The same boundary problem shows up in franchisee versus franchisor rollout decisions, and it is worth settling before the first store goes live rather than during it.
Write the split down as a document before you write it into a configuration. Franchisees who see the boundary in advance argue about it once, which is much better than arguing about it per store.
The compliance conversation with franchisees
A mandated system provokes a predictable objection, and it is rarely about the brand standard. It is about cost and about control, in that order.
The cost answer is arithmetic they can do themselves. Plans start at $250 per month. If their average ticket is $30 and they are missing even a handful of calls a day during peak, the recovered orders cover the subscription well before the end of the month. Have them run their own numbers rather than yours, using something like the phone ordering ROI calculator, because a number they computed is a number they believe.
The control answer is the split above. A franchisee who keeps hours, delivery rules and local promotions does not feel like a store is being run from headquarters. A franchisee who has to file a ticket to change Christmas Eve hours will work around the system within a quarter.
There is a third objection you should take seriously: some operators genuinely believe their regulars want a person. In a full-service room with a real bar crowd, that can be true. The compromise is a routing rule rather than an exemption, so those calls reach a human by design while the brand still owns what the caller hears first.
Auditing what you can finally see
Once the greeting is identical everywhere, mystery shopping the phone stops being a useful spend. Four sampled calls a year told you almost nothing anyway.
The audit that replaces it reads transcripts. Pull escalated and failed calls by location, monthly, and sort them: callers asking for a person, complaints, questions the agent could not answer, and menu items the agent could not parse. The last two are configuration defects with named fixes. The first two are volume you can compare across stores, and a location whose complaint rate runs well above the system is telling you something about its operations that no field visit was going to surface. Multi-location reporting rollups covers how to structure that view so it stays readable at fifty stores.
Then do the thing you could not do before. Compare answered-call rate by store and by daypart. It is the first phone metric in franchising that is measured the same way at every location, which makes it the first one you can hold anyone to.
Pick your ten worst stores on that metric next month and look at their staffing, not their scripts. The phone will already be saying the right words.