Most operators approach this by looking for the list. Which states are two-party, which are one-party, print it out, done. The list exists, it is easy to find, and it is close to useless on its own.
It is useless because a restaurant call rarely stays inside one state. A caller on a mobile phone with an out-of-state area code, standing in your parking lot, ordering from a restaurant in a third state where your recording server happens to live, is an ordinary Tuesday. The list tells you the rule in each place. It does not tell you which place governs, and that second question is the one that actually decides what you are allowed to do.
This is general education and not legal advice. The point of the article is to hand you better questions for the attorney who can answer them.
Two-party is a misleading name for the rule
Start with the vocabulary, because the common shorthand causes real errors.
"Two-party consent" describes a rule that everyone on the call must consent to the recording. On a two-person call that means two people, which is where the name came from. On a call where your manager joins to resolve a complaint, it means three. On a conference call with a catering client and their event planner, more.
Practitioners generally say "all-party consent" for this reason, and it is worth adopting the phrase internally because it prevents a specific mistake: a staff member who believes they have "the second consent" once the caller agrees, and then loops in a third person without repeating anything.
The contrasting framework, where the consent of a single participant to the conversation is treated as sufficient, is covered in one-party consent and the disclosure you should still give. The two frameworks are not as far apart in practice as they sound, because the operational answer in both cases usually ends up being the same announcement.
There is a second piece of vocabulary that matters more than most operators expect. Several of these frameworks do not protect all communications, only ones the participants had a reasonable expectation would remain private. What counts as private, and whether a business call to a public-facing restaurant line qualifies, is exactly the kind of question that has been litigated and is exactly the kind of question you should not resolve yourself.
Why the list is the wrong tool
Imagine a call to your restaurant. To know which rule governs it, you would need to know at minimum where the caller physically is, where your business is, where the recording is created and stored, and how the relevant jurisdictions have treated calls that touch more than one of those places.
You know approximately none of that at the moment the phone rings. You cannot know where a mobile caller is standing. Area codes have not indicated location for two decades. And you certainly cannot make a routing decision about disclosure based on facts you will not learn until after the conversation ends.
Different jurisdictions have taken different positions on the interstate question. Some have applied their own law to calls involving their residents. That variation is the reason the list of states, which looks like a clean answer, produces an unclean one.
The practical response most multi-state operators land on is to stop trying to sort calls at all. Apply the strictest standard that could plausibly reach any call you handle, on every line, all the time. It costs you one sentence at the top of a call. It removes an entire category of decision-making from people who are simultaneously trying to seat a four-top.
The two states restaurants ask about most have their own articles, because the details differ enough to be worth separate treatment: California call recording and Florida call recording.
What "consent" looks like on a real restaurant phone
The mechanism nearly every business uses is a disclosure at the very start of the call, before any substantive conversation, followed by the caller continuing to talk. The theory is that a caller who hears the notice and stays on the line has agreed. Whether that theory holds in your jurisdiction is a legal question, and it is the single most important one to bring to counsel.
What that means operationally, if you decide to record:
- The announcement plays before anything else, including your greeting, since a disclosure that arrives after the caller has already stated their order is protecting nothing.
- It is on every line that gets recorded, including the catering line, the second line nobody remembers exists, and any overflow path.
- It is short and plain. A long notice invites people to hang up, and a vague one may not say what it needs to say.
- Anyone who joins the call later hears it or is told, which is the part manual processes fail at most often.
- Callers who object have a path that does not require them to hang up, meaning someone can continue without recording.
That last item is the one to think hardest about, because it determines whether your system is genuinely obtaining consent or merely announcing a policy.
If an automated agent answers your phone, the announcement is easier to get right, because software does not forget it during a rush and does not shorten it on the four hundredth call of the day. That consistency is a real advantage and it is not the same thing as compliance. The general framing is in call recording and consent for restaurants.
The part that outlasts the call
Consent governs whether you may capture the conversation. It says much less about what happens to the recording afterward, and the afterward is where restaurants accumulate risk without noticing.
Ask three questions of yourself and of any vendor holding this data. What exactly is retained, audio or transcript or both. How long is it kept, by default, without anyone deciding. Who can access it, including which vendor employees. A recording you kept for four years because nobody set a deletion rule is a liability you chose by accident, and a transcript is not meaningfully safer than audio for this purpose. Setting an actual rule is the subject of call recording retention policy, and the ownership questions sit in voice AI data ownership and privacy.
Here is a reasonable place to land. Decide first whether you need recordings at all, because many of the reasons operators want them, such as settling order disputes, are served by an order record rather than audio. If you do need them, apply an all-party-style disclosure on every line regardless of which states you think you are in, write down how long you keep things and actually delete on that schedule, and take that written practice to an attorney in your state for review. That review is an hour of someone's time, and it is the only step in this article that produces an answer rather than a better question.