If your agreement carries a fifteen percent phone-order rate, a $42 order placed by a caller who found your number on a marketplace listing nets you about $35.70. If it carries thirty percent, the same order nets $29.40. Neither figure appears on the ticket. It surfaces at the end of the month as a line on a statement, grouped with everything else, at which point nobody on your staff can reconstruct which calls it referred to.
That is the part operators find frustrating, and it is worth understanding mechanically before deciding whether to do anything about it.
The forwarding number is the entire mechanism
A marketplace listing displays a phone number. On many platforms, that number is not your restaurant's line. It is a tracking number owned by the platform that forwards to your line. The call rings at your host stand, sounds completely normal, and your staff takes the order the way they always do.
The platform, meanwhile, knows the call happened, how long it lasted, and that it came from their listing. That record is what a phone-order commission is billed against.
Nothing about this is hidden in a legal sense. It is generally described in marketplace marketing terms. It is hidden in a practical sense, because the mechanism is invisible to the person answering the phone and the charge is invisible on the ticket.
Read the agreement, then read the statement
Two documents, in that order.
The agreement tells you which rate applies to phone orders, whether it differs from the delivery rate, and what counts as an attributable call. Definitions vary. Some agreements bill on calls of a minimum duration whether or not an order was placed. Others bill only on confirmed orders. That distinction matters enormously if your phone gets a lot of "are you open on Monday" calls.
The statement tells you whether the count matches reality. Pull one month. Count the phone-order charges. Then pull your own call log for the same days and count the calls that arrived on the marketplace number. If your phone provider does not break out the dialed number, that is the first thing to fix, because you cannot audit a bill you have no independent record of.
Do this once. If the numbers line up and the volume is small, stop and go do something more valuable. Auditing is only worth repeating when it has already found something.
What routing around it actually involves
Assume you decide the volume justifies the work. The steps are unglamorous.
- Update the phone number on every listing you control, starting with Google Business Profile, Yelp, Apple Maps, and your own website footer, so that the number a customer finds most easily is yours.
- Ask the marketplace, in writing, whether your listing number can be set to your direct line, and keep their answer in a file with the agreement.
- Print your direct number on menus, takeout bags, receipts, and delivery inserts, since a repeat customer who calls the number on the bag never touches the marketplace again.
- Make sure the direct line is actually answered, because moving calls to a line that rings out at seven on a Friday costs more than the commission did.
That last one is where most of these projects fall over. Operators shift traffic to their own number, the number goes unanswered during exactly the hours the volume arrives, and the orders go somewhere else entirely. The economics of voice AI versus delivery app commissions only work if the direct channel picks up.
The answered-call problem is the real constraint
A commissioned order at 85 cents on the dollar beats an unanswered call at zero. That comparison is unflattering but honest, and it is the reason marketplaces retain phone volume even from restaurants that resent the fee.
So the sequence matters. Fix your ability to answer first, then move the traffic. A restaurant that reliably answers during its dinner rush can take the marketplace number off its listings and keep most of the revenue. A restaurant that misses a third of its peak calls will simply convert commissioned orders into lost ones, and will not find out, because a missed call leaves no record.
If you are already consolidating orders from several platforms, the same logic applies across all of them, and consolidating third-party delivery phone orders covers the operational side of running one intake process instead of four.
What you give up
Being fair about this: the marketplace listing is doing something for you. It is discovery. Someone who did not know your restaurant existed found it in an app, and the phone number they called was the one the app gave them.
If a meaningful share of your marketplace phone orders are first-time customers, routing around the fee costs you those customers, not just the commission. The number to look at is repeat rate. If callers from that number are largely people ordering from you for the fifth time, the platform is charging you for a customer you already own, and moving them to your own line is straightforward value. If they are mostly new, you are paying an acquisition cost, and acquisition costs are allowed to be expensive.
Most independents land somewhere in between, which is why the honest recommendation is usually to keep the listing and move the repeat customers, rather than to cut the channel off.
Porting, and when not to bother
Some operators respond to all of this by wanting a new number entirely, or by wanting to bring an existing number under a system they control. That is a reasonable end state, and phone number porting for restaurants walks through how it works and what breaks during the transition.
It is also, frequently, more disruption than the problem justifies. Changing your primary number means updating every listing, every printed menu, every delivery insert, and every regular's contact card. If your goal is to stop paying commission on calls, you usually do not need a new number. You need your existing number on the listings customers actually use, and a line that answers.
Here is the test worth running this week. Pull last month's statement, count the phone-order charges, multiply by your average ticket, and multiply that by the rate in your agreement. If the result is under a couple hundred dollars, file it and move on. If it is over a thousand, spend the afternoon on your listings, and check your peak-hour answer rate before you touch anything else. Compare the annual figure against what a phone system that answers every call costs and the decision usually makes itself.