My area manager says everything is fine but orders are down
Restaurant chains hit this disagreement constantly and treat it as a credibility problem, which it usually is not. The area manager is reporting what they can see: the kitchen is staffed, the tablet is on the counter, the team is working. The numbers are reporting what customers did. Between those two sits the storefront, which neither of them observes, and which can be closed, restricted, mispriced or missing a promotion while everything visible from inside looks exactly as it should.
Why can both reports be honest?
Because they are answering different questions and neither one is the one you asked.
“Is the restaurant operating?” is what a manager can answer. “Could a customer order?” is what the numbers depend on. Those diverge whenever the listing stops matching the kitchen, and a listing can stop matching the kitchen in half a dozen ways that produce no signal inside the building.
The disagreement is therefore evidence of a measurement gap rather than of anybody being wrong.
What could be true that neither of them would see?
Five candidates, commonest first.
The listing was not orderable during part of the trading day, from a device or a state nobody cleared. The delivery zone narrowed, so a slice of the catchment cannot see the site while nearby customers can. A promotion stopped appearing, so the marketing that was driving volume is not running. Prices on the listing differ from what was intended. Or the site’s ranking in the app fell, so fewer customers ever reach it.
Each of those produces a fall in orders and none produces anything visible from the counter.
How do I settle it quickly?
Open the listing as a customer, from an address a few miles away, during trading hours. Not the portal, the customer app.
That one action resolves a surprising share of these disputes on the spot. If the site is not there, or shows as closed, or has no offers where you expected them, the argument is over and the work begins.
If it looks correct at that moment, you have learned something narrower than it feels: it was fine at that minute. Availability problems are usually intermittent and concentrated in the evening, which is exactly when nobody is checking.
What if it keeps looking fine when I check?
Then the check interval is the problem rather than the site.
An outage from 19:10 to 21:40 is invisible to anyone who looks at ten in the morning. Spot checks find persistent problems and miss episodic ones, and episodic ones are where most lost trading lives.
Continuous observation from outside is the only thing that closes it. With a Kitchain (kitchain.co) record behind the conversation, the manager’s account and the sales figures stop being two versions to choose between and become one timeline that explains both.
How should the conversation be run once you have the data?
Take the accusation out of it, because the manager was not wrong.
The productive framing is that the site was operating and the storefront was not, and that nobody at the site could have known. That keeps the manager as an ally in fixing the cause rather than as a defendant. And the cause is nearly always something they can fix in seconds once told.
The unproductive framing is asking them to be more vigilant, which asks a person to watch something they cannot see.
What if the data shows the listing really was fine?
Then you have ruled out the largest single explanation and can look at demand honestly.
That is worth something on its own. Most soft months are investigated without ever eliminating the storefront, so the analysis proceeds on an assumption nobody tested. Ruling it out in an afternoon makes the rest of the investigation credible.