Every AI platform in the restaurant category reports metrics. Most of the metrics measure the platform, not the operation. This is a partial list. Feel free to send it to your account representative.
Vanity Metric 1 — Alert Volume
How many alerts did the platform fire this week? The platforms celebrate this number. It proves the system is watching. What it does not prove is that the watching produced anything. An alert that fires to an operator at capacity is an alert that died. Alert volume measures platform activity. It does not measure operational change.
Vanity Metric 2 — Insights Surfaced
The dashboard surfaced 847 insights this month. That number lives on a slide. The question the number does not answer is whether any of those insights caused an action to occur. Insight count is a proxy for "the AI is producing something." It is not a proxy for "the operation is different than it would have been."
Vanity Metric 3 — Forecast Accuracy
Our demand forecast was 94% accurate. Congratulations. The 94% accuracy improved the manager's schedule draft. The manager still drafted the schedule. The 94% was the floor her judgment built from. Forecast accuracy in isolation is a technical metric for the platform. It is not a metric for what the operation did differently because the forecast existed.
Vanity Metric 4 — User Engagement
Your team logged in 112 times this week. The platform measures the login. The platform does not measure whether the login led to an operational change that would not have happened without the platform. Engagement is how SaaS measures itself. It is not how a restaurant measures whether the platform earned its keep.
Vanity Metric 5 — Integration Count
We connect to 47 systems. That is an infrastructure fact. It is not a value fact. A platform can be connected to every system you run and deliver zero autonomous action. The integrations count measures plumbing. Plumbing does not operate the restaurant.
Vanity Metric 6 — Customer Satisfaction with the Platform
NPS 72 from our user base. The operators like the platform. The question is what the operators did differently because of the platform. A platform that is pleasant to use and changes nothing about the operation is a pleasant dashboard. The NPS measures the dashboard experience. It does not measure the restaurant.
The One Metric That Isn't Vanity
Autonomous actions taken during service, without human-in-the-loop approval, in production deployments, per service hour.
That is the metric. Count the actions the system took without anyone clicking confirm. Not the alerts sent. Not the insights surfaced. Not the recommendations made. The actions the platform executed on its own recognizance, within parameters the operator set, in live service, when the manager was carrying something else.
If the answer is zero, the platform is an awareness tool. That is a legitimate category. It is not an execution layer. If the answer is non-zero, every number that follows — hospitality loss recovered, yield captured, compliance maintained, labor rebalanced — is a consequence of that first number. Autonomous actions per service hour is the metric the industry does not report because most of the industry cannot report a non-zero number.
Ask the Question
The session at RLC will make the case for outcomes over vanity numbers. That is the right instinct. The outcome that matters is whether the platform acts. Every other metric is a proxy for that one question. The platforms that are built to act will be happy to quote the number. The platforms that stop at the arrow will reframe the question.
Watch how they reframe it. That is where the category lines are drawn.