Your dashboard is not wrong. It just waits.

A med spa action center is the part of a business intelligence system for a medical spa that looks for what matters without being asked and then does something about it. Nabu's runs every night across every location, keeps only the findings big enough to be worth an owner's time, ranks them by the money they represent, builds the work each one calls for, holds that work until a person approves it, sends it to the team as real tasks, and measures the result four weeks later. A dashboard shows numbers and stops there. An action center closes the gap between the number and the result.

Start with a short, specific list rather than a message to everyone on file. Nabu builds that list for you: clients who are past the gap your own clinic rebooks that treatment at, who have not been in for anything else since, who have nothing on the books, and who have agreed to be contacted. It then drafts a message for each person that refers to their own last visit, creates one follow-up task per client for someone on your team, and four weeks later reports how many actually came back and what they spent. Repeat visits are roughly two thirds to three quarters of med spa business on the American Med Spa Association's published numbers, which is why lapsed clients are usually the cheapest revenue in the building.

Use your own clinic's history rather than an industry table, because the gap that matters is the one your clients actually keep. Nabu works out the typical gap for each service from your own repeat visits, and it will only state one when there is enough history to trust it: at least ten gaps across six different clients. Below that it falls back to a default for that treatment type and says on the finding that it did. If you would rather set the number yourself, enter it on the service in Setup and yours wins over both.

Usually because what you keep per treatment changed, not how many you sold. Product cost went up, a discount quietly became the standard price, or the treatment started taking longer than it is priced for. Nabu splits a drop in contribution into three parts: how much came from volume, how much from the price you actually realized after discounts, and how much from what it costs you to deliver. Those three have to add back up to the change or Nabu refuses to label it at all, and the review it prepares matches whichever one is driving it, so a cost problem gets a cost review rather than a generic look at the numbers.

Yes, and that is the point of it. Nabu sits on top of the booking system you already run and reads from it, so you keep Zenoti, Boulevard, Mangomint or Vagaro exactly as it is, including how your team books and charges. What Nabu adds is the layer those systems stop short of: what each treatment costs you to deliver, what is left after that, which clients have quietly stopped coming, and what to do about the parts that are slipping. Nothing about how your front desk works has to change.

An alert fires when a number crosses a line somebody set, and then it is your problem. A finding names something you recognize, says what it is worth and how that was worked out, shows the evidence and the date your data ends, and arrives with the work already built. Nabu ranks every actionable finding and leaves out small changes and anything that does not yet have a useful next step.

A dashboard hands you a chart and leaves the thinking, the work and the proof to you. Nabu does the work and then grades it. Three things follow from that. The med spa cost model is already built in, so you are not configuring what a device hour or an injector minute costs before you get anything useful out of it. Nabu will not propose a move it cannot also carry out and measure, so you never get advice it has no way of checking. And the price is published on this site rather than quoted on a call.

No. Roughly four in five med spas are single location businesses on the American Med Spa Association's 2024 numbers, which means the owner is usually the analyst, in between patients. Nabu is built for exactly that: findings are ranked, each one reads as a sentence, and the work under it is already prepared and assigned. The alternative most owners are weighing is a fractional CFO at several thousand dollars a month, or another Sunday with the spreadsheet.

Nabu Intelligence covers the Action Center and all analytical Ask Nabu chat, and costs $99 a month per location on an annual plan or $125 month to month. It is an add-on, so it attaches to any Nabu plan, including the free Solo Founder plan for a single provider at one location. Every location you turn it on for is priced the same as the first, and this is a launch price held for anyone who starts on it.

Detection runs every night, so the first findings appear the morning after your data has synced. What they can cover depends on how much history came across with it, because Nabu learns each service's rebooking gap from your own repeat visits rather than from a table, and it needs enough of them before it will state one. Every finding is stamped with the date your data actually ends, not the date the job ran, so you always know how current it is. A window your data does not cover produces no finding at all.

No. In this version Nabu sends nothing at all. An approved outreach becomes a list, a message written for each person, and assigned tasks for someone on your team to work and send. Nobody is contacted on a channel they have not agreed to, a do not contact flag beats everything else, and consent is checked again the moment before the work is created rather than only when the plan was written. Direct sending comes later, and it will sit behind the same approval when it does.

Whoever already has the permission the equivalent button in Nabu needs, at every location the work touches. There is no separate role to set up. Somebody who cannot see a clinic's financials cannot approve work against it, and the staff you can assign work to are the active employees at that clinic rather than everyone in the business. Every approval records who approved it, when, what was proposed, what ran, and what it created.

Into the Task Center, which is the surface your team actually opens. Each task carries the client, the reason, the approved message, an owner and a due date, and the person working it records what happened: contacted, booked, no answer, declined, or do not contact. Each person sees only their own list in Nabu Today. The Action Center reads all of that back, so an outreach whose tasks were never worked is graded as outreach that never went out rather than outreach that failed.

Often it cannot, and it is built to admit that. On any list of 20 or more clients, Nabu holds one in five back at random and never contacts them. Four weeks later it compares the two groups on how many came back and what they spent. If the people nobody contacted came back at about the same rate, the verdict is that Nabu cannot attribute the result, not that it worked. Without that rule, anything you did in a good month would look like a win.

Nabu says so. Four weeks after the work actually starts it measures the exact people the action targeted, over the 28 days after against the 28 days before, and gives one of four verdicts: improved, worsened, unchanged, or cannot attribute. The finding closes either way, because leaving it open would put the same list in front of you again next week. Outcomes are written once and never edited, so the record of what happened survives Nabu later concluding something different.

Yes, it uses AI, and no, the AI cannot touch your numbers. Nabu is built as two engines with separate jobs. Ordinary code finds the problems, calculates what each one is worth, and builds the moves Nabu can actually carry out, so no figure is ever estimated by a model. The AI then reads that evidence and the prepared plans, recommends the one that fits your situation, explains why in plain English, and writes each client's message the way a person at your clinic would. It cannot change a figure, a name or a recipient, or offer a move Nabu has no way to carry out, and a sentence containing a number the evidence does not hold is rejected before you see it.