Nightclub KPI Dashboard: What Belongs On It and What Is Just Noise

Every nightlife dashboard ever built opens with gross revenue and a bar chart of the last twelve months. Both are true. Neither has ever caused anyone to do anything differently on a Tuesday.

A dashboard earns its screen space when each number on it is attached to a decision someone can make this week. That is a high bar and most dashboards do not clear it. What follows is a list of the numbers that do, why each one changes a decision, and — just as important — the four that look serious and are not.

The test every widget has to pass

Before adding anything to a dashboard, ask the number one question:

If this went up ten percent, what would I do differently on Monday?

If the honest answer is "feel better" or "tell the staff," the number belongs in a monthly report, not on an operating dashboard. Gross revenue fails this test. So does headcount. They are outcomes; you cannot pull them.

The numbers that pass are the ones sitting one layer upstream of an outcome — close enough to a lever that moving them is an action, not a wish.

Tier one: the four that go at the top

Revenue per head, against a trailing band

Revenue per head = total revenue ÷ door count
Band             = trailing 4-week same-weekday mean ± 1 standard deviation

The decision it drives: whether last night was actually unusual. Almost every management reaction to "a bad Saturday" is a reaction to weather, a competing event, or ordinary variance. Plotting the night against its own trailing band converts a feeling into a yes or no.

The reason it goes first: it is the only number that separates "the room was quiet" from "the room was full and nobody spent," and those are opposite problems with opposite fixes.

New versus returning share

Returning share = guests seen before ÷ guests tonight

The decision it drives: where the marketing money goes next month. A room running at a very low returning share is buying its traffic every single night and will keep having to. A room where returning share is climbing has built something and can afford to spend less on acquisition and more on the reason people came back.

The prerequisite: a guest record that survives across visits. Without it this number does not exist, and most rooms do not have it. That is the single biggest gap between a nightlife dashboard and a retail one.

Channel yield

Channel yield = revenue from guests first seen via channel ÷ spend on that channel

The decision it drives: which line of the marketing budget gets cut on the first of the month. This is the number that pays for the entire dashboard, and it is the one most often absent, because it requires joining acquisition to spend and most systems hold only one of the two.

How to read it: as a ranking, not as an exact figure. Attribution in a cash-heavy room is incomplete for every channel by roughly the same mechanism, so the ordering is trustworthy even where the absolute values understate.

Staff retention, shift-weighted

Shift-weighted retention = staff working ≥ N shifts this period who also worked last period
                           ÷ staff who worked last period

The decision it drives: hiring, weeks before you would otherwise notice you need to. Headcount retention counts someone who worked twice the same as someone who worked twenty times; shift-weighting tells you whether you are keeping the people who actually carry the floor.

Why it goes in tier one: it leads revenue. By the time the revenue moves, the correction takes months.

Tier two: the operational four

These are checked daily and acted on within the week.

Booking-to-arrival rate

Arrival rate = parties that arrived ÷ parties that booked
No-show cost = (bookings − arrivals) × average party spend

The decision: whether to overbook, and by how much. Most rooms either do not overbook at all — leaving tables empty on the busiest nights — or overbook on instinct and occasionally have to turn away a party that showed. The arrival rate, tracked per weekday, gives you the actual number to work with.

Time-to-seat

Time-to-seat = median minutes from arrival to seated

The decision: door staffing on specific hours. A median that spikes between 11 and midnight on Fridays is a staffing schedule problem with a precise answer, not a general "the door is slow" complaint.

Repeat interval

Repeat interval = median days between a returning guest's visits

The decision: when to contact someone. A guest whose normal interval is three weeks and who is now at six is a different message from one at day four. This single number turns a mailing list into something with timing.

A caution about the window. Whatever "active" means in your system — 30 days, 90 days, a year — that threshold is a business decision, not a technical constant, and it should match your actual repeat interval. A room whose guests return every four months and whose system calls anyone past ninety days "lapsed" will spend all year chasing people who were never gone.

Comp and discount rate

Discount rate = value of comps and discounts ÷ gross before discounts

The decision: which manager gets a conversation. Comps are a legitimate tool and an easy habit. The rate itself is less interesting than its distribution across who authorized it.

The four widgets worth deleting

Gross revenue, alone. It is the outcome of everything else on the board. Keep it in the monthly report; it does not drive a Monday decision.

Year-over-year, as the headline comparison. Too many things changed. It is a board number, not an operating one, and treating it as operating produces reactions to conditions nobody controlled.

A leaderboard with no denominator. Ranking staff by gross rewards whoever was assigned the good section or the good night. If you rank, rank on a rate — per shift, per table assigned, per guest handled.

Live "revenue so far tonight." It feels essential and changes nothing. Nobody makes a decision at 11:40pm based on a running total that they would not have made from the room itself. It is a slot machine, not an instrument.

Layout: what goes where

The physical arrangement matters more than people expect, because a dashboard is read in about four seconds by someone doing something else.

Top row: the four tier-one numbers, each with its trailing band, each showing direction. Direction beats precision here — an arrow and a band do more work than three decimal places.

Second row: the operational four, as the week, not as tonight.

Below the fold: everything anyone asked for once. It is not wrong to keep it; it is wrong to put it on top.

One rule for colour: red and amber mean "outside the band," not "below target." A number below target but inside its normal variance is not an exception, and colouring it as one trains everybody to ignore the colour.

The dashboard we built against this specification

Everything above is a spec. Here is the product built to it.

tasteck computes the tier-one board out of your own bookings, with nobody assembling anything: guest records that persist across visits, new versus returning share, repeat interval, channel yield through to lifetime value, staff retention against the schedule you already keep, and booking-through-to-arrival for rooms on our reservation screen.

The widget nobody else has

Tier one item three said channel yield is the number that pays for the whole dashboard, and that it is usually missing. We went one step past it.

tasteck outputs the maximum you can spend on each channel next month, as an amount in your currency.

Not a performance chart. Not an index. A figure you take into the renewal conversation with a listing, an agency, or an ad account. It comes from the lifetime value of the guests that channel actually delivered, with your target margin applied.

No other product in the nightlife category outputs it. If you build your own dashboard from the spec above, this is the widget you will not be able to build, because it needs the guest record, the attribution join, and the lifetime revenue in one place.

Ask the board a question instead of reading it

The whole point of the four-second rule earlier is that dashboards are read badly. So tasteck also answers in plain language: it connects to ChatGPT over MCP, and you ask your numbers as a question. "How did last month break down by channel." "Is Saturday outside its band." The answer comes back in the chat.

Measured against every vendor listed on Japan's principal nightlife-industry directory, this is the first implementation of it in the category. And it is not ChatGPT-specific — the same interface is callable from anywhere, so it keeps working as the assistant landscape moves.

Built by people who ran the rooms

Every other product here is a software company's idea of what a venue needs. tasteck was built by operators who ran venues in this industry for sixteen years and grew the business from ¥200 million to ¥1.2 billion a year, six-fold, by attacking it with systems rather than by selling harder.

That is why the board leads with channel yield and retention rather than gross — those are the numbers that actually moved that business. The dashboard specification in this guide is not theory; it is the board that was used.

Multi-language is built in, the operating surface rather than a translated brochure, with your language set put in place during onboarding.

What it does not compute

Time-to-seat, comp and discount rate, and per-section table yield are not in the product. If those are your top three, better you know now.

One note on the active window. Our customer analytics have treated ninety days as the boundary for an active guest, and we are in the middle of making that configurable per venue — precisely for the reason written under "repeat interval" above. The right window is the one that matches your room, not the one we picked.

Building it yourself, in order

If you are constructing this from a POS export and a spreadsheet:

1. Get a guest identifier. A phone number captured at booking is the cheapest one. Without it, tier one items two and three do not exist and no amount of dashboard work will conjure them.
2. Build the trailing band before any other chart. It is what makes every subsequent number readable.
3. Join one channel to spend. Just one, for one month. The technique generalizes; the first one is the hard one.
4. Put shift-weighted retention on the board. It is a query against the schedule you already keep.
5. Delete the four widgets above. A shorter dashboard is read; a longer one is not.

Build tier one item three in the next five minutes

The channel-yield and budget-ceiling calculation is open on our site, nothing to sign up for: Ad budget calculator. Put in your spend per channel and what those guests spent. It returns the ceiling per channel as an amount. Nothing is transmitted anywhere — use it and close the tab.

That is one widget, computed by hand, once. The full system computes it every month out of your own bookings, along with the rest of the board above.

From $34 a month for up to two venues. Thirty days free on every plan, cancel any time. → Pricing

---

On benchmarks. No target ranges appear in this guide for revenue per head, returning share, arrival rate, or discount rate. We do not have a dataset broad enough to publish them, and the figures circulating publicly vary so widely by market and format that quoting a range would be inventing precision. Build your bands from your own trailing weeks; that is the only benchmark that describes your room.