I have built the same dashboard in four different companies now. Each time the first version tried to show too much. Each time the general managers stopped opening it inside a month. Each time I stripped it down and it started getting used.
Here is the version I would build today, on the first day, in any multi-unit restaurant group with more than three locations.
The dashboard is not a data problem
Most operators think dashboard adoption is about the tool. Power BI versus Tableau. Looker versus a custom build. Which POS extract, which labor system, which pipe.
The tool matters last. What decides whether a general manager opens the dashboard tomorrow morning is a design question and a management question. Design question: can they read it in a minute on a phone. Management question: are they going to be asked about a number on it this week.
If both answers are yes, adoption sticks. If either answer is no, the dashboard dies. I have watched Power BI dashboards fail and a Google Sheet succeed inside the same company. The Google Sheet had five columns and was tied to a Monday meeting. The Power BI dashboard had 34 tiles and no meeting behind it.
Five tiles, and only five
The general manager's home screen dashboard shows exactly five numbers. Not four. Not seven. Five. Any more and the eye stops knowing where to land.
Fig. 1 · The five-tile home view on a phone.
Tile one: sales versus forecast, yesterday
Not sales versus last year. Not sales versus budget. Sales versus the forecast the general manager submitted for yesterday. This one shift matters. The general manager owns the forecast, so the general manager owns the miss. Sales versus last year is a corporate number and it teaches nothing about operator decision quality.
Tile two: labor as a percent of sales, yesterday
A single number, with the target next to it. Not a chart. Not a trend line. Yesterday's number. The general manager looks at this and knows in one glance whether they overstaffed a slow section or got caught short on a busy one. If the number is bad, they call the scheduler before the shift starts today.
Tile three: food cost variance, week to date
Food cost is not a daily number, it is a running number. Show the week-to-date variance against theoretical, in points. Positive is bad. Green under a threshold, red over. Anything more is decoration.
Tile four: comps and voids, yesterday
The dollar amount and the count. Comps that spike overnight are almost always a service recovery event, a training gap, or a theft pattern starting. The general manager should see the number before the area director sees it.
Tile five: the variable KPI
This is the tile that changes. Rotating quarterly. Table turn time in a full-service restaurant. Catering on-time rate in a catering-heavy unit. Drive-through window time in a quick service. This is the tile you tie to the weekly one-on-one. This is the tile that drives learning.
What gets watched gets moved. What gets watched together, with a real conversation on Monday, gets moved fast.
Design for a phone, in one hand, with coffee
The general manager opens this at 7am. In a car. On a train. Standing in the walk-in doing a temperature log. If the dashboard needs a login prompt, a two-factor code, a filter dropdown, or a scroll, they will not open it a second time.
What that means in practice:
- No login on the phone. Persistent session, biometric unlock. The dashboard opens straight to the five tiles.
- No filters. The dashboard already knows which location this general manager runs. It filters itself.
- No horizontal scroll. If a tile does not fit on a portrait screen, it is not on the home view.
- Big numbers, small labels. The number is the point. The label is the context.
- Color is a signal, not decoration. Green good, red bad, gold "focus this week." Nothing else.
A general manager who reads the dashboard in sixty seconds standing up will open it every morning. A general manager who has to sit down and think about it will not.
The data has to be ready before they wake up
This is the boring piece that decides everything else. The dashboard has to be refreshed by 6am local time, seven days a week, no exceptions. If yesterday's numbers land at 10am, the dashboard becomes a review tool instead of a decision tool, and general managers do not use review tools.
What that requires in the pipeline:
- Toast (or Square, or whatever POS) posts closing data by midnight local. This is a POS setup question, not a data question.
- The extract job runs at 3am, pulling the previous day's tickets, labor, comps, voids, and menu mix.
- 7shifts labor data pulls in the same 3am window.
- QuickBooks Enterprise or the accounting system feeds actuals for anything past week to date. Weekly rather than daily is fine here.
- Power BI (or Looker Studio, or Metabase, or the tool of your choice) refreshes at 5:30am.
- The dashboard is live on the phone at 6am.
None of that is exotic. It is Zapier, Make, or a scheduled query. If the operator does not have someone in-house to build it, a decent freelancer builds this pipe in a week. The rate-limiting step is the design discipline, not the engineering.
Fig. 2 · From closing shift to morning phone in six hours.
Tie one tile to the weekly meeting
This is the piece nobody wants to talk about, because it is a management piece and not a data piece. The dashboard does not get used because the numbers are interesting. The dashboard gets used because someone asks about them.
Here is the actual mechanic. Every Monday one-on-one between the area director and the general manager starts with the same three questions:
- What happened with the sales-versus-forecast tile last week? Where were we off, and what changed?
- What did the labor tile look like on Friday and Saturday, and why?
- What is happening with the focus tile this week?
The general manager knows those questions are coming. So they open the dashboard on Sunday night. Then on Friday. Then, inside three weeks, every morning.
You do not need a training program for the dashboard. You need a meeting that references it. That is the whole adoption strategy.
Where operators go wrong
The mistakes I have made and the mistakes I see other operators make, in order of how much time they cost:
Building for the operator, not the general manager
The operator wants to see year-over-year comparisons, cohort trends, franchisee benchmarks, and a heat map. The general manager wants to know if they overstaffed the lunch. Build the general manager's view first. Build the operator's view on a separate tab, or a separate tool entirely, because those two audiences will never share a dashboard well.
Waiting for perfect data
The reason most dashboards never ship is that the data has three known issues and the team wants to solve them first. Ship the dashboard with the known issues documented. Fix them in place. A working dashboard with a footnote beats a perfect dashboard that ships in nine months.
Adding a tile every time somebody asks
The corporate finance team wants a tile. The marketing team wants a tile. The training team wants a tile. Every added tile costs three general managers their weekly open. Say no. Give those teams their own view. Keep the phone view at five.
Treating the dashboard as done
The variable tile has to change every quarter. The thresholds have to move as the business changes. The dashboard is a living operating tool, not a finished product. Someone has to own it and edit it. Usually that is the operator or the area director, not IT.
The point
A general manager who opens a five-tile dashboard every morning is running a different restaurant than a general manager who checks numbers at the end of the month. That gap is enormous, and it costs almost nothing to close.
Design for the phone. Refresh by 6am. Five tiles. Tie one tile to a real conversation on Monday. That is the whole thing.
The tools will change. The design principles do not.