Should I add a mid shift? Ask your sales by hour
The lunch line has been out the door three days running, and the opener is calling for help. The question is the oldest one in scheduling: is that a real pattern worth a shift, or a busy week? To answer it properly, you'd pull hourly sales from the POS and line them up against who was working each hour, which lives in a different app. Most of us skip that and go with our gut.
When Mays has both, the sales by hour and the schedule, your AI app can line them up in one answer.
Part of our series Run your restaurant with your AI: one real situation per post, with the words you type and what happens in Mays. The prompts and results are illustrative examples.
What the manager types
"At Midtown, over the last 4 weeks on weekdays, what do sales look like hour by hour, and how many people were scheduled and actually on the clock each hour? Would a 10:30 to 2:30 mid shift make sense?"
What the AI does in Mays
It asks Mays for sales by hour at Midtown, with staff. That's the data behind the Staffing vs Revenue heatmap on Insights, and it's read-only. For each hour of the day on Midtown's clock, Mays returns:
- Net sales, orders and average ticket, as totals for the range and as a typical day (the average per day).
- Each hour's share of the range's sales.
- The average number of people scheduled in that hour and the average on the clock. An open shift counts as one scheduled person.
- Labor cost and labor % per hour, only if you can see pay and included pay when you connected.
It can split the same numbers by weekday (each weekday's average over the range), or group them into dayparts instead of hours: breakfast 6–11 am, lunch 11 am–2 pm, afternoon 2–5 pm, dinner 5–9 pm and late night 9 pm–6 am. One question covers up to 31 days.
What you see
The AI reads the pattern back, for example (illustrative):
"On a typical weekday at Midtown, 11 am to 1 pm is 31% of sales, with 2.1 people on the clock. 7–9 am has 3.4 people on for 18% of sales. The busiest hour is 12–1 pm, about 46 orders on average with 2 people on. Thursday and Friday lunch are the strongest. A 10:30–2:30 shift would put a third person on through the whole lunch peak."
The numbers come from Mays; the suggestion is the AI's reading of them. You decide.
If the answer is yes
"OK. Add a 10:30 to 2:30 open shift at Midtown on Thursday and Friday next week, barista."
The AI asks Mays for a preview of each shift. Shifts are added as drafts, like in the schedule builder: nobody is notified and staff don't see them until you publish the week. The preview also tells you what the shift does to the week:
- the week's scheduled hours before and after (and the day's);
- which coverage gaps it closes, if any;
- who it would put into overtime at that location, for a shift with a person on it;
- with pay access, the labor cost before and after and labor as a % of the week's sales forecast, against the location's labor target.
Overlaps, time off, availability, minors' rules and overtime are warnings, never blocks: the AI tells you, and you decide. When the week is right, publish it: "Publish next week at Midtown." Publishing has its own preview: what staff see now against what they'll see after, and who gets notified (by default, only the people the changes affect).
Put a person on it instead of an open shift if you already know who: "…for Maya L."
What it replaces
Before | With Mays connected to your AI |
|---|---|
Export hourly sales from the POS | Sales by hour come from Mays |
Count heads per hour from the schedule by hand | Scheduled and clocked-in people per hour, in the same answer |
Guess whether it's worth the labor | Labor % per hour, with pay access |
Add the shift in another app | A draft shift with its effect on the week, then publish |
Who can do it
Sales by hour need access to Insights in Mays. Adding shifts needs permission to edit the schedule at that location, plus the organization's Allow AI apps to take actions setting and make changes ticked when you connected.
What Mays doesn't do here
- It doesn't forecast a new shift's sales. It shows what happened. A typical day is an average, so a single big catering day can pull it up; look at the per-weekday view before you decide.
- Salaried pay isn't split by hour, so hourly labor cost covers hourly staff only.
- It doesn't publish when it adds. Draft shifts stay drafts until you publish the week.
For the numbers behind the decision, see The numbers a new restaurant manager owns.
Try it on your busiest daypart
- Connect Mays to ChatGPT, Claude or Grok from mays.app/ai (ChatGPT, Claude, Grok). Connect Mays to ChatGPT, Claude or Grok has the details.
- Ask for sales by hour with staff for the last few weeks.
- Add the shift as a draft, check the preview, and publish when the week is right.
Try Mays free for 30 days, with no credit card, and connect it to the AI you already use.