Every operator I know has run the same pre-shift meeting for years. The GM walks in with a clipboard, reads yesterday's sales, calls out labor, mentions a VIP, points at a couple of prep items, and asks if there are any questions. Nobody has questions. The team goes to their stations. Eleven minutes have gone by and no coaching happened.
The meeting is not broken because managers are lazy. It is broken because the numbers-recitation part swallows the coaching part every single time. There is a fix, and the fix is not a better meeting agenda. The fix is to move the numbers to the phone before anyone walks in the door.
I have watched this play out in two very different environments. Inside 21 franchise units in six states, running out of Walmart, Sam's Club, Whole Foods, and Target boxes with a $36M P&L, the pre-shift meeting had to move fast because turnover of shift leads was the constraint. In a $30M Michelin-recognized Bay Area group serving 215 employees and enterprise catering for Stanford, Google, Apple, Meta, LinkedIn, Salesforce, Cisco, Adobe, and Nvidia, the pre-shift meeting had to move deep because standards were the constraint. Same automation solved both. The template is the same because the problem is the same: the leader's fifteen minutes is the scarcest resource in the operation, and it gets spent on reading numbers out loud.
What the automated brief actually contains
Eight blocks. Not twelve. Not fifteen. If the brief scrolls past two thumb-flicks on a phone, half the shift leads will stop reading at block five.
The eight, in order, because order matters:
- Yesterday's sales versus forecast. One line. Actual dollars, variance in percent, and a tiny arrow. Nothing else.
- Today's forecast. Cover count if it is a full-service unit. Transaction count if it is a counter concept. Same one-line format.
- Weather. High, low, chance of rain. Weather is not a novelty item for a patio unit. A 20 percent chance of rain moves covers.
- Staffing versus need, by hour. A three-column table for lunch, mid, and dinner blocks. Scheduled versus modeled need versus the delta.
- Prep priorities. The top three items from the prep sheet, in the order the kitchen should attack them.
- VIP reservations and notes. Name, party size, time, tag (repeat guest, allergy, birthday, press).
- Red flags from the closer. Two-line summary from whoever locked up. What broke, what got 86'd, what the guest complained about at 10:47pm.
- Focus KPI for the week. One number, one target. Same number for a full quarter.
That is the whole brief. It fits in a Slack message. A shift lead reads it standing at the bus stop or sitting in a parked car. By the time they walk into the building, they know what today looks like and what yesterday cost.
The wireframe
Here is what the rendered brief looks like on a phone at 8:01am. It is intentionally boring. Boring is what gets read every day.
Fig. 1 · The brief as it renders in Slack, 8am local.
Two things worth pointing at. The staffing block uses red only where there is a gap. The rest of the shift is plain cream. The eye lands where it needs to. The closer's red flag has its own color band, because it is the one item that came from a human at 11pm and has to survive.
Where the data comes from
None of this is exotic. Every input is a system the group already pays for. The only thing being built is the connector between them.
Fig. 2 · Five inputs, one glue, two delivery targets.
The scenario runs once per unit at 7:55am local. Time zone is per-unit, not global. In a group spread across Northern California, Nevada, and Texas, the trigger fires four times a morning at different clock times so each unit gets the brief on their own 8am.
Build time for the first unit is about a weekend, most of it spent shaping the message template and testing the Toast auth. The second unit is a duplicate scenario with a variable swap. Unit 21 takes about eleven minutes to add.
Why manual pre-shift meetings drift
I have sat in hundreds of pre-shift meetings across 12+ new-unit openings and six states of franchise territory. The pattern is almost always the same.
The GM opens with numbers because the numbers feel like the responsible thing to lead with. Sales, labor, comps, top movers. The team stands there nodding. Half of them are thinking about the walk-in they need to break down. The other half are checking the clock because their side work is not going to do itself.
By the time the GM finishes the numbers, they have four minutes left. The coaching moment they planned, the one about handling the two-top that walked out unhappy on Saturday, gets compressed into a sentence. "Let's make sure we are checking back at the two-minute mark." Nobody remembers it by 6pm.
The pre-shift meeting was never supposed to be a data recitation. It was supposed to be the fifteen minutes a leader owns before the noise starts.
When the numbers arrive on the phone at 8am, the GM walks in already knowing what everyone else knows. The meeting starts at the coaching moment. Fifteen minutes of standing time gets spent on the thing that actually moves service, not on reading a report out loud.
What breaks first, and how to fix it
The pipe itself is stable. Toast, 7shifts, Resy, weather, Airtable, Make. Those APIs do not fall over on a random Tuesday. What breaks is the human input at the front and the human attention at the back.
The closer's notes go stale
Everything except the red-flag field is a system pulling from another system. The closer's note is a person writing in an Airtable form at 11pm after a long shift. If the form is optional it does not get filled in. If it does not get filled in for three nights in a row, the red-flag block starts showing "no notes" and the shift lead trains themselves to ignore that section.
The fix is to make the closer's note a required step before the deposit report runs. Airtable form with three fields: what broke, what got 86'd, what a guest said. Two minutes. Non-negotiable. Once it is part of the close checklist it stays part of the close checklist.
Slack starts feeling like noise
If the brief posts into a busy channel with 40 messages an hour, it scrolls past. The fix is a dedicated channel per unit, muted for everyone except the shift leads for that unit. The DM to the GM sits in their sidebar. Both places, no email.
The focus KPI stops changing
The eighth block is supposed to rotate quarterly. Table turn time in Q1, catering on-time in Q2, first-guest response in Q3. If the KPI never rotates, the brief becomes wallpaper. The rotation has to be a calendar item the operator owns, or it will not happen.
The GM starts editing the template on their own
Six weeks in, a well-meaning GM will decide they need a ninth block. Then a tenth. If the template drifts per unit, comparability is gone and the area director cannot use the brief to spot patterns across the fleet. Lock the template at the group level. If a unit needs a custom line, put it in the closer's note field for that unit only, not in the standing brief.
Slack outages, API rate limits, and Sundays
Slack has an outage every six months or so. Toast throttles pulls when a big software update hits. Sundays have half the reservation volume and skew the forecast if the model is untuned. None of those break the brief permanently, but each one shakes confidence for a week. Build a fallback that catches the exception and posts a plain-text "brief unavailable, check the dashboard directly" message. Silence is worse than a stub. A shift lead who opens Slack at 8:03am and sees nothing assumes the tool is broken.
What it cost, what it returned
Rough numbers from the Hana Group build, spread across the $36M P&L, 21 units, six states. Build cost was one weekend of my time plus a Make subscription that runs about $34 a month for the volume. Airtable was already in the stack.
The return is harder to price because it is a time reallocation, not a cash line. Every shift lead across every unit reclaims about eight minutes of the pre-shift huddle, five to seven times a week. Across 21 units that is roughly 100 hours a week of newly available leader attention. What that time gets spent on decides whether the return shows up in retention, in guest recovery scores, or nowhere.
In the units where the GM used the freed time for a specific coaching topic each week, hourly-team retention moved. In the units where the freed time went into a longer smoke break, nothing moved. The tool is neutral. The management practice around it is not.
The rollout
The pattern that works is not "build once, deploy to all units on Monday." It is more like this:
- Build for one flagship unit. GM who is a builder. Let them shape blocks four through seven with you.
- Run it for two weeks. Watch what gets read and what gets ignored. Cut a block if you have to.
- Duplicate to two more units. Different demographics, different volume. Watch again.
- Standardize the template. Duplicate across the rest of the fleet in a single day.
- Add the pre-shift huddle expectation to the operating rhythm. Area directors reference block eight on their weekly one-on-one.
The rollout that fails is the one that skips step five. Without the operating rhythm behind it, the brief becomes a beautiful thing nobody reads. With the rhythm behind it, the brief becomes the phone-first version of the operating meeting.
A quick word on the AI layer
A lot of operators want to skip straight to "have ChatGPT summarize the numbers." I have built custom GPTs for exactly this use case and they work, but the sequence matters. Get the plain-text brief working with hardcoded blocks first. Prove the operating discipline. Then, if you want, add a small language model step at the end that writes a two-sentence narrative on top of the numbers. Something like "labor ran two points hot on Saturday dinner because two servers picked up shifts they were not scheduled for. Watch the mid tomorrow, it is projected to be slower than the prior week."
The AI layer is a nice addition. It is not the point. The point is that the numbers land on time in the right place. A brief with no AI narrative that lands at 8am beats a brilliantly summarized brief that lands at 9:15am, every day of the week.
The point
Automation in restaurants is not about replacing people. It is about moving the boring parts of a leader's job into a machine so the leader can spend the fifteen minutes they have with their team on something that matters.
An 8am brief on a phone is not a technology story. It is an operating discipline story with a small technical wrapper. Toast, 7shifts, Resy, weather, Airtable, Make, Slack. None of that is new. The new thing is deciding that the pre-shift meeting is worth protecting from itself.
Build the brief. Route it to the shift leads. Then walk into the huddle without the clipboard.