Everyone blamed the new kitchens. The cause was a lunch menu.

It started with a question.

Amara Reidruns growth at Simcorp Kitchensin Veloccy

Orders per user has been sliding for three or four weeks, and I can’t tell if it’s the Gurgaon kitchens ramping or something else.

AI with access

Access

Experience

none

−6.2% orders per user per month, four weeks running

With access to the data, it is the new kitchens.

Three kitchens had just opened in a new city, and new kitchens start slow. Total orders were up, so the launch looked like it was working and frequency would recover by itself.

AI with experience

Access

Experience

Lunch launch · 7 kitchensSessions feed dropping eventsand more

132.6% of the decline sits in the seven launch kitchens

With experience of the company, it is a menu.

The new kitchens had improved. On Veloccy’s calendar was something another team had done: kitchen operations had launched six lunch items in seven of the older kitchens. Those seven carry more than the whole decline.

How it happened in Veloccy.

A simulated ten-minute food company with fifteen of its own kitchens. We built it and planted the cause, so we know what the right answer is. These are a few moments from the work.

Simcorp Kitchens Arcs GRO-1Orders per user is sliding

Learned

What it read What it led to Who is on it

Orders per userLunch launch · 7 kitchensGurgaon kitchens openedSessions feed dropping eventsLaunch kitchensOffice customers

GRO-1 · Orders per user is slidingArc · learned

Menu fulfilment issue rateOPS-1Second lunch lineThe kitchen team’s dashboardThe second line works

Amara ReidGrowthRafael VegaKitchen operations

The arc for this work, as the team sees it: what Veloccy read, what the work led to, and the thread it happened in.

01It starts from what it knows

Veloccy checked what every team had changed before it looked for a cause.

It set a known data problem apart first, so the drop that was left was real. Then it ruled the new kitchens out with a number, and went to the lunch launch that operations had put on the calendar.

Simcorp KitchensArcsGRO-1

Amara Reid

Frequency looks down. I can’t tell if it’s the Gurgaon kitchens ramping or something else.

Started from

Lunch launch · 7 kitchensGurgaon kitchens openedSessions feed dropping events

The answer

It is not Gurgaon. The seven kitchens that launched the lunch menu carry more than the whole decline.

Gurgaon frequency rose 38.7% and is hiding part of the drop. In the launch kitchens, lunch customers hit “not available” far more often, and office customers are ordering much less.

Every number opens the query behind it

02It finds what is missing

The number that explained it did not exist, so Veloccy proposed it.

Nobody was measuring how often the menu let a customer down. Veloccy defined the metric, worked out its usual range, and gave it to the person who runs the kitchens. It has been checked every morning since.

Simcorp KitchensBrainMetrics

Metric

Menu fulfilment issue rate

Proposed by Veloccy · owned by Rafael Vega

Usual range: 2.73% to 3.27%
Launch kitchens: 12.44%
Everywhere else: 3.10%
Checked: Every morning

03It comes back

Operations made a bet on four kitchens, and Veloccy read the result.

A second lunch prep line went into four of the seven kitchens. The other three were held back on purpose. What would count as working was written before the first cook was hired.

Simcorp KitchensArcsOPS-1Readout

Veloccy came back with the result

Readout · OPS-1

The second lunch line worked. Frequency is back inside its usual range.

Second line3.27%

Held back10.09%

Expected at most 8.0%, and at least 3 points better than the kitchens held back. Both started at 12.44%.

Other meals held
Order volume held
Basket size held

DecisionShip. Extend to the other three kitchens.

Why access to the data was not enough.

01The obvious answer fits the data.

New kitchens do pull an average down. Knowing these had not takes knowing what another team changed that month.

02The metric did not exist.

You cannot query a number nobody defined. Something has to notice that it is missing.

03The fix was tested, not assumed.

Veloccy said what was not proven, too: seven kitchens is a small group, and they were not chosen at random.

What stayed in the Brain.

Every one of these is experience the next question starts from, in every team. See the Brain

KSimcorp Kitchens

MetricMenu fulfilment issue rate
A new metric with an owner and a usual range.
CalendarLunch launch · 7 kitchens
The lunch launch and the fix, with what each did.
DashboardThe kitchen team’s dashboard
A dashboard the kitchen team opens themselves.

Next use case · Performance marketing Return on ad spend fell. Nobody on the team had changed anything.

Which bet is on your desk this month?

Show us, and we will show you what Veloccy does with it.

Talk to us