Turn a losing week into a plan
A worked example: describe a bakery that lost money, attach its week as a workbook, solve next week's plan, test it with what-ifs, verify one and compare them on a workspace.
Where: New Model
Harbor Point Bakery lost money last week. It bakes sourdough loaves and croissants in one oven, buys flour and butter, and sells to walk-in customers and to Pier Cafe. This walk turns its numbers into next week's plan, checks the answer, and pressure-tests it before anyone acts on it. Allow about an hour; most of it is waiting.
Download the week: harbor-point-week.xlsx. Its eight sheets hold the recipes, what the vendors charge, the oven's 60 hours and costs, what each customer buys and pays, and delivery costs. They're listed on Start from your workbook.
Describe the business
In the rail, click New Model, then Next, Basic, Next, and name the model. On Describe your
goals, set What to optimize to Maximize weekly profit and Hard limits to
the oven runs at most 60 hours a week. On Describe your business, type:
Harbor Point Bakery is a neighborhood bakery with one oven. We bake sourdough loaves and croissants, buy flour from Mill Supply and butter from Dairy Co-op, and sell to walk-in customers and to Pier Cafe. Last week we lost money, and we want the plan for next week that makes the most profit. The attached workbook has our recipes, prices, oven hours and what each customer will buy.
Attach the week's workbook
Click Add on Upload or select relevant datasources, then Upload new datasets and attach them to this model. Choose Microsoft Excel, drop the file, click Continue → and Add source, name the dataset, and click Import data →. Click Finish.
If the window answers Build failed — connection expired or unknown — reconnect your source, close it and import the file again; our second try went through.
Agree how the sheets were read
Click Next; after about a minute and a half Step 2 lights up Next. Click it to open Your data, as I read it. WHERE THE NAMES COME FROM should list the four items, the two customers, the five places and one period, Week 1. Under What to call things, pick a name for each part (we took Stock at each site, Customer orders, Purchases, Production), click Use all sheets, then Next.
Say what to decide
What you want to decide starts with your description. Replace it with one sentence:
Decide how many sourdough loaves and croissants to bake next week, and for which customers, to make the most profit within the oven's 60 hours. Click Next.
Settle the terms and the gaps
On The columns and the terms, answer each marked term with its button. Ours asked units and currency: kg per loaf or croissant, dollars per kg, per hour, per week, per item. Click Next.
On What your data does not say, answer each card with Type your own answer and Save answer, or Scope it out. We told it flour and butter don't use the oven, there's no stock on hand and nothing is kept to next week, the $150 is paid every week, and demand figures are ceilings; we scoped out vendor minimums. Click Design on this reading.
Plan one week, then sign off the design
The design took about ten minutes. Click Next on The design is drafted. The card assumed 13 weeks: click Approve on that value, then One combined period under Planning horizon. The redraw took about three and a half minutes; reload the page to see SCOPE read one period.
Under Rule choices from your brief, waiting for your word, click Confirm on The model is built for the most profit, and Leave this out on the rest (a row can come back once; click it again). Approve the last assumed value, Confirm each section to 5 / 5 confirmed, and click Confirm and map your workbook. See Review the design.
Build on your rows and solve
About forty seconds later Step 4 shows Your data, bound to the model, with every sheet placed. Click Build and solve on your data. It solved in about a minute; click Next.
Your baseline reads $1,422. That's the hand answer: an oven hour earns most on walk-in croissants, then Pier Cafe croissants, then walk-in loaves, so the plan sells all 200 + 150 croissants and 80 walk-in loaves (55 hours) and spends the last 5 hours on 20 of Pier Cafe's 40 loaves.
Ask what more oven time is worth
Click Go to Model Overview, then Scenarios. Click + New, name it Oven runs 70 hours, Create,
and type What if the oven could run 70 hours a week instead of 60? in What if…. Click Run scenario.
It took about three and a half minutes.
Bakery profit reads $1,524, +$102 vs base. The editor's text says why: the extra hours bake Pier Cafe's other 20 loaves, and then the oven has 5 hours to spare.
Ask what a flour price rise costs
Add a second what-if the same way: Flour costs $2 a kg, with
What if Mill Supply raises flour to $2.00 a kg? It returned $1,354, −$68 vs base: the plan doesn't
change, and the 85 kg of flour cost 80 cents more each. Click a scenario in the list to see its tiles.
Verify the one you'd act on
Select Oven runs 70 hours and click Verify. The drawer says how far to trust the number: a confidence badge with its reason, the result, each KPI marked Verified or Estimated, and under Inputs the two oven-hour changes your what-if made, marked ← you (override). Export gives Evidence JSON or Evidence Markdown for the record. See Verify.
Put them side by side on a workspace
Click Browse all 3, tick both what-ifs, and click Compare in the bar that shows 2 selected. eode opens a workspace, Docs walkthrough — harbor point · compare in our case, with Base as the anchor.
Scenario comparison ranks the three objectives and lists every KPI's change; Financials comparison lays out the income statement for each. Base's adds up to the hand answer: Gross Sales $2,105, less Net Purchases $242, Labor $240 (the oven's running cost), Fixed Processing $150 and Distribution $51, is $1,422.
To arrange the tiles or share the workspace, see Workspaces.
Related
- Start from a workbook: the reading, the terms and the gaps, screen by screen
- Write and run a what-if and Read a scenario's results
- Scenarios: Browse all, selecting and comparing
- Widgets: what each tile on the compare workspace shows