Start

Guides / Infor exports / Guide 182

Free guide

Build a Weekly Dashboard From Your Infor Export

You will end with: an open-orders dashboard that a checked script rebuilds the same way from every week's Infor export, with every set-aside row listed.

Difficulty: 3 of 5 stars, Build+25 pts, points for this guide25-minute guide
ForPlannerControllerOwner
Build a weekly dashboard from your Infor export. An export grid becomes four dashboard tiles.
  • TimeAbout 25 minutes
  • ToolsInfor CloudSuite Industrial export and Claude, with code execution on
  • Cost$0 to practice
Tools in this guideAll tools
  • Your ERP system
    Work tool. The company system that runs orders, inventory, and purchasing. Guides use exports from it, not a live link.
  • Claude
    AI app. An AI chat app. You type or paste text, or add a file, and it writes back.
  • ChatGPT
    AI app. An AI chat app. You type or paste text, or add a file, and it writes back.

Answer. To build a weekly dashboard from an Infor export, export the Customer Order Lines grid to Excel, have Claude map its columns once, and then have Claude write a small script that turns any week's export into the dashboard page. You check the first page against an answer key. Every Monday after that, the same script rebuilds the page the same way.

Time and cost. About 25 minutes the first time, and about 5 each week after. $0 to practice with the sample files on the free Claude plan. Who this is for. Planners, controllers, and owners at shops that run Infor CloudSuite Industrial, who want the open-orders picture every Monday without building it by hand.

This guide assumes Infor CloudSuite Industrial, which Infor also sells as SyteLine. It is the Infor ERP that most small and mid-size discrete manufacturers run. The steps work the same in the cloud version and in recent on-premises versions (10 and later), because they use the same forms. If your company runs Infor LN or M3, the idea is the same, but the screens and column names are different.

In this guide, Hector Lee, the planner at Mesquite Ridge Metal Works (a made-up company), exports Monday's open customer order lines. Carol, the owner, wants four numbers every week: how many lines are open, what they are worth, how many are past due, and what those are worth. Claude does the fuzzy part once, which is reading Infor's column names. A script does the counting every week after that.

Diagram: the Infor export goes to a column map, written once by Claude and checked by a person. The map and the export go into infor_dashboard.py, which makes the dashboard page. Next Monday, a new export goes into the same script and map. Rows the script cannot use go to a set-aside list with the reason. Same export, same page, every time.

What will I learn besides building a dashboard?

You will learn six habits that work for any report you build from an ERP export, with a short "why" for each later in this guide.

  1. Map the columns once. AI reads the export's column names one time, and you check the map.
  2. Have AI build the tool once. Claude writes a script, and the script makes the page every week.
  3. Describe the page you want, not the code. You list the tiles and tables, and Claude works out how.
  4. Set bad rows aside, and never guess. A row the script cannot trust goes on a list with its reason.
  5. Check the first page against an answer key. You compare it with numbers you know are right.
  6. Rerun the same script every week. Same rules, same page, so the numbers can be compared week to week.

What do I need before I start?

You need Claude, the two sample exports, and the answer key. There is nothing to install.

Safe use. These files are made up. A real export holds customer names and prices, and what you attach goes to the AI company. Check your company's AI rules before you attach a real one. Read the safe use notes.

How do I build a weekly dashboard from an Infor export?

Look at the export first, have Claude map its columns, have Claude write and run the dashboard script, check the page against the answer key, and run the same script on next week's export. Six steps.

Step 1. Look at the export before Claude does

Open mrmw-co-lines-2026-11-16.xlsx in Excel and close it without saving. Two minutes here makes every later check faster, because you will know what the traps are.

Row 1 is a sample-data note, and the column names are on row 2. The Order numbers have spaces in front, because Infor pads them to a fixed width. Rows 20 and 21 are the same line twice. Row 23 has no due date. Order 48195 shows up three times, as releases 1, 2, and 3. That is not a duplicate, because each release is its own delivery. The Status column has four values: Ordered, Planned, Filled, and Complete.

The Infor Customer Order Lines export in Excel, rows 19 to 26. Red dots mark the traps: order numbers with spaces in front, rows 20 and 21 the same order 48210 line twice, a blank due date on row 23 circled, a Planned status on row 25, and a quantity stored as text 4.000 on row 26.

Step 2. Have Claude map the columns, and nothing else

Open a new chat at claude.ai, attach mrmw-co-lines-2026-11-16.xlsx, and paste the prompt below. It asks Claude to fill in a small map with a fixed shape: which column holds the order, the line, the due date, and so on. If a field is not in the export, Claude writes null instead of guessing. The prompt also fills in what each Infor status means. In CloudSuite Industrial, Ordered is a firm open line, Planned is not firm yet, Filled has shipped, and Complete is done.

I attached a Customer Order Lines export from Infor CloudSuite Industrial (SyteLine), made with Actions > To Excel. I want a weekly open-orders dashboard from exports like this one. First, only map the columns. Do not build the dashboard yet.

Read the file. Row 1 may be a note, so find the row that holds the column names. Then make a file called infor-column-map.json in exactly this shape. Use the column names exactly as they are spelled in the export. Use null for any field you cannot find. Never guess.

{
  "header_row": <Excel row number of the column names>,
  "order": "", "line": "", "release": "",
  "customer_name": "", "item": "", "status": "",
  "qty_ordered": "", "qty_shipped": "", "due_date": "", "unit_price": "",
  "status_values_found": [],
  "open_statuses": ["Ordered"],
  "not_firm_statuses": ["Planned"],
  "closed_statuses": ["Filled", "Complete"]
}

Then tell me in plain words, in a short table: which column you picked for each field, and each status value you found with how many rows have it. After the table, list anything odd you saw in the data, such as numbers stored as text, spaces in front of values, dates written different ways, blank cells, or the same line twice. Keep it short.

Claude makes infor-column-map.json and replies with a table of the columns it picked and how many rows have each status. Check two things. The customer field should be Name, not Customer, because the Customer column holds codes like C000101. The status counts should be Ordered 23, Planned 1, Filled 2, and Complete 1. Download the map. You can compare it with my map.

Claude chat with the Infor export attached. Claude made infor-column-map.json and shows a table: order is Order, customer_name is Name (circled), due_date is Due Date. Status counts: Ordered 23, Planned 1, Filled 2, Complete 1. Odd things: spaces in front of order numbers, 40.000 and 4.000 stored as text, rows 20 and 21 the same line, row 23 no due date, and the three releases of order 48195 are not duplicates.

Step 3. Have Claude write and run the dashboard script

In the same chat, paste the build prompt. You are describing the page you want, not the code: four number tiles, a table by due week, a table by customer, the past due lines, and the rows set aside. The six rules make the page safe. The most important one is rule 1. It says one order line is Order plus Line plus Release, so the three releases of order 48195 stay separate and the true duplicate gets set aside.

Now write a Python script called infor_dashboard.py that builds the dashboard from any week's export in this shape, using infor-column-map.json. Find every column by its name from the map, never by its position, so the script still works if the columns move.

What goes in: the export file and infor-column-map.json. The export date is in the file name (mrmw-co-lines-YYYY-MM-DD.xlsx), and that date is "today."

Rules:
1. One order line is Order + Line + Release. Take the spaces off the order number. If the same Order + Line + Release is in the file twice, keep the first one and set the second aside.
2. Open qty is Ordered minus Shipped. Some quantities are text, like "40.000". Read them as numbers.
3. A line is open when its status is in open_statuses, its open qty is above 0, and it has a due date. Set aside every other row with the reason. Lines in not_firm_statuses also get their own count and value, "Planned, not firm," which is not in the totals.
4. Due dates can be real Excel dates or text like 11/20/2026 (month/day/year). If a due date is blank or cannot be read, set the line aside. Never guess a date.
5. If a column named in the map is not in the export, stop with a plain message that names the column, and do not make a page.
6. Open value is open qty times Unit Price.

What comes out: one HTML page, open-orders-dashboard-YYYY-MM-DD.html, that opens in a browser with no internet.
- Four number tiles at the top: open lines, open value, past due lines, past due value.
- Open lines and open value by when they are due: Past due (before today), This week (today through 6 days later), Next week (the 7 days after that), and Later.
- Open lines and open value by customer, largest value first.
- The past due lines, oldest due date first.
- Planned, not firm: lines and value.
- Rows set aside: every row not counted as open, with its Excel row number and the reason.
- A check line at the bottom: rows in the export = open lines + rows set aside.

Run the script on the export I attached. Give me the page and the script to download. Then run it a second time and tell me whether the page came out the same.

This takes a minute or two. Claude writes infor_dashboard.py, runs it, runs it a second time to make sure the page comes out the same, and gives you open-orders-dashboard-2026-11-16.html. Download both files and double-click the page to open it in your browser.

The open-orders dashboard page for 2026-11-16. Four tiles: 21 open lines, $333,580.00 open value, 3 past due lines, $31,044.00 past due value. Below, a table by when due: past due 3, this week 6, next week 6, later 6. A customer table starts with Prairie Wheel Equipment at $121,260.00.

Step 4. Check the page against the answer key

Check the page itself, not Claude's summary of it. Open answer-key/dashboard-2026-11-16.md and compare these by hand.

  • The four tiles: 21 open lines, $333,580.00 open value, 3 past due lines, $31,044.00 past due value.
  • The past due lines are orders 48150, 48166, and 48171.
  • Six rows are set aside: rows 6, 7, and 8 (Filled or Complete), row 21 (the same line as row 20), row 23 (no due date), and row 25 (Planned, not firm).
  • The check line at the bottom says 27 rows = 21 open lines + 6 set aside.

If a number does not match, do not keep this version. Tell Claude which number is wrong and what the answer key says, and ask it to fix the script and run it again. (How to test an AI tool against answers you know.)

Step 5. Run the same script on next week's export

This is the step that makes it a weekly dashboard. Start a new chat. Attach mrmw-co-lines-2026-11-23.xlsx, infor_dashboard.py, and infor-column-map.json. Paste the prompt below, with last week's four numbers at the end.

I attached this week's Customer Order Lines export, plus infor_dashboard.py and infor-column-map.json from last time. Run infor_dashboard.py on the new export exactly as it is. Do not change the script or the map.

If the script stops, tell me its message and what it means, and do not work around it. If it runs, give me the new page to download and tell me in four short lines: open lines, open value, past due lines, and past due value, each next to last week's number.

Last week's numbers: [paste the four numbers from last week's page]

Next week's export has a surprise in it. Someone added a Promise Date column to the grid, so every column after Due Date moved one place to the right. The script still works, because it finds each column by its name from the map, not by its place. The new page shows 19 open lines, $328,254.00 open value, 2 past due lines, and $14,950.00 past due value, which matches answer-key/dashboard-2026-11-23.md.

Step 6. Save the script and the map together as version 1

Make a folder called infor-dashboard. Put infor_dashboard.py, infor-column-map.json, and a plain text file called what-changed.txt in it. Write one line in the text file, such as "v1. Open-orders dashboard from Customer Order Lines. Checked against the answer key for 11-16 and 11-23." (Save every working version.)

Every Monday, export the grid, save it with the date in the file name, and do Step 5. Save each week's page too, so you can look back. That takes about five minutes.

What does good look like?

With the two sample exports, the dashboard shows these numbers. Answer key, sample data.

Export Open lines Open value Past due lines Past due value Set aside
2026-11-16 21 $333,580.00 3 $31,044.00 6
2026-11-23 19 $328,254.00 2 $14,950.00 3

For 2026-11-16, the lines by when due are: past due 3 ($31,044.00), this week 6 ($83,660.00), next week 6 ($112,900.00), and later 6 ($105,976.00). The one Planned line, worth $56,850.00, is shown on its own and is not in the totals. You can open my sample page to compare.

What can go wrong?

These came up while building and testing this dashboard.

  1. The duplicate is counted twice. If the script keys on Order and Line only, it either drops releases 2 and 3 of order 48195 or keeps the real duplicate. Rule 1 says Order plus Line plus Release.
  2. Closed lines count as open. Infor keeps Filled and Complete lines on the grid until someone moves them to History. The status lists in the map keep them out.
  3. The "odd things" list is vague. A reply can say only "some dates may be in different formats" and miss the quantities stored as text. That is why you look at the export yourself in Step 1.
  4. Claude adds a rule you did not ask for. For example, it may also set aside rows with a blank or unreadable price. That is a safe choice, but read what it says it added before you keep the script.
  5. A column is renamed in Infor. If someone renames a column the map uses, the script stops and names the missing column. That is on purpose. Fix the map, not the export.
  6. The page is right but the export was not. If the grid filter left out a warehouse, the page will be wrong in a tidy way. Use the same saved filter every week.

How do I check the answer?

Check the page against the export and the answer key, not against Claude's description.

  • The four tiles match the answer key.
  • Every set-aside row has a reason you can see in the export.
  • The check line adds up: rows in the export = open lines + rows set aside.
  • Running the script twice gives the same page.

Why does this way work?

It works because AI does the one fuzzy job once, which is reading Infor's columns, and a script you can check does the counting every week. Here is the short "why" behind each habit.

Map the columns once. Infor exports come out with columns in data order, extra hidden columns, and numbers stored as text. Reading those is the part a person or AI has to think about, so you do it once, write it down in a map with a fixed shape, and check it. That is the same reason you clean the data before AI uses it, and a fixed shape is the idea behind structured output.

Have AI build the tool once. If you asked a chat for the dashboard every Monday, the numbers could drift from week to week, even on the same export. A script gives the same page from the same file every time, which is what deterministic means. You spend AI once on building the tool, and the tool does the repeat work for free.

Describe the page you want, not the code. You know what Carol needs to see on Monday, and Claude knows how to write the code. When you describe the output, not the code, each of you does the part you are good at.

Set bad rows aside, and never guess. A line with no due date is a question for customer service, not a blank to fill in. The script lists it with the row number, so a person sees it, which is how you make bad data stop the script instead of hiding in a total.

Check the first page against an answer key. A dashboard can look tidy and still be wrong. When you already know the right numbers for the sample, a wrong tile shows up in minutes. (Test an AI tool against answers you know.)

Rerun the same script every week. You decided the rules, Claude built the tool, and the tool runs the same way each Monday. That split is humans plan, AI does, and it is why this week's number can be compared with last week's.

Frequently asked questions

Can the dashboard pull from Infor by itself, with no export?

Yes, with help from IT. Infor's API can return the same customer order lines that the form shows, and your script can read that instead of a file. Get your data out of Infor has the request to send IT. Start with the export, because the script's rules stay the same either way. (What is an API? and how AI connects to your systems.)

Our grid has different columns. Does this still work?

Yes. That is what the map is for. Run Step 2 on your own export, and Claude fills the map with your column names. If your company counts Planned lines as open, move Planned into the open list before Step 3.

Can I ask Claude questions about the export instead?

Yes, for one-off questions such as "which customer has the most past due?" (Ask your data in plain words.) Use the script for the numbers you report every week, because a chat answer can change from one run to the next.

Can I make it an Excel file instead of a web page?

Yes. Change "What comes out" in the build prompt to an Excel file with one sheet per table. Check the first one against the answer key the same way.

Does this work with ChatGPT?

Yes, if your plan can run code on files. The prompts are plain text and the script is plain Python, so it runs the same no matter which AI wrote it. Check its first page against the answer key the same way.

Take this with you

Three prompts and a saved pair of files, the script and the map. Together they turn any Monday's Infor export into the same dashboard.

1. Map the columns once. Attach your export.

I attached a Customer Order Lines export from Infor CloudSuite Industrial (SyteLine), made with Actions > To Excel. I want a weekly open-orders dashboard from exports like this one. First, only map the columns. Do not build the dashboard yet.

Read the file. Row 1 may be a note, so find the row that holds the column names. Then make a file called infor-column-map.json in exactly this shape. Use the column names exactly as they are spelled in the export. Use null for any field you cannot find. Never guess.

{
  "header_row": <Excel row number of the column names>,
  "order": "", "line": "", "release": "",
  "customer_name": "", "item": "", "status": "",
  "qty_ordered": "", "qty_shipped": "", "due_date": "", "unit_price": "",
  "status_values_found": [],
  "open_statuses": ["Ordered"],
  "not_firm_statuses": ["Planned"],
  "closed_statuses": ["Filled", "Complete"]
}

Then tell me in plain words, in a short table: which column you picked for each field, and each status value you found with how many rows have it. After the table, list anything odd you saw in the data, such as numbers stored as text, spaces in front of values, dates written different ways, blank cells, or the same line twice. Keep it short.

2. Build the dashboard script. Same chat, after you check the map.

Now write a Python script called infor_dashboard.py that builds the dashboard from any week's export in this shape, using infor-column-map.json. Find every column by its name from the map, never by its position, so the script still works if the columns move.

What goes in: the export file and infor-column-map.json. The export date is in the file name (mrmw-co-lines-YYYY-MM-DD.xlsx), and that date is "today."

Rules:
1. One order line is Order + Line + Release. Take the spaces off the order number. If the same Order + Line + Release is in the file twice, keep the first one and set the second aside.
2. Open qty is Ordered minus Shipped. Some quantities are text, like "40.000". Read them as numbers.
3. A line is open when its status is in open_statuses, its open qty is above 0, and it has a due date. Set aside every other row with the reason. Lines in not_firm_statuses also get their own count and value, "Planned, not firm," which is not in the totals.
4. Due dates can be real Excel dates or text like 11/20/2026 (month/day/year). If a due date is blank or cannot be read, set the line aside. Never guess a date.
5. If a column named in the map is not in the export, stop with a plain message that names the column, and do not make a page.
6. Open value is open qty times Unit Price.

What comes out: one HTML page, open-orders-dashboard-YYYY-MM-DD.html, that opens in a browser with no internet.
- Four number tiles at the top: open lines, open value, past due lines, past due value.
- Open lines and open value by when they are due: Past due (before today), This week (today through 6 days later), Next week (the 7 days after that), and Later.
- Open lines and open value by customer, largest value first.
- The past due lines, oldest due date first.
- Planned, not firm: lines and value.
- Rows set aside: every row not counted as open, with its Excel row number and the reason.
- A check line at the bottom: rows in the export = open lines + rows set aside.

Run the script on the export I attached. Give me the page and the script to download. Then run it a second time and tell me whether the page came out the same.

3. Every Monday. Attach the new export, the script, and the map.

I attached this week's Customer Order Lines export, plus infor_dashboard.py and infor-column-map.json from last time. Run infor_dashboard.py on the new export exactly as it is. Do not change the script or the map.

If the script stops, tell me its message and what it means, and do not work around it. If it runs, give me the new page to download and tell me in four short lines: open lines, open value, past due lines, and past due value, each next to last week's number.

Last week's numbers: [paste the four numbers from last week's page]

4. My sample files: infor_dashboard.py and infor-column-map.json. Your own script will differ, so have Claude write yours from the prompts.

What's next?

Get your first win this week.

One minute. Three guides picked for your job.

Start my first win