Build a workflowWorkflow

Turn a spreadsheet rule into a formula you can check

Build a Google Sheets stock flag with an eight-case answer key, then test blanks, zero, the exact boundary and invalid inputs.

Illustrative resourceNot tool-testedSpreadsheet beginners who need a stock-threshold formula they can inspect and test

What you’re working towards

A proposed single-cell Google Sheets formula, a pasteable eight-case synthetic grid, an independent answer key and a procedure for checking actual results and recalculation. The downloadable pack is documentation-reviewed; Google Sheets execution and the AI workflow remain unperformed.

Download playbook .md

Your choice also saves this guide’s current worked-example checkmarks. Only this browser; no account or sync. View saved guides

Saving needs JavaScript. Reading and downloading this guide does not save it.

Included example · illustrative MD file

Download the formula check pack

Owned Markdown guide with a tab-separated synthetic test grid, explicit Google Sheets formulas, eight expected outcomes and reversible test instructions. Spreadsheet app execution and the AI workflow are unperformed.

Not tool-tested. Review and adapt before using it for a real project. Your browser may download this file when you choose View example file.

View example file Download example

Before you begin

A disposable Google Sheets spreadsheet you can edit in the computer/web interface; United States locale, English function names and automatic calculation for testing
The explicit rule: individual gift bags, a strict below-five threshold, whole non-negative counts only, and separate MISSING and CHECK INPUT statuses
The supplied synthetic test grid and an untouched expected-results column; no real customer or stock files are needed
A text AI assistant if you want help explaining or adapting the rule; its account tier and limits must be checked separately

Possible tools

Google SheetsComputer/web app · required
Access and plan limits unverified
An existing editable spreadsheet is needed. Account-tier eligibility, quotas and free-plan access have not been checked. Target United States locale and English function names. Documentation supports component functions, not execution of this combined formula.
A text AI assistantReader-selected text chat · suggested
Access and plan limits unverified
Use an assistant already available to the reader. No provider capability, pricing, upload limit or account tier has been verified. Returning formula text is not app execution.

Use the sample first. Remove private information before taking a prompt into another tool.

Make a start

One step at a time.

5 steps

Start with the synthetic worked example or other non-sensitive inputs. Run one prompt at a time in your chosen AI tool. Replace any bracketed placeholders with the requested material, check each reply, and supply the source or earlier outputs that step needs. Paste them again if the tool lacks that context.

Set up the worked example

Open: Open a new text chat and the formula-check-pack.md download in a text editor or Markdown reader. For the actual app test, use a disposable Google Sheets spreadsheet in the computer/web interface that you may edit. The exercise targets United States locale, English function names, comma-separated arguments and period decimals. Do not change a production or shared sheet. If this app/setup is unavailable, you can inspect the text but must leave the spreadsheet test not run.
Add the sample: In step 1, replace [paste the input above] with all of “The starting point” (Synthetic input): TARGET, APPROVED RULE, the complete PASTE INTO A1 grid and the final clear-cell instructions. For step 2, replace [paste] with that same complete input. For step 3, replace [paste] under APPROVED RULE with the full approved-rule paragraph, not the answer key. In step 4, paste the actual formulas from your sheet, original rule/grid and only results or errors you observed into the two labeled sections. In step 5, replace [paste] with your real observations, including not-run tests. Use “The intended result” (Illustrative output) and the pack as expected answers, never as an observed log. Keep the chat open between steps.

Start with the synthetic example, then adapt each prompt to your task.

Write down what every kind of input means

Agree the strict boundary and the meaning of missing or invalid values before asking for a formula. In this fixture, a blank is unknown, zero is known stock and exactly five is OK.

Your starting prompt

Turn this stock rule into a decision table before writing a formula. Include a genuinely empty cell, zero, exactly five, ordinary whole numbers below and above five, text, a negative number and a fraction. Preserve the strict less-than boundary. Do not convert text, round a count or fill blanks. Flag missing rules. Target Google Sheets on the web, United States locale and English function names. Do not claim execution.

RULE AND SYNTHETIC GRID:
[paste the input above]

Expect: A decision table before a formula: blank, zero, the exact boundary, ordinary counts, text, negative and fractional inputs.
Check: A truly empty cell means MISSING; valid whole counts 0–4 mean RESTOCK; 5 and above mean OK. Text, including numeric text, negative numbers and fractions mean CHECK INPUT. Exactly 5 does not trigger restocking. These are separate scenarios, not stock to add together.
If it differs: If the rule is relaxed, send: “Restore the approved strict below-five rule. Keep genuinely empty, numeric zero, numeric text, negative and fractional inputs distinct. Do not coerce text or round counts. Return the corrected decision table before formula work.”

Paste the eight-case grid into a disposable sheet

Download the pack. In a new Google Sheets file, use the specified locale and English function names. Paste the tab-separated grid into A1. Verify six columns and nine rows including the header; clear B2 completely. Keep C and E blank and preserve the expected statuses in D.

Your starting prompt

Review the test grid against the rule. State the exact A:F column mapping and eight expected statuses. Explain how to distinguish a truly empty B2 from zero or a text value. Identify any paste/setup checks I must do in Google Sheets. Do not claim you inspected or edited the sheet.

RULE AND GRID:
[paste]

Expect: An explanation of A:F: case_id, quantity, actual_status, expected_status, check, reason. In your disposable file, inspect File → Settings for United States locale and English function names. Keep automatic calculation if that setting is shown. Copy only the pack’s tab-separated grid into A1, clicking once rather than editing inside the cell.
Check: Check A1:F9: six columns and eight cases T01–T08. Clear B2 completely with the clear/delete contents action; typing a space or the word blank is not empty. Leave C2:C9/E2:E9 empty and D unchanged. B3/B4/B5/B6/B8/B9 should hold numeric 0/5/3/12/−1/2.5; B7 holds text pending. The pack’s spare-cell ISBLANK/ISNUMBER probes let you check types before proceeding.
If it differs: If columns shift or everything lands in one cell, undo and re-enter the six columns from the pack without dropping empty fields. If a numeric case was pasted as text, retype that original numeric value; do not convert arbitrary future text inputs. Send: “Explain this exact setup mismatch: [cell and observed value/type]. Keep the original rule and expected column unchanged.” If setup cannot match, stop with the app test not run.

Get one formula and inspect its references

Ask for a single-cell formula for C2. Compare it with the supplied version, paste it and fill down to C9. Inspect C3 and C9 to ensure their references point to B3 and B9; the status formula must not read the answer key.

Your starting prompt

Write a single-cell Google Sheets formula for C2 reading B2, to fill down through C9. Use English function names, comma separators and the literal threshold 5. A truly empty cell returns MISSING; text, negative values and fractions return CHECK INPUT; valid whole counts below 5 return RESTOCK, otherwise OK. Do not coerce text, round values, use scripts or add external references. Explain the check order, return the exact formula separately, and list tests to run. Do not claim execution.

APPROVED RULE:
[paste]

Expect: One formula for C2 using B2, with the empty/type/whole-number checks before the strict threshold. Copy only the formula, including its leading =, into C2. Fill down through C9 using the cell’s fill handle.
Check: Inspect the formula bar at C2, C3, C5 and C9: each must read its matching B-row. It must use <5, not <=5, and must never read column D. Compare with the pack’s proposed formula. Expected C2:C9 labels are MISSING, RESTOCK, OK, RESTOCK, OK, CHECK INPUT, CHECK INPUT, CHECK INPUT. Seeing a formula in a document is not executing it.
If it differs: For a parse error or wrong label, send: “Review this exact formula-bar text and error against the approved rule and target locale: [paste formula and observed error]. Explain the smallest correction without changing the rule, answer key or numeric/text policy.” Paste a corrected formula only after understanding it, refill and inspect row references again. Do not install scripts or add external references.

Compare every result and make one test fail

Enter the supplied EXACT comparison in E2 and fill down. Check all eight statuses, then change B4 from 5 to 4 while D4 stays OK: expect RESTOCK and FAIL. Restore 5 and run the other three reversible checks in the pack. Eight PASS rows are expected only after you actually observe them.

Your starting prompt

Audit my observed Google Sheets results against the supplied expected grid. Report each case ID, relevant cell, expected value and observed value. Separate observations from hypotheses. Check blanks, numeric/text types, the strict threshold, formula references and whether deliberate changes caused a FAIL. Do not edit the answer key to hide a mismatch. Give the smallest correction and exact cases to rerun.

RULE, FORMULAS AND EXPECTED GRID:
[paste]

OBSERVED RESULTS AND ERROR TEXT:
[paste the synthetic results actually observed]

Expect: A comparison of your actual observations with the answer key, with errors kept distinct from guesses about their cause. First enter =IF(EXACT(C2,D2),"PASS","FAIL") in E2 and fill to E9. EXACT checks letters and spaces in the two labels.
Check: Observe every C/E row; baseline should show eight PASS labels, not just zero failures. Perform these changes separately, restoring each: B4 5→4 gives RESTOCK/FAIL; B3 0→pending gives CHECK INPUT/FAIL; B2 empty→0 gives RESTOCK/FAIL. Finally D6 OK→WRONG gives E6 FAIL while C6 stays OK; restore D6. Recheck baseline. Write down actual results and exact errors; do not borrow the pack’s expected results as observations.
If it differs: If a deliberate change still shows PASS or a status fails, send the exact cell, formula-bar text, original expected label and actual result with step 4’s prompt. Ask: “Check whether these are formulas or fixed text, whether references move by row and whether recalculation occurred. Do not edit the key to hide the mismatch.” Apply a reviewed minimal correction, repeat every affected test and restore/recheck all eight rows.

Keep a checked copy before adapting the rule

Record the actual app, locale, date, formula text and test results. Resolve missing or invalid counts with their owner. Another app, threshold, unit or input policy requires its own revised answer key and app checks before use on real stock.

Your starting prompt

Make a short test record from only the observations below: app/interface, locale, function language, date, exact formulas, baseline results, deliberate changes, failures, corrections and unresolved checks. Label any unperformed test as not run. Separate spreadsheet execution from any text-AI drafting session. List what must be retested before changing the threshold or moving this to real stock data.

OBSERVATIONS:
[paste]

Expect: A test record containing only observed app/interface, locale, language, date, exact formulas, baseline results, four deliberate changes, repairs and unrun checks. Keep the disposable workbook under a distinct test name and save a separate text record. Reopen the sheet and confirm the restored baseline remains.
Check: No app test is passed without actual observations there. A text-AI response, documentation review and local fixture model are separate activities. The formula cannot prove that a numeric date/percentage represents a bag count, and RESTOCK is only a review flag, not an order quantity or purchasing action.
If it differs: If the record calls an unperformed test passed, send: “Replace unsupported pass claims with not run. Preserve every actual failure and correction. List the rule, answer-key, type, reference and recalculation tests needed after any change of app, locale, threshold, unit or input policy.” Resolve missing/invalid stock with its owner separately; do not copy this test into live stock as production-ready.

Optional: improve the result

Add type and recalculation tests before generalizing

In a copy, add numeric text, an empty-string formula, a space and a logical-value probe. All should be CHECK INPUT under this rule. If moving the threshold to a fixed cell, define and test invalid threshold behavior before changing the baseline formula.

Advanced prompt

Extend the stock-rule test plan with numeric text, an empty string, a space and a logical value while retaining the eight baseline cases. Explain each expected result. Propose a separate revision that uses a fixed threshold cell with a non-negative whole-number requirement, explicit behavior for blank/text/invalid thresholds and a fixed reference. List every formula and expected result that would change. Propose fill-down and recalculation checks for the first, middle and last rows. Keep unexecuted cases marked untested; do not silently change the approved rule.

CURRENT RULE, FORMULA AND TEST GRID:
[paste]

Optional: use your own input

Check a formula for your own simple numeric rule

Start a new text chat with a synthetic rule and no private stock file. State your actual spreadsheet app/interface, locale and function language; do not assume the sample setup applies. Replace all MY INPUT fields together, including threshold, equality rule, units, missing/invalid input behavior and output labels. Keep this to a simple row-by-row numeric review flag. After checking the decision table, send:
“Create a small synthetic test grid for MY INPUT, with stable case IDs, input, blank actual/check columns and an independently explained expected column. Cover my boundary, ordinary values, missing values and invalid types according to my rule. Specify the actual target cells and paste/type checks for my app. Stop if syntax or input behavior is unconfirmed.”
Prepare a disposable sheet you may edit; preserve the expected column. Check its actual columns, rows and input types, then send:
“Using my confirmed grid layout, draft one status formula for the first data row and a separate exact-label checker. Explain the checks and relative/fixed references in my app’s syntax. No scripts or external references. Do not claim execution.”
Enter and fill formulas only in that test copy, checking first/middle/last references. Send:
“Give reversible recalculation tests using MY INPUT’s actual boundary, invalid-input rules and labels, plus a deliberate wrong expected label that must fail the checker. State what to restore after each test; do not fill in my observations.”
Perform the tests and then send the original rule/grid, actual formula-bar text and your observed results/errors with:
“Audit these observations, distinguish hypotheses, suggest minimal fixes and give the exact rerun cases. Preserve the original key. Produce a test record with not-run checks and limits.”
Save the test copy and record, reopen and recheck restored baseline. If app access or confirmed syntax is unavailable, keep the written proposal and leave app tests not run. These personal checks replace the worked-example checks. The original advanced extension is sample-only unless separately adapted to your rule and expected results.

Your own-input prompt

Help me turn MY INPUT into a checkable spreadsheet formula proposal. First build a decision table, not a formula. Extract the exact unit, threshold, strict or inclusive comparison, output labels and rules for genuinely empty cells, zero, text/numeric text, negative values, fractions and other supplied cases. Ask about missing or contradictory rules instead of adopting the guide’s rule. Do not silently coerce, round or repair inputs. Distinguish stored values from formatting; a numeric date or percentage is not necessarily a valid count. Ask about unclear target app/interface, locale, function language and layout before giving app-specific instructions. Stop for my review. Later we will make a synthetic answer key, draft a formula, run manual tests in my disposable copy and record actual observations. Do not edit a live sheet, access accounts, install anything, order stock or claim app execution from a text answer.

MY INPUT: RULE
[State the item/unit, threshold, whether equality triggers the flag, exact output labels, and desired behavior for each missing/invalid input type. Mark undecided rules unknown; do not copy sample values just to fill this section.]

MY INPUT: TARGET AND TEST LIMITS
[State the spreadsheet app, interface, locale, function language and any known cell layout. Use small synthetic values and a disposable test sheet. Say unknown or unavailable where appropriate. List any additional cases you need checked.]

Check your own result

Your decision table, formula and independently explained answer key agree on your actual unit, boundary, equality behavior, labels and input policy; missing rules are resolved before testing.
The disposable grid has your actual case count and cell layout. Input types are checked, status formulas read the matching input cells, and no status formula reads the expected column.
You personally observe baseline results, a boundary change, applicable invalid/missing cases and a deliberate checker failure, restoring and rerunning the baseline afterward.
Your saved record distinguishes observed spreadsheet behavior, text-AI drafting and unperformed tests. A new app, locale, rule or input meaning needs revised expectations and fresh tests; a passed small fixture is not production certification.

Show, then try

See the shape of the result.

This input is synthetic. The output is a written illustration of the intended result, not a verified AI-tool run.

The starting point

Synthetic input

TARGET
Google Sheets on the web, computer interface. United States locale, English function names, comma separators, period decimal.

APPROVED RULE
Fictional stock item: paper gift bags, counted individually. Restock only when a valid count is strictly below 5. A valid count is a directly entered, non-negative whole number. A truly empty quantity cell returns MISSING. Text, including a number stored as text, negative numbers and fractions return CHECK INPUT. Valid counts 0 through 4 return RESTOCK; 5 or more return OK. Do not replace unknowns with zero, round fractions or convert text automatically. The threshold is the literal number 5 in this pilot. Each row is a separate test scenario, not an item to sum.

PASTE INTO A1
case_id	quantity	actual_status	expected_status	check	reason
T01			MISSING		No count supplied
T02	0		RESTOCK		Zero is a known count below five
T03	5		OK		Exactly five does not trigger restocking
T04	3		RESTOCK		Ordinary whole count below five
T05	12		OK		Ordinary whole count above five
T06	pending		CHECK INPUT		Text is not a count
T07	-1		CHECK INPUT		Negative counts are invalid
T08	2.5		CHECK INPUT		Fractional counts are invalid

Clear B2 after pasting so it is truly empty. Keep C2:C9 and E2:E9 blank until you enter the formulas. These eight rows are scenarios, not a total stock list.

Copies text only. For image or file tasks, use the source files named in the steps too.

Download source .txt

The intended result

Illustrative output

ILLUSTRATIVE FORMULA AND EXPECTED RESULTS; NOT CAPTURED APP OUTPUT
C2, then fill down to C9:
=IF(ISBLANK(B2),"MISSING",IF(ISNUMBER(B2),IF(OR(B2<0,MOD(B2,1)<>0),"CHECK INPUT",IF(B2<5,"RESTOCK","OK")),"CHECK INPUT"))

E2, then fill down to E9:
=IF(EXACT(C2,D2),"PASS","FAIL")

EXPECTED STATUS BY CASE
T01 | quantity=[truly empty] | MISSING
T02 | quantity=0 | RESTOCK
T03 | quantity=5 | OK
T04 | quantity=3 | RESTOCK
T05 | quantity=12 | OK
T06 | quantity=pending | CHECK INPUT
T07 | quantity=-1 | CHECK INPUT
T08 | quantity=2.5 | CHECK INPUT

After running the formulas in the named app, all eight checks should be PASS. Optional H2 =COUNTIF(E2:E9,"PASS") should show 8; I2 =COUNTIF(E2:E9,"FAIL") should show 0. These values are an answer key, not observed execution.

RECALCULATION CHECKS TO PERFORM
Change B4 from 5 to 4: C4 should become RESTOCK and E4 FAIL while D4 stays OK. Restore 5. Change B3 from 0 to pending: expect CHECK INPUT and FAIL; restore 0. Enter 0 in B2: expect RESTOCK and FAIL; clear B2 to restore MISSING and PASS. Temporarily change D6 to WRONG: C6 stays OK and E6 becomes FAIL; restore D6 to OK. Restore the baseline and check all eight rows again.

CURRENT LIMITS
No Google Sheets formula execution, paste/fill-down observation or AI-workflow run has been recorded. Local fixture checks do not establish app compatibility. No Excel, mobile, alternate-locale, free-tier or time-saving claim is made.

Download result .txt

You’re the final check

Check the worked example.

These checks use the worked example and its particular facts. With your own material, check your actual names, numbers, sources and rules; do not reuse the sample’s values as the answer.

The grid contains T01 through T08 exactly once. B2 is truly empty, B3 is numeric 0 and B4 is numeric 5; C and E contain formulas after setup.Expected results are T01 MISSING; T02 RESTOCK; T03 OK; T04 RESTOCK; T05 OK; T06, T07 and T08 CHECK INPUT. Check every row in Google Sheets.The status formula uses <5, reads the matching B-row and never reads the expected-status column D. Inspect the first, middle and last filled rows.With the baseline unchanged, all eight exact-match checks should be PASS. A zero FAIL count alone cannot establish success if formulas or results are missing.Changing B4 from 5 to 4 must produce RESTOCK and FAIL against unchanged D4; restoring it must recover OK and PASS. Perform and restore the other input/checker perturbations in the pack.Record actual spreadsheet observations separately from the authored answer key, documentation checks and local fixture checks. Do not claim app compatibility or an AI-workflow test without execution evidence.

Unsaved checks stay in this tab and reset when you reload. If you choose Save guide, its current example checkmarks are kept in this browser when storage is available.

A few things to watch for.

Using <=5 when the approved rule is strictly below five.
Replacing an unknown count with zero, accepting numeric text automatically or rounding fractions without permission.
Treating a space or a formula returning an empty string as a truly empty cell.
Pasting fixed statuses instead of formulas or filling every row with a reference to B2.
Assuming ISNUMBER proves the value is a stock count; dates, percentages and other numeric meanings require source checks.
Treating a documentation review, an unchanged PASS label or local arithmetic as Google Sheets execution.

What has been checked?

Documentation: Reviewed 2026-10-14. Source review is not workflow execution.
Included files: Scoped checks passed on 2026-10-14. File parsing; Fixture checks. See scope below.
AI workflow: Not run in a recorded consumer-AI workflow.
Time savings: Time savings not measured

Read the scope and evidence

formula-check-pack.md

2026-10-14 · parser · passed

Markdown/JSON/TSV structure and narrow formula grammar only; no spreadsheet app execution.

UTF-8 Markdown and JSON parsed; code fences paired
A1:F9 TSV shape, eight unique IDs and intentionally empty B2/C/E fields
Four formula strings parsed with a narrow local grammar and correct known arities
Formula references and strict <5 boundary inspected
Fixture and formula text match record input/output
Artifact SHA-256 recomputed

Environment: Python 3.12.14; standalone validator, no Google Sheets session.

SHA-256: 492c73b43e85f74b69d1500a6d41088db182ffba0220b2037c3b3474f10bcf58

formula-check-pack.md

2026-10-14 · semantic fixture · passed

Authored expected results compared with an independent Python rule and a narrow local formula model. This is not a spreadsheet engine, app run or AI-workflow run.

Eight baseline answer-key cases matched an independent Python rule and a narrow local formula model
Four additional text/empty-string/space/logical-value cases matched the local model
Four deliberate mismatch cases and their restorations checked in the local model
Small integer range -10 through 100 and six fractional probes checked in the local model

Environment: Python 3.12.14; no spreadsheet library or native app invoked.

SHA-256: 492c73b43e85f74b69d1500a6d41088db182ffba0220b2037c3b3474f10bcf58

IF function ↗ (opens in a new tab)

2026-10-14 · confirmed

IF(logical_expression, value_if_true, value_if_false) chooses between two outputs based on a logical test.