- Find and flag duplicate values →
Mark every repeated registration, keep blank rows blank, and see which values need a closer look.
- Calculate percentage change from a positive base →
Compare old and new values, distinguish growth from decline, and handle zero or negative starting values explicitly.
- Convert imported text into numbers →
Convert amounts with an explicit decimal convention, keep missing values blank, and flag text that cannot be parsed.
- Weight a workload by its case counts →
Combine per-case estimates using the number of cases instead of giving every service the same influence.
- Classify deliveries against an SLA →
Separate unfinished deliveries, missing measurements and genuine zero-hour observations before deciding whether a completed delivery is late.
- Build an operational review with the right denominator →
Compare two legitimate averages for one selected population and base the review flag on the agreed case-weighted metric.
- Choose a rate from quantity bands →
Find the rate attached to the largest threshold at or below each quantity, with explicit rules for missing bands and negative inputs.
- Build a filtered priority queue with a stable tie rule →
Select one team at or above a points threshold, then sort the complete records by points descending and numeric ticket ascending.
- Return one numeric review mean for every record →
Use BYROW to reduce each three-review row, then keep the record labels beside the results with HSTACK.
- Summarise a numeric sample without hiding missing data →
Keep the numeric count, median and sample standard deviation together; distinguish a missing sample from a measured zero.
- Compare a chosen percentile with an allowance →
Calculate an inclusive percentile, count observations strictly above it and apply an explicit allowance rule.
- Flag observations outside the IQR fences →
Calculate inclusive quartiles and two 1.5-IQR fences, then count measured values strictly outside them.
- Split a pasted log into rows and fields →
Turn a small literal pipe-delimited log into a rectangular report while keeping empty records, blank fields and leading zeroes.
- Extract both sides of the final separator →
Use the last literal separator to separate a record prefix from its final field without losing earlier separators or repeated records.
- Normalize and review structured reference codes →
Create a three-column audit that validates a declared ASCII code shape before extracting a four-character suffix.
- Reconcile grouped totals from two exports →
Keep the union of both ID lists, combine signed repeated rows on each side, then report both totals and their difference.
- Group posted net units by region →
Filter order status before summing repeated regions, including returns and numeric zeroes.
- Cross-tabulate posted order means →
Build a region-by-channel mean report whose row, column and grand means use the original records.
- Report each group share of net units →
Make the grand-total denominator explicit, retaining signed returns, missing intersections and a zero-net state.
- Convert fixed-width local date and time text →
Build a real numeric timestamp from explicit components, then reject values that normalize to a different date or clock time.
- Calculate hours for a shift across midnight →
Use one-day clock wrap, whole-minute arithmetic and a declared break rule to calculate numeric paid hours.
- Count dates shared by two inclusive intervals →
Clip each original interval to a reporting window, count both endpoints and preserve a distinct review result for reversed records.
- Flatten numeric readings in a declared scan order →
Turn a 4 by 3 readings grid into one column while keeping numeric zeroes and excluding text notes and booleans.
- Return the last N active records →
Skip empty-ID slots, reorder the fields and take the final requested records without sorting or reversing them.
- Accumulate posted charges after rounding each line →
Use a declared zero initial state, select posted records and add individually rounded signed charges.
- Classify real errors without misreading text →
Return a category and classic error code for each imported value while preserving source order.
- Replace only missing-value errors →
Use one literal fallback for real #N/A while keeping other errors visible and preserving ordinary value types.
- Summarize measurements that contain errors →
Report a signed total, numeric count, numeric mean and genuine error count without converting other value types to zero.
- Evaluate dated cash flows using actual day intervals →
Keep date-value pairs aligned and use 365-day discounting for an irregular cash-flow schedule.
- Compare fixed and accelerated declining balances →
Keep DB rate rounding, first-year months, DDB factors and the final fiscal stub explicit.
- Compare exact, cumulative and interval event counts →
Build a complete binomial report for a fixed number of independent trials with one constant event probability.
- Compare finite sampling with a replacement model →
Calculate a sample-count report without replacement, then compare it with independent draws at the same starting proportion.
- Scale an event-count model to its observation window →
Convert a rate per base interval to a window mean, then report exact, cumulative, interval and upper-tail event probabilities.
- Refactor a nested IF policy into a maintained rule table →
Compare a frozen multi-branch policy with a current editable lower-bound table, then identify every request whose selected flat rate changes.
- Rebuild dates and measure complete calendar intervals →
Extract and rebuild real date components, compare DATE overflow with month-end clamping, then report complete years/months and actual elapsed days.
- Compare OFFSET and INDEX dynamic ranges →
Use one row-count input to resize two real reference methods, then compare their totals, means and selected populations.
- Choose wildcard searches or literal product codes →
Compare intentional pattern matching with literal identifiers, returning the first physical source record and preserving numeric zero.
- Count numbers, text, logical values and blank-like cells →
Reconcile four disjoint stored types and explain why COUNTA plus COUNTBLANK can exceed the number of cells.
- Compare mean populations without changing stored types →
Keep numeric zeroes, excluded text and explicit mean denominators visible before choosing an averaging rule.
- Build and inspect line-separated address labels →
Use actual LF characters, compare compact and fixed-slot joins, then inspect an ASCII marker without losing the original text.
- Resolve overlapping request rules with explicit priority →
Compare nested IF and first-match IFS on the same requests. Combine AND, OR and NOT, expose overlapping rules and change the priority in a real worksheet cell.
