22 Manus Prompts for Data Analysis & Spreadsheets (2026)

Goal: Profile the attached dataset [file name] and give me a plain-language summary of what's in it.
Constraints: Analyze every column. For the timeframe [start date] to [end date], focus on [key metric columns] and segment by [segment column, e.g. region]. Run code on the actual file — do not estimate.
Deliverable: A one-page findings report with these sections — (1) dataset overview (row count, column list, data types, date range), (2) per-column profile (min/max/mean/median, % missing, distinct values), (3) top 5 things worth my attention.
Audit rules: Show a short methodology note listing every assumption and filter applied. Flag any data-quality issues (missing values, duplicates, outliers, inconsistent formats). Do not guess missing values — report them as gaps. Use Manus 1.6 Max.

Goal: Find the trends, spikes, and outliers in the attached dataset [file name].
Constraints: Analyze [metric column] over [date column] for the period [timeframe], broken down by [segment column]. Define an outlier as any value more than 2 standard deviations from the segment mean, and a trend as a sustained move over [3+] consecutive periods.
Deliverable: A short report with (1) a table of detected trends (segment, direction, % change, period), (2) a table of outliers (row, value, expected range, likely reason if evident from the data), (3) a 3-line takeaway.
Audit rules: State the exact method and thresholds used in a methodology note. Tag any suspected cause as Low confidence unless the data proves it. Flag columns too sparse to analyze rather than forcing a result. Use Manus 1.6 Max.

Goal: Calculate my core KPIs from the attached dataset [file name] and write them into a findings report.
Constraints: Compute [list KPIs, e.g. total revenue, average order value, conversion rate, churn] for [timeframe], with a period-over-period comparison against [prior period]. Segment KPIs by [segment column].
Deliverable: A findings report with (1) a KPI summary table (metric, current, prior, % change, direction arrow), (2) a 2-sentence read on each KPI, (3) the three KPIs most off-target.
Audit rules: Show the formula used for each KPI in a methodology note. Compute all figures from the supplied data only; if a KPI can't be derived, mark it "not computable" and say why. Leave no cell as a guess. Use Manus 1.6 Max.

Goal: Break the attached dataset [file name] down by segment and tell me which segments over- and under-perform.
Constraints: Group by [segment column, e.g. plan tier / channel / region] and compute [metrics, e.g. count, revenue, average value, retention] for [timeframe]. Rank segments by [primary metric].
Deliverable: A comparison table (one row per segment, columns for each metric plus rank and share of total) and a short narrative naming the top 2 and bottom 2 segments with the likely driver from the data.
Audit rules: Note any segment with a sample size below [n] as low-reliability rather than ranking it confidently. List assumptions and any excluded rows in a methodology note. Do not fabricate segments not present in the data. Use Manus 1.6 Max.

Goal: Tell me what variables in the attached dataset [file name] move together with [target metric].
Constraints: Test the relationship between [target metric] and [candidate columns] over [timeframe]. Report correlation coefficients and note direction and strength. Do not claim causation.
Deliverable: A ranked table (variable, correlation with target, direction, strength label) plus a 3-line summary of the strongest signals and any that surprise.
Audit rules: State the correlation method and sample size in a methodology note. Tag every relationship as correlation only, not cause, and mark weak or noisy relationships as Low confidence. Flag columns dropped for being too sparse. Use Manus 1.6 Max.

Goal: Clean the attached spreadsheet [file name] and return a tidy version.
Constraints: Deduplicate rows using [key columns, e.g. email + name] as the match key. Normalize [columns, e.g. dates to YYYY-MM-DD, phone numbers to E.164, text to Title Case, currency to USD]. Trim whitespace and standardize inconsistent category labels in [column].
Deliverable: A cleaned spreadsheet with the same columns plus a "change_notes" column, and a separate summary tab: rows in, rows out, duplicates removed, values normalized per column.
Audit rules: Never merge two rows unless the match key is identical — flag near-matches for my review instead. Do not invent values to fill blanks. Log every transformation rule applied in a methodology note. Use Manus 1.6 Max.

Goal: Handle the missing values in the attached dataset [file name].
Constraints: For each column with gaps, use the rule I specify: [column A] fill with the segment median, [column B] fill with "Unknown", [column C] leave blank and flag. Never impute [critical column, e.g. revenue].
Deliverable: The dataset with a "was_imputed" flag column marking every filled cell, plus a summary tab: % missing before and after per column, and the method used for each.
Audit rules: Only impute where I explicitly allowed it; everywhere else leave the cell blank. Tag every imputed value as Medium confidence at most. State each fill rule in a methodology note so I can reverse it. Use Manus 1.6 Max.

Goal: Merge the two attached files [file A] and [file B] into one enriched dataset.
Constraints: Join on [key column, e.g. customer_id] using a [left/inner] join. From [file B] bring in [columns to add]. If a key appears more than once in either file, resolve by [rule, e.g. keep most recent by date].
Deliverable: A single merged spreadsheet with all requested columns, plus a summary tab reporting matched rows, unmatched rows from each side, and any duplicate keys resolved.
Audit rules: List every unmatched key rather than silently dropping it. Do not fabricate values for rows with no match — leave the added columns blank and flag them. Document the join type and dedupe rule in a methodology note. Use Manus 1.6 Max.

Goal: Add a category and tags to every row in the attached dataset [file name].
Constraints: Read [text column, e.g. product description / support ticket] and assign one category from this fixed list: [category 1, category 2, category 3, Other]. Also add up to [3] tags from [tag column or free tags]. Do not create categories outside my list.
Deliverable: The dataset with new "category" and "tags" columns, plus a summary tab showing the count and % per category.
Audit rules: When a row is ambiguous, assign "Other" and flag it rather than forcing a category. Tag any low-certainty classification as Low confidence in a "confidence" column. Explain the rules used in a methodology note. Use Manus 1.6 Max.

Goal: Build a KPI dashboard from the attached dataset [file name].
Constraints: Compute [KPIs, e.g. revenue, active users, conversion rate, churn, AOV] for [timeframe], each with a period-over-period delta and a spark trend. Segment the top row of tiles by [segment column].
Deliverable: A dashboard with (1) a row of KPI tiles (value + % change + up/down arrow), (2) a trend line per KPI, (3) a segment breakdown table beneath, and a hidden "data" tab holding every calculated figure.
Audit rules: Show the formula behind each tile in a methodology note. Compute all values from the file only; mark any KPI it can't derive as "n/a" with a reason. No placeholder numbers. Use Manus 1.6 Max.

Goal: Chart the trends in the attached dataset [file name] over time.
Constraints: Plot [metric columns] over [date column] at [daily/weekly/monthly] granularity for [timeframe], one series per [segment]. Add a 7-period moving average where it aids readability.
Deliverable: A line chart per metric (segments as separate series, clear legend and axis labels), plus a small table of start value, end value, and % change per series.
Audit rules: State the granularity, rollup method, and any smoothing in a methodology note. If periods are missing from the data, show gaps rather than interpolating. Flag series with too few points to trend. Use Manus 1.6 Max.

Goal: Build a cohort retention grid from the attached dataset [file name].
Constraints: Group users by [signup/first-purchase date column] into [weekly/monthly] cohorts. Measure retention as the % of each cohort active in each subsequent period, based on [activity column]. Cover [N] periods.
Deliverable: A cohort table (rows = cohort, columns = period 0..N, cells = retention %) with conditional shading, plus a short read on which cohorts retain best and where the drop-off is steepest.
Audit rules: Define "active" and the cohort logic explicitly in a methodology note. Show cohort sizes; mark cohorts below [n] users as low-reliability. Do not extrapolate periods that haven't fully elapsed — leave them blank. Use Manus 1.6 Max.

Goal: Build a conversion funnel from the attached event dataset [file name].
Constraints: Order these steps [step 1 → step 2 → step 3 → step 4] using [event/stage column], counting unique [user/session id] per step for [timeframe]. Segment the funnel by [segment column] if provided.
Deliverable: A funnel chart plus a table (step, count, step conversion %, overall conversion %, biggest drop-off flag), and a 2-line summary of where users leak most.
Audit rules: State how a user is counted per step and how out-of-order events are handled in a methodology note. Do not infer steps that aren't in the data. Tag any assumption about step order as an assumption. Use Manus 1.6 Max.

Goal: Visualize how [items, e.g. products / channels / regions] compare in the attached dataset [file name].
Constraints: Rank [dimension column] by [metric] for [timeframe]. Show the top [10] and bottom [5]. Add a second metric [secondary metric] as a comparison.
Deliverable: A horizontal bar chart ranked by the primary metric, a paired chart or column for the secondary metric, and a ranking table (item, metric 1, metric 2, rank, share of total).
Audit rules: Note the ranking metric, timeframe, and any excluded items in a methodology note. Compute shares from the file total, not an assumed total. Flag items with sparse data rather than ranking them confidently. Use Manus 1.6 Max.

Goal: Build a [12/36]-month financial model spreadsheet for [business/product].
Constraints: Drivers are [starting customers], [monthly growth rate], [price], [churn rate], [gross margin %], and [fixed monthly costs] — put all of these on an editable "Assumptions" tab. Model monthly across [timeframe].
Deliverable: A spreadsheet with tabs (1) Assumptions, (2) Monthly P&L (revenue, COGS, gross profit, fixed costs, EBITDA per month with live formulas referencing Assumptions), (3) Summary (annual totals + key ratios), (4) a simple chart of revenue vs. costs.
Audit rules: Use real cell formulas so I can change any assumption and watch it flow through — no hard-coded results. List every assumption and its source or "user-provided" in a methodology note. Flag any driver left blank rather than guessing it. Use Manus 1.6 Max.

Goal: Build a monthly budget tracker spreadsheet for [personal / team / project] use.
Constraints: Categories are [list categories]. I want a budgeted amount and an actual amount per category per month for [year]. Currency [USD].
Deliverable: A spreadsheet with (1) a monthly grid (rows = category, columns = Budget, Actual, Variance, Variance % — all live formulas), (2) a totals row, (3) a summary tab with total budget vs. actual and a variance chart, (4) conditional formatting that turns over-budget cells red.
Audit rules: Leave Actual cells blank for me to fill — do not invent spending. Use formulas for every Variance and total so edits recalculate. Document the sheet structure in a short notes tab. Use Manus 1.6.

Goal: Build a [project task / habit / content] tracker spreadsheet.
Constraints: Each row is a [task/habit/item] with columns [Name, Owner, Status, Priority, Start date, Due date, % complete, Notes]. Status options are [Not started, In progress, Blocked, Done].
Deliverable: A spreadsheet with (1) the tracker grid with a dropdown-style Status column, (2) auto-calculated columns for days remaining and an overdue flag, (3) a summary tab counting items by status and a simple progress bar or chart.
Audit rules: Pre-fill only structure and example rows clearly marked "EXAMPLE" — no fake real data. Use formulas for days-remaining and overdue so they stay live. Explain each column in a notes tab. Use Manus 1.6.

Goal: Build a weighted comparison spreadsheet to help me choose between [options, e.g. tools / vendors / plans].
Constraints: Options are [list options]. Criteria and weights are [criterion 1 = 40%, criterion 2 = 30%, criterion 3 = 20%, criterion 4 = 10%] — weights on an editable tab. Score each criterion 1–5.
Deliverable: A matrix (rows = options, columns = each criterion score, weighted score, and total) with a live weighted-total formula, plus a ranked summary and a bar chart of totals. Highlight the winner.
Audit rules: Leave score cells for me to fill, or fill only from data I provide and cite it. Recompute totals via formula when weights or scores change. If you pre-fill any score, tag it as an assumption at Low confidence in a notes column. Use Manus 1.6 Max.

Goal: Analyze the attached dataset [file name] and write a report that ends in a clear decision.
Constraints: Answer this specific question: [decision question, e.g. "which two products should we discontinue?"]. Use [metrics] over [timeframe], segmented by [segment]. Decision threshold: [state the rule, e.g. discontinue if margin < X% and volume declining].
Deliverable: A report with sections (1) Executive summary + the recommendation, (2) Key findings with supporting figures, (3) The decision applied against the threshold, (4) Risks and what would change the call, (5) Methodology.
Audit rules: Every figure computed from the file; cite any outside figure's source URL. Tag each recommendation with a High/Med/Low confidence level. State assumptions and data-quality caveats plainly. Do not fill gaps with guesses. Use Manus 1.6 Max.

Goal: Compare [options] using the attached data [file name] and recommend one, showing your weighting.
Constraints: Evaluate each option on [criteria] with weights [criterion = %...]. Pull the underlying numbers from the file for [timeframe]; where the file lacks a value, note it rather than estimate.
Deliverable: A memo with (1) a weighted scoring table (option × criterion, weighted total, rank), (2) a one-paragraph rationale per option, (3) a clear top pick with runner-up, (4) the conditions under which the pick flips.
Audit rules: Show the weighting math and every score's source in a methodology note. Tag the final recommendation and any estimated input with a confidence level. Flag any criterion you couldn't score from the data. Use Manus 1.6 Max.

Goal: Investigate why [metric, e.g. conversion / revenue / retention] [changed, e.g. dropped 12%] and report the likely cause.
Constraints: Compare [affected period] against [baseline period] using the attached data [file name]. Break the change down by [dimensions, e.g. channel, segment, product] to isolate where it originated.
Deliverable: A report with (1) the headline change quantified, (2) a decomposition table showing each dimension's contribution to the change, (3) the most likely cause(s) ranked, (4) recommended next checks, (5) methodology.
Audit rules: Distinguish clearly between what the data proves and what is inference — tag inferred causes Low/Med confidence. Show the decomposition math. Do not assert a single cause if the data supports several; list them. Use Manus 1.6 Max.

Goal: Turn the attached dataset [file name] into a one-page executive KPI briefing.
Constraints: Cover [core KPIs] for [timeframe] with period-over-period change. Keep it to one page; lead with the answer. Audience is [leadership / board].
Deliverable: A briefing with (1) a 3-bullet "what matters this period" up top, (2) a KPI table (metric, value, change, status: on/off track), (3) two or three "watch items" with a recommended action each, (4) a one-line methodology footnote.
Audit rules: Every number from the file; mark any KPI not computable as such. Tag each "watch item" action with a confidence level. Keep language plain — no hedging that hides the message, no invented figures. Set up this same briefing to run on a schedule if I ask. Use Manus 1.6 Max.