Skip down to the research record

Pipeline aggregation tools: 4 options for deal rows

Quota Forecast Editorial Team · Published · 11 min read

PostgreSQL window functions keep separate deal rows: they do not group rows into a single output row like non-window aggregate calls, according to PostgreSQL.

Key takeaways

  • Use window functions to keep deal rows: each row keeps its separate identity, according to PostgreSQL.
  • Power BI CSV ceiling is 30,000 rows: The maximum number of rows that Power BI Desktop and Power BI service can export to a .csv file is 30,000, according to Microsoft.
  • Power BI XLSX ceiling is 150,000 rows: The maximum number of rows that the applications can export to an .xlsx file is 150,000, according to Microsoft.
  • Metabase exports flat tables: Pivoted results will display as a flat table in Excel, not a native Excel PivotTable, according to Metabase.

Pipeline aggregation: which options preserve deal rows

When a coverage review requires separate deal rows rather than collapsed totals, PostgreSQL window calculations provide a documented path that retains individual record identities. The documentation states that "window functions do not cause rows to become grouped into a single output row like non-window aggregate calls would" according to PostgreSQL. This distinction matters when you need to calculate values across related deals while keeping each deal visible as its own line item in the final report. Non-window aggregates collapse multiple inputs into one result, but window operations maintain the original row structure for downstream inspection.

For revenue leaders checking pipeline coverage, this means you can trace specific deal amounts back to their source records. The tool preserves the granularity needed to verify whether a coverage ratio reflects actual open opportunities or merely aggregated sums. If your current workflow relies on grouped results, consider whether the underlying data supports row-level verification before finalizing the forecast number.

Check whether your current coverage report needs separate deal rows or grouped results, then compare that requirement with the documented behavior of the tool you use. For forecast category definitions, see Pipeline, Best Case, Commit: Forecast Categories Explained.

Comparison method and reference table from publisher documentation

The comparison below maps each tool to its documented row-handling behavior, using only the publishers' own statements. For a coverage review, the key distinction is whether an operation retains separate deal rows or collapses them into grouped results. The table lists four tools: Power BI, Metabase, PostgreSQL, and pandas. Each cell reflects what that specific publisher documents about export limits or grouping mechanics.

ToolRow Preservation BehaviorDocumented Limit or Output Type
Power BIExports rows up to a defined ceiling for each file format30,000 rows for .csv; 150,000 rows for .xlsx according to Microsoft
MetabaseExports pivoted results by default; unpivoted option availablePivoted or unpivoted output according to Metabase
PostgreSQLWindow functions do not group rows into a single output rowRows retain separate identities in window calculations according to PostgreSQL
pandasGroups Series/DataFrame by mapping, function, label, or list of labelsGrouping through explicit mapping or labels according to pandas

Power BI documentation specifies distinct ceilings for different export formats. The maximum number of rows that Power BI Desktop and Power BI service can export to a .csv file is 30,000, while the maximum for an .xlsx file is 150,000, according to Microsoft. These are separate allowances based on file type, not a single universal limit.

Metabase documentation states that by default, it exports the pivoted results, but you’ll have the option to export the unpivoted results, according to Metabase.

This means related deal rows keep their individual identities when using window calculations.

Pandas documentation describes grouping as mapping a Series or DataFrame by a function, label, or list of labels, according to pandas. This is an explicit grouping operation distinct from other data manipulation steps.

Check which of these behaviors matches your current report needs before selecting a tool. For a wider tool comparison, see sales forecast calculation tools.

PostgreSQL retains rows while pandas distinguishes grouping from reindexing

When a coverage review requires calculations across related deal rows without collapsing them into single grouped outputs, window functions provide the mechanism. A window function performs a calculation across a set of table rows that are somehow related to the current row, according to PostgreSQL. Unlike standard aggregate operations that reduce multiple inputs to one output value, the rows retain their separate identities, according to PostgreSQL. This structural difference means each deal row remains distinct in the result set even after the calculation completes.

The PARTITION BY clause within OVER divides the rows into groups, or partitions, that share the same values of the PARTITION BY expression(s), according to PostgreSQL. Identifying this grouping expression explicitly is essential because it defines which deals are considered related for the calculation. If the partition expression matches the sales rep identifier, calculations span only that rep’s deals. If it matches the account identifier, that account’s rows share one partition. The partition definition controls the scope of the "related rows" mentioned in the window function definition.

In contrast, pandas documentation distinguishes between two different operations on time-series data. One operation groups Series/DataFrame by mapping, function, label, or list of labels, according to pandas. This grouping operation organizes data by the specified keys. Another operation reindexes a Series/DataFrame with the given frequency without grouping, according to pandas. Frequency reindexing aligns data points to a regular time interval but does not combine multiple rows into a single summary row. These are not interchangeable ways to summarize deal rows. Grouping organizes entries by key; reindexing adjusts the temporal index structure.

Metabase pivot exports differ from native Excel PivotTables

A pivoted export from the question interface becomes a flat table in Excel, not a native Excel PivotTable. By default, the tool exports the pivoted results, but you have the option to export the unpivoted results instead Metabase. If your coverage review requires a native Excel PivotTable, you will instead want to export the raw, unaggregated rows of data that you’ll need Metabase. This distinction matters because the reason the system does not try to export results as a native Excel PivotTable is that Excel doesn’t support all of its aggregation functions Metabase.

Trying to reconstruct these pivot tables as Excel PivotTables without those specific functions can lead to correctness issues, which defeats the purpose of looking at the data in the first place Metabase. When you choose to drop summarizations to access raw rows, one thing to watch out for is that some summarizations include implicit joins Metabase. Consequently, when you drop the summarizations, you may need to join tables to include all of the columns you’ll need Metabase. Inspect whether removing summarizations also removes implicit joins needed for the desired columns before relying on the exported dataset Metabase.

Does a Metabase pivot export become a native Excel PivotTable?

No, the pivoted results will display as a flat table in Excel, not a native Excel PivotTable Metabase. To use a native Excel PivotTable, Metabase suggests you export the raw, unaggregated rows of data instead Metabase.

Power BI limits depend on rows, intersections and data size

Power BI export ceilings are not a single allowance; they vary by file format, visual type, and connection method. Treat each limit as a distinct constraint before exporting deal-level totals for a coverage review.

The maximum number of rows that Power BI Desktop and Power BI service can export to a .csv file is 30,000, according to Microsoft. This row ceiling applies specifically to CSV output and does not extend to other formats.

For Excel files, the maximum number of rows that the applications can export to an .xlsx file is 150,000, according to Microsoft. This higher limit allows more deal rows in a single spreadsheet but remains separate from the CSV cap.

Matrix visuals introduce a different constraint based on data intersections rather than simple row counts. For export from matrix visuals using Data with current layout, the export limit is 150,000 data intersections, according to Microsoft. An intersection represents a unique combination of row, column, and value elements within the visual.

To calculate available capacity, consider a matrix with 1,000 products (rows) × 12 months (columns) × 2 measures (values) = 24,000 data intersections, according to Microsoft. This example shows how multiple dimensions multiply to consume the intersection budget.

If a matrix has three value columns in each row, the maximum row count is 150,000 ÷ 3 = 50,000 rows, according to Microsoft. Additional value columns reduce the number of rows you can export because each row consumes more of the intersection limit.

Connection type also affects export capacity. When you're using DirectQuery, the maximum amount of data that Power BI can export is 16-MB uncompressed data, according to Microsoft. This size-based ceiling operates independently of row or intersection counts and may restrict exports even when those numeric limits are not reached.

Check which limit binds your specific export: row count for CSV, row count for XLSX, intersection count for matrices, or data size for DirectQuery connections.

Illustrative example: a 1,000-deal coverage list

Suppose a coverage list holds 1,000 deal rows. That is far below the .csv row ceiling, so one export can hold every deal. If the same deals sat in a matrix with 3 value columns, each row would use 3 data intersections, and 1,000 rows would use far fewer intersections than the matrix limit allows.

Which option fits the coverage evidence you need

Match your tool choice to the specific row-level evidence required for the review. If the goal is to inspect individual deal rows while calculating values across related records, use PostgreSQL window functions. These calculations do not cause rows to become grouped into a single output row like non-window aggregate calls would, according to PostgreSQL. This behavior retains separate deal identities, which is essential when verifying that each pipeline entry remains distinct in the final report.

When the requirement shifts to building an interactive analysis in Excel, the approach changes. If you want to use a native Excel PivotTable, you’ll instead want to export the raw, unaggregated rows of data that you’ll need, according to Metabase. Exporting pre-pivoted or aggregated results from Metabase produces a flat table. To maintain full flexibility for slicing and dicing the data later, pull the base dataset before applying any summarization logic in the reporting tool.

For teams using Power BI, the decision depends on whether the visual needs to display every underlying record or just summarized intersections. Matrix visuals handle row and column intersections differently than standard tables, and export limits vary by format. Review the specific ceiling for your file type and data size before assuming the entire dataset will transfer. If the report must show every deal row, verify that the export method supports that volume without truncating the list.

Next step: Today, count the deal rows in your current coverage report and compare that count with the export limit for the tool that produces it.

Related guides: Pipeline report filters: 4 tools for coverage reviews · Sales forecast notes: 4 tools for deal review context · Check deal stage duration across 4 CRM tools.

FAQ: pipeline aggregation and deal-row access

Which pipeline aggregation options retain separate deal rows?

PostgreSQL window functions retain separate deal rows, according to PostgreSQL.

Does Power BI apply the same export ceiling to my CSV file and matrix visual?

No. Matrix exports using Data with current layout are capped at 150,000 data intersections, according to Microsoft.

Can a Metabase pivot export become a native Excel PivotTable?

No. The pivoted results will display as a flat table in Excel, not a native Excel PivotTable, according to Metabase. Reconstructing this flat table as a native PivotTable carries a risk of incorrectness if missing aggregation functions are involved.

How does pandas grouping differ from frequency reindexing?

Grouping involves organizing data by mapping, function, label, or list of labels, according to pandas. Frequency reindexing reindexes a Series or DataFrame with the given frequency without grouping, according to pandas.

What does PostgreSQL PARTITION BY control in a window calculation?

PARTITION BY divides rows into partitions that share the same values of the specified expression(s), according to PostgreSQL.

Sources