Check which date field each system uses. Kapiche filters by your analysis's default date field. Your other system may use a different date, for example a survey's recorded date instead of its start date, or the call date instead of the call processed date. Survey platforms often have their own default too: Forsta's responses page, for example, filters by interview start date, so the same month can show a different total there than in Kapiche. To change the date field Kapiche uses, see How do I set a default date field?

Check the date range and time zone. A record near midnight at the end of a month can fall into different months in systems that use different time zones.

Check your dates were read correctly. Numeric dates such as 01/02/2024 can be read day first or month first. See Understanding date formats in Kapiche.

Check the start date of your integration. Integrations may only import data from a set start date, so older records may not be in Kapiche.

Filters in Kapiche. Workbench filters, segments, compare mode and saved views all change the numbers you see. Clear them, or apply the same ones you used in your other report. Smart Reports can also have their own filters, set when the report was edited.

Exclusions in your other system. Many reports leave records out by default, such as abandoned calls, test or partial survey responses, or transferred calls counted only once. Check whether those records were included in the data sent to Kapiche.

Rows skipped at upload. If you chose to skip bad or missing dates, rows with missing or unreadable dates weren't loaded.

File limits. Uploads are limited to 500K records per file, and 500MB for CSV files (100MB for xlsx). Larger exports need to be uploaded in batches, so check every batch was added. See Formatting your Data.

Integration syncs. Check the import status of your integration for failed or partial syncs. Common causes are changed permissions, the owner of a survey leaving your organisation, and fields renamed in the source system.

Blank values in the source. If a field is blank or NULL in the file or system you exported from, it stays blank in Kapiche. Comments that are blank at the source won't appear in theme counts.

Overlapping uploads. If you add files that overlap in time and your project doesn't use a unique ID, the overlapping rows are added twice. With a unique ID and record updates enabled, a row with an ID that already exists updates that record instead. See How do I update records in my Project data?

Combined projects. A record that appears in two source projects is counted once for each.

Records, comments and conversations. A record can have several text fields, and a call or chat can be split into several records. Theme counts are based on comments, so they can differ from your record count.

Themes overlap. One comment can belong to several themes, so theme counts add up to more than your total, and theme percentages won't sum to 100%.

Exports. Some exports have one row per text field rather than one row per record, so their row count can be higher than your record count. See How do I export data from Kapiche?

Pivot tables. See Why do my Pivot Table numbers look different from my Workbench?

Blank and non-numeric values. Scores only include valid numeric values. Values such as "10 - Great" are cleaned, as described in How does Kapiche clean numerical and NPS data?

Score field settings. Check the scale and the values counted as top or bottom box in your score field configuration match how your other system calculates the score.

Weighting. If your other system weights results, compare against a weighted metric in Kapiche rather than an unweighted one.

Rounding. Small differences, such as 0.1 points, are often caused by rounding at different stages.

the project name

the number you see in Kapiche and the number in your other system

the date range and filters used in each

a screenshot or export from both, if you can