> AI agents: this is one page from Mammoth Analytics documentation. The index of all pages as Markdown is https://mammoth.io/docs/llms.txt. Append `.md` to any docs URL, or send `Accept: text/markdown`, to get Markdown.

# Investigate a pattern with Explore Cards, filter by filter

> Find the cause of a spike, outlier or odd segment in Mammoth by filtering one Explore Card at a time and reading how the cards to its right change.
>
> Applies to: Roles: Edit access to the workspace to filter. Where: View › Explore › Add card

Each explore filter narrows the rows the cards to its right describe. Filtering one card at a time and reading the others is called progressive discovery. You need the **Explore** panel open with a card for each column you want to study, and edit access to the workspace to filter. How filtering works is on [Explore Cards](https://mammoth.io/docs/learn/explore/explore-cards/).

## Worked example: find the cause of a defect spike

1. Double-click "Fabrication Line Beta" on the line_name card. The grid drops from 2,847 rows to 892.
2. Read the cards to the right. The shift card now shows Day 60%, Evening 25%, Night 15%. The defect_severity card shows Critical 35%, which is unusually high. The date card shows a spike in the weeks of March 15–22.
3. Double-click "Critical" on defect_severity. The operator_name card now shows 3 operators behind 80% of critical defects.

![quantity card with a filter pill moved to the first position, beside city and order_date cards that recalculate on the filtered rows](https://mammoth.io/docs/images/20261009_193721_ExploreCardsNumericFilter.webp)

## Other patterns to follow across cards

| Pattern | What to do |
| --- | --- |
| Category, then numbers | Double-click "Enterprise" in customer_tier; the revenue card to its right shows the Enterprise revenue spread |
| Time, then category | Double-click "Q4 2024" on a date card; the product_category card shows Q4 performance |
| Outliers | Double-click the highest bar on an amount card (for example prices at $999,999); the customer_name and region cards show who and where. Check the other cards for date, category and source, then decide whether to exclude the rows with a Conditional Filter |
| Several filters | "West" (3,200 rows), then "Electronics" (1,850), then "Q4 2024" (420) |
| Inconsistent values | Spot variants such as "Apple Inc.", "Apple Inc" and "Apple Incorporated" on a text card, click them, then use **Replace all values like this** (Bulk Replace) to standardize |
| Time patterns | Double-click a spike on a date card and read the cards to its right for the cause; save the period as its own View if you need it again |
| Segments | Filter by a main attribute, refine with more filters, then save the segment as a View or a task |

Only cards to the right of a filtered card recalculate, so drag the cards you want to read after the one you filter. To keep a segment, [save the filters as a View or a task](https://mammoth.io/docs/learn/explore/save-explore-filters/).

## Related

- [Explore Cards](https://mammoth.io/docs/learn/explore/explore-cards/): select, filter, exclude and clear filters.
- [Text, number and date cards](https://mammoth.io/docs/learn/explore/card-types/): what each card shows, including outliers and spelling variants.
- [Find and replace values](https://mammoth.io/docs/learn/clean-and-shape/find-and-replace/): the Bulk Replace task for spelling variants.
- [Filter rows](https://mammoth.io/docs/learn/clean-and-shape/filter-rows/): exclude outlier rows for good.

---
Source: https://mammoth.io/docs/learn/explore/investigate-with-explore-cards.md · Updated: 2026-11-08