Field types and mapping

When you upload data to Genuics, every column is assigned a semantic type that tells the platform how to treat it - in charts, filters, aggregations, and AI analysis. Getting the types right is one of the most important steps for accurate results.

How auto-inference works

Genuics scans your data during upload and automatically assigns a type to each column based on the values it finds. For example:

• A column with values like 4.2, 3.8, 5.0 is detected as number
• A column with 2025-01-15, 2025-02-20 is detected as date
• A column with yes, no, true, false is detected as boolean
• A column with email addresses is detected as email

Auto-inference is correct the vast majority of the time. When it misses, you can override the type manually from the Schema tab.

Semantic type reference

Type	What it means	Example values
string	Free-form text	"John Smith", "North Region"
number	Numeric values for aggregation (sum, average, etc.)	42, 3.14, -7
date	A date or datetime value	2025-06-15, 2025-06-15T14:30:00
main_date	The primary date column used as the default time axis in charts	Same as date - but only one per dataset
boolean	True/false values	true, false, yes, no, 1, 0
sentiment	Pre-computed sentiment scores or labels	"positive", "negative", "neutral", 0.85
nps	Net Promoter Score ratings (0-10 scale)	9, 6, 3
csat	Customer Satisfaction scores	4, 2, 5
categorical	Values from a fixed set of options	"Enterprise", "SMB", "Startup"
open_end	Long-form text responses (analyzed by AI)	"The onboarding was confusing..."
ID	Unique identifiers (excluded from analysis)	"resp_8a3f2", "12345"
email	Email addresses	"jane@example.com"
phone	Phone numbers	"+1-555-0123"
URL	Web addresses	"https://example.com/page"

Why types matter

The type you assign to a column directly affects how Genuics uses it:

• Charts: A "number" column can be summed, averaged, or used as a Y-axis. A "categorical" column becomes a grouping dimension. A "date" column enables time-series charts.
• Filters: Number columns get range sliders. Categorical columns get dropdown pickers. Date columns get date-range selectors.
• AI analysis: Geni and the Insights engine treat "open_end" columns as text to mine for themes and sentiment. They treat "nps" and "csat" columns as score metrics with built-in calculation logic.
• Reports: Aggregation options (SUM, AVG, COUNT) only appear for number-typed columns.

When to override auto-inference

There are a few common situations where you should manually change a column's type:

• Numeric IDs detected as numbers. A column like "ticket_id" with values 1001, 1002, 1003 might be inferred as "number," but you don't want to sum or average ticket IDs. Change it to ID.
• Categorical values detected as strings. If a column has a small set of repeating values like "North", "South", "East", "West", changing it to categorical improves filter dropdowns and chart grouping.
• Scores detected as generic numbers. If you have an NPS column (0-10 scale) or CSAT column, explicitly typing it as nps or csat unlocks built-in scoring calculations (Promoter/Detractor breakdowns, CSAT percentages).
• Open-ended text detected as strings. If a column contains free-text survey responses, change it to open_end so the AI engine analyzes it for themes and sentiment.

How to change a column type

1. Open the dataset and switch to the Schema tab.
2. Find the column you want to change.
3. Click the type dropdown next to the column name.
4. Select the correct type from the list.
5. Click Save. Genuics re-processes the column with the new type.

Tips for clean typing

• Set main_date early. Before building dashboards, make sure the right date column is marked as main_date. This saves you from manually selecting the time axis in every widget.
• Use categorical for anything with fewer than ~50 unique values. Region, department, product tier, plan name - these all benefit from categorical typing.
• Mark open-ended columns explicitly. AI comment mining only runs on columns typed as open_end. If your survey has a free-text question, make sure Genuics knows about it.
• Don't type everything as string. String is the fallback type. The more specific you are with types, the better your charts, filters, and AI analysis will be.

Next steps

Learn how to connect external data sources and map their fields in Field mapping for connectors, or head back to Managing datasets to explore other dataset operations.