Column Detection

Automatic column type detection and mapping

Smelt automatically analyzes your CSV headers and sample data to identify standard field types.

How Detection Works

1. Header analysis — Column names matched against known patterns
2. Confidence scoring — 0.9 for exact matches, 0.6 for fuzzy matches
3. Sample verification — Sample values displayed for confirmation

15 Recognized Field Types

Smelt automatically detects these column types:

#	Field Type	Example Headers
1	First Name	first_name, FirstName, First
2	Last Name	last_name, LastName, Surname
3	Full Name	full_name, Name, Contact Name
4	Email	email, Email Address, Contact Email
5	Phone	phone, Phone Number, Mobile
6	Company Name	company, Company Name, Organization
7	Job Title	title, Job Title, Position
8	Industry	industry, Sector, Vertical
9	City	city, City, Location
10	State/Region	state, Region, Province
11	Country	country, Country, Nation
12	LinkedIn URL	linkedin, LinkedIn URL, Profile
13	Website/URL	website, URL, Company Website
14	Company Size	company_size, Employees, Size
15	Revenue	revenue, Annual Revenue, ARR

Viewing Detected Columns

After upload, the file detail page shows:

• Each column name from your CSV
• Detected field type (if recognized)
• Confidence score
• Sample values from your data

If Detection Is Wrong

Column detection is for display purposes only. It doesn’t affect how templates work.

In templates, always use the exact column name from your CSV:

{{Company Name}}     ✅ Uses your actual column
{{Company}}          ❌ Won't work if column is "Company Name"

The detection just helps you understand your data—it doesn’t rename or modify columns.

Unrecognized Columns

Columns that don’t match known patterns are still fully usable:

• They appear in the column list
• You can reference them in templates with {{Column Name}}
• They just won’t have a “detected type” label

Custom Column Names

You can use any column names you want. Smelt will:

1. List all columns from your CSV
2. Attempt to detect standard types
3. Let you reference any column in templates

Use descriptive column names like Pain_Point_Research or Recent_News for custom enrichment data. You can reference these directly in templates.