Load Data from a CSV File
Load Data from a CSV File
The fastest way to get data into CRM Analytics is a CSV upload. It's a manual, one-time load — no sync, no dataflow, no recipe — and it's perfect for data that lives outside Salesforce or that you just need to get in quickly.
Our example file, Opportunity External Amounts, ships in the EA training files (the zip from the first video). It mimics data arriving from a separate system: a Name, an Original Amount, and an External Invoice Amount. Imagine finance tracks an external invoice figure that Salesforce doesn't have, and you've exported it to CSV.
The upload flow
From Analytics Studio, the path is short and worth memorizing:
- 1
Open Analytics Studio
App Launcher → scroll to Analytics Studio → open it. You'll land on the Studio home.
- 2
Create → Dataset
Click Create, then Dataset. Pick the first option: CSV file.
- 3
Choose your file
Select your CSV (here, the opportunity external amounts file) and click Next.
- 4
Name the dataset
Give it a clear name, and add a suffix if you like — it lands in your private app unless you choose another.
- 5
Review field types, then upload
Check the dimensions, measures, and dates the engine detected, adjust if needed, and click Upload File.
When you land back in the Studio, the dataset is ready to use. Notice something important on the data-landscape diagram from the last lesson: a CSV dataset sits on its own — it's not wired into a dataflow or sync, because it was a manual, one-time load. The data is simply there, ready for you.
The metadata schema (a little advanced)
When you upload, the data engine reads your CSV and guesses each field's type: text becomes a dimension, numbers become measures. You can override those guesses right in the UI.
The classic example is zip code. It's digits, so the engine reads it as a measure (something to sum or average) — but you'd never add up zip codes. You slice and group by them. So change it from measure to dimension on the spot.
Dimension
A field you group or filter by — account name, stage, region, zip code. The categorical stuff.
Measure
A numeric field you aggregate — amount, quantity, count. The stuff you sum and average.
Most of the time the UI is all you need. But some things — like multi-value fields — can't be defined in the UI. For those, download the generated metadata schema file, edit it by hand (for example, set isMultiValue: true on a field), re-upload it, and finish the upload.
Automating CSV uploads
A manual upload is great once. But what if a system drops a fresh CSV every night? Two options practitioners actually use:
Dataset Utils (GitHub)
A community tool that runs on a local or virtual machine. Set up a folder listener: when a CSV is dropped in, it auto-uploads, upserts, or replaces the dataset.
Trailblazer Community thread
An older but useful thread collecting several approaches to scripting and automating CSV loads.
CSV upload covers external and one-off data beautifully. But the heart of CRM Analytics is Salesforce data — and for that we need a smarter approach. Before we build one, the next lesson covers the single most important concept in dataset design: grain.
Discussion
No comments yet — be the first to start the discussion.
The Data Landscape
Before you build a dataset, understand the CRM Analytics data landscape: the Connect/Sync staging layer, how data comes in from Salesforce, Snowflake, CSVs, and ETL tools, and where dataflows and recipes fit.
Grain, Lookups & Joins
The one concept that makes or breaks a CRM Analytics dataset: grain. Understand grain, why you start from your lowest-level object, and how lookups and joins pull related data onto each row without multiplying it.