Uploading your first Campaign dataset (CSV or Excel)

This article walks you through preparing your data file, uploading it to the SuntheticsML platform, and avoiding the most common errors first-time users run into.

What You’ll Need

A dataset with at least 5 experiments (rows) — start with what you already have today!
Your data file saved as .csv or .xlsx

⚡ New: Lithium is now the default algorithm (*starting April 2026)

Lithium consolidates Campaign Details and the data file upload into a single step, so this guide reflects that streamlined flow. Existing Helium and Hydrogen campaigns are unaffected.

Step 1 — Prepare Your Data File

Your upload file is the foundation of every prediction SuntheticsML makes and the ML model that is generated.

Structure rules for the Upload File

Each row = one experiment (one observation, one data point)

Each column = one variable (input or output)

First row = column headers — name your variables here

Inputs = what you control (temperature, equivalents, solvent, catalyst…)

Outputs = what you measure (yield, selectivity, purity, cost…)

Example dataset

Temp_C	AA_eq	Solvent	Yield
25	1.2	DMF	68
50	1.5	THF	82
75	2.0	DMF	91
60	1.8	MeCN	77
40	1.0	THF	54

Inputs: Temp_C, AA_eq, Solvent

Output: Yield

Variable naming rules

✅ Do’s	❌ Don’ts
Keep variable names 1–8 characters	Do not exceed 20 characters (breaks plot readability in finished Sunthetics campaigns)
Start with a letter	Do not start with 'is' or 'has'
Use unique names	Do not repeat or duplicate column names
Keep level names ≤ 8 characters	Do not use long labels like: Tetrahydrofuran_anhydrous
Stick to letters, numbers, underscores	Do not use °, %, µ, spaces, or emojis in headers

Step 2 — Run the Pre-Upload Checklist

Skip any of these key data upload requirements and you’ll hit blocking warnings during campaign setup. Your file must adhere to the following:

☐ At least 5 rows of input data (5 experiments minimum)

☐ No duplicate column headers (variable names must be unique)

☐ No empty cells in input columns (every input value must be populated)

☐ Every input and output column has variance — values actually change across rows. A column where every row says “25” will not provide enough useful information for modeling and optimization.

☐ No two columns are perfectly correlated (e.g. Temp_C and Temp_F containing the same information)

☐ Categorical labels are spelled consistently — DMF ≠ dmf ≠ “DMF ” (with a trailing space)

☐ Numerical columns should only contain numbers — no text like N/A, trace, <5 in numeric columns

☐ No formulas left over from Excel — paste as values before saving

☐ Output variables/columns must have continuous numerical values

💡 Not sure if your data is clean enough?

Upload it anyway. SuntheticsML will surface Data Quality Warnings automatically, and you can fix issues before launching. The Sunthetics team can also help — email support@sunthetics.io.

Step 3 — Upload to SuntheticsML

From the Dashboard, click Create a Campaign in the top left.
In the first step (for the Lithium ML algorithm campaigns), fill in the following fields:
1. Campaign Title — be specific. Helps distinguish campaigns for yourself and your team/colleagues in the future.
2. Description (optional) — what you’re optimizing and why throughout the campaign.
3. Campaign Type — Reaction, Process, Formulation, or Other
4. Tags (optional) — pick from frequently used tags, or type your own to create a custom tag and press Enter
Click Upload data and select your .csv or .xlsx file from your computer. You can also drag and drop a file into the window directly for easier use.
Click Next to move into the Data Configuration step.

🚀 Tip

The new Quick Guide info modal built into Lithium walks you through best practices at each step. Use it the first few times — it’s faster than jumping back to docs.

Step 4 — Confirm What the Platform Sees

Once the file is uploaded, SuntheticsML displays your dataset in a table and asks you to:

Tag each column as an Input or Output (default is Input — flip the toggle to the right (green) for outputs).

Pick a variable type for each input:

Type	Use it for
Continuous	Any numeric value in a range (mass, temperature, concentration)
Discrete Numerical	Numbers limited to specific discrete values (e.g., pH = 6, 7, 8)
Categorical	Non-numeric attributes, allowed for alphanumeric characters (e.., names of solvents, bases, catalyst types)
Time	Measurements taken at intervals over time
Chemical	Molecules defined by SMILES strings - valid SMILES are required in these columns

Common Issues & Fast Fixes

Issue	Fix
⚠️ “Column has no variance” warning	All values in that column are identical — drop the column, or vary the experiments
⚠️ “Potential variable correlation” warning	Two inputs move together. Usually safe to keep one and drop the other — but check with your team first
❌ Upload fails on .xlsx	Re-save as .csv (UTF-8). Eliminates hidden formula and format issues
❌ Special characters break headers	Stick to letters, numbers, underscores. No °, %, µ, spaces, or emojis
❌ Numeric column rejected	Replace text values like N/A or trace with actual numbers, or leave the cell blank if allowed
❌ File too large or has many sheets	Keep one sheet. Rows = experiments. Columns = variables. Remove other file content and ensure upload size is within the maximum allowable limit (maximum of 4 MB).

What Happens Next

After your upload, you’ll move through the rest of the Lithium campaign setup:

Variables Configuration — set min/max ranges for inputs and outputs and update variable levels

Modeling — pick Automatic (recommended) or Manual model selection for each output variable
Optimization — define goals (Maximize, Minimize Track) for each output, input constraints, and your overall campaign optimization goal
Summary — review configuration, set number of suggested + exploratory experiments, and launch

Need Help?

📧 Technical support: support@sunthetics.io
📚 More guides: browse the full Knowledge Base