The 5 Controls Every Experiment Needs

Missing controls are the most common reason experiments fail peer review — and one of the most preventable. Here are the five control groups every experiment should have, and how to think about them.

1. Negative control

A negative control shows what happens when your treatment is absent. Without it, you can't know whether your observed effect is real or an artifact of your experimental system.

Example: In a cell viability assay testing a drug, your negative control is vehicle-treated cells (DMSO at the same concentration as your drug solution).

2. Positive control

A positive control confirms your assay is working. If your positive control doesn't respond as expected, your entire experiment is invalid — regardless of what your treatment group shows.

Example: In an apoptosis assay, staurosporine at a known-lethal concentration should reliably induce apoptosis.

3. Isotype control (for antibody-based assays)

Isotype controls match the species, class, and concentration of your primary antibody but lack specificity for your target. They reveal non-specific binding.

Example: In flow cytometry with a mouse anti-human CD3 antibody, use a mouse IgG1 isotype control at the same concentration.

4. Housekeeping control (for molecular assays)

A housekeeping gene or protein confirms equal loading or input across samples. Without it, you can't normalize your data.

Example: GAPDH or beta-actin for western blots; 18S rRNA for qPCR.

5. Technical replicate controls

Run the same sample multiple times to distinguish biological variability from technical error. Three technical replicates is the minimum for most assays.

Using AI to audit your controls

When you generate a protocol in Olto Discovery, the AI scores your control design from 0-100 and specifically flags missing control types. You can also use the "Strengthen Controls" refinement to automatically add any missing controls to an existing protocol.

Good controls aren't overhead — they're what makes your data publishable.