--- title: "Control Group: Definition and PM Examples" description: "A control group is the unchanged baseline an experiment is compared against. See why it matters and how A/B Testing for Product Managers teaches it in practice." canonical_url: "https://builderscamp.com/guides/glossary/control-group-experimentation" date_published: "2026-09-16" date_modified: "2026-09-16" author: "Andre Albuquerque" publisher: "Builders Camp" guide_class: "glossary" --- # What Is a Control Group in an Experiment? **TL;DR:** A control group is the version of a product an experiment's variant is compared against, kept unchanged so it serves as a baseline. Without a real control group, a team cannot tell whether a metric moved because of the change or because of something else entirely. ## What does control group mean? A control group is the set of subjects in an experiment that do not receive the treatment or change being tested, and it serves as the benchmark against which the treated group's results are compared. Per [Britannica](https://www.britannica.com/science/control-group), the control group is essential because it lets researchers isolate and measure the true effect of the treatment by providing a stable baseline; without one, it becomes difficult to tell whether an observed change is real or the result of unrelated factors. In product experimentation, this baseline role is exactly what makes an A/B test meaningful. The variant group experiences the change; the control group experiences the existing product, unchanged, for the same period. ## Why control group matters for product managers A/B Testing for Product Managers treats the control group as a foundational concept, not a technical footnote. The bootcamp's certification quiz defines it precisely: the control group represents the baseline version used for comparison against test variations, distinct from a group that sees no product at all or a random mix of users. The bootcamp also warns against a subtler failure: running multiple overlapping tests on the same audience without separating who lands in which control and variant group, which introduces interaction effects that make it impossible to cleanly attribute a result to any single change. ## Control group example In the bootcamp's Booking.com practical challenge, the test setup splits traffic 50/50 between control and variant, with roughly 40,000 users per variant per day and about 560,000 users per variant accumulated by the midpoint of the test. The control group sees the existing hotel listing page without the new "Price Match Guarantee" badge; the variant group sees the badge added. That clean 50/50 split is what allows the team to attribute the 9.4 percent conversion lift, and the drops in booking value and add-on purchases, specifically to the badge rather than to some other factor happening in the market at the same time, like a seasonal booking pattern that would have shown up in both groups equally. ## How Builders Camp teaches control group Builders Camp teaches the control group concept inside the [A/B Testing for Product Managers bootcamp](https://builderscamp.com/bootcamps/ab-testing-for-product-managers), directed by Andre Albuquerque, as part of its experiment design module covering variants, guardrails, and how to avoid hidden bias in a test's setup. Builders Camp runs live and self-paced bootcamps in product management and AI product building. [See the A/B Testing for Product Managers bootcamp](https://builderscamp.com/bootcamps/ab-testing-for-product-managers) for the next cohort dates. ## Frequently asked questions ### What is the difference between a control group and a variant? The control group is the existing, unchanged version. The variant is the new version being tested. Users are randomly split between the two so the only meaningful difference between groups is the change itself. ### Why can't you just measure a metric before and after a launch instead of using a control group? Because other factors, seasonality, marketing pushes, or unrelated product changes, can move a metric at the same time, and without a control group there is no way to separate those effects from the change you actually made. ### Does the control group ever see the new feature? No. By definition, the control group experiences the existing, unmodified product for the length of the test, which is what makes it a valid comparison point. ### What breaks a control group's usefulness? Any leakage between groups, such as users in the control group somehow being exposed to the variant, or a control group that is not comparable in size and composition to the variant group. ### Is a control group only used in A/B testing? It comes from broader scientific experiment design, but in product work it appears anywhere a team wants to isolate a change's effect, including feature rollouts, pricing tests, and messaging tests. ### How large should a control group be? Large enough, alongside the variant group, to reach the sample size needed for statistical significance on the primary metric, calculated before the test starts rather than adjusted afterward. ## Sources - [Britannica: Control Group](https://www.britannica.com/science/control-group) ## How this guide was made Researched from Builders Camp's bootcamp, track and masterclass material and the sources listed on this page, drafted with AI, and fact-checked against every source cited.